Servo magazine 09 2007

Tạp chí Servo

Order 24 hours a day, 7 days a week

www.Jameco.com

Or call 800-831-4242 anytime

OTHER JAMECO ADVANTAGES:

I More major passive, interconnect and mechanical brands than other distributors

electro-I 99% of catalog products ship the same day

I Lowest prices guaranteed, or we pay 10%

I Major brand names and generic equivalents

for even greater cost savings

We’re semi nuts

We’ve got semis on the brain Jameco offers more major brands of semiconductors than anyone

— almost twice as many as these catalog distributors.* It’s another Jameco advantage

Cypress Semiconductor Avago Technologies Avago Technologies Atmel SemiconductorDiodes, Inc Cypress Semiconductor Freescale Semiconductor Avago Technologies Fairchild Semiconductor Fairchild Semiconductor Infineon Technologies Cypress SemiconductorFreescale Semiconductor Freescale Semiconductor Integrated Devices Diodes, Inc

ST Microelectronics National Semiconductor NXP (formerly Philips) IntersilTexas Instruments ST Microelectronics ST Microelectronics Lattice Semiconductor

Texas Instruments Texas Instruments Linear Technology

Lite-On SemiconductorMaxim

Micron TechnologyMicrosemiNational SemiconductorNEC CorporationNXP (formerly Philips)Renesas TechnologySharp Microelectronics

ST MicroelectronicsTexas Instruments Toshiba

:cignaZkZaXdcigdaaZg hj^iZYidhbVaagdWdi^Xh VcYXdcigdaVeea^XVi^dch

35 BASHBALL!

by Steve Judd

BotBash returns with a new game,

a radically updated “Active Arena,”

and a dif ferent approach to extreme robotic sports.

by Dennis Hong, Karl Muecke, Robert Mayo, Jesse Hurdus, and Brad Pullins

DARwIn’s first soccer tournament.

by Peter Smith

Recap of this year’s event

Features & Projects

SERVO Magazine (ISSN 1546-0592/CDN Pub Agree

#40702530) is published monthly for $24.95 per year by T &

L Publications, Inc., 430 Princeland Court, Corona, CA 92879 PERIODICALS POSTAGE PAID AT CORONA, CA AND AT ADDITIONAL ENTRY MAILING OFFICES POSTMASTER:

Send address changes to SERVO Magazine, P.O.

Box 15277, North Hollywood, CA 91615 or

Station A, P.O Box 54, Windsor ON N9A 6J5; cpcreturns@ servomagazine.com

PAGE 62

09.2007 VOL 5 NO 9

SERVO 09.2007 5

ENTER WITH CAUTION!

Columns

08 Robytes by Jeff Eckert

Stimulating Robot Tidbits

10 GeerHeadby David Geer

20 Ask Mr Roboto by Pete Miles

Your Problems Solved Here

Published Monthly By

T & L Publications, Inc.

430 Princeland Court Corona, CA 92879-1300

(951) 371-8497

FAX (951) 371-3052 Product Order Line 1-800-783-4624

www.servomagazine.com

Subscriptions

Inside US 1-877-525-2539 Outside US 1-818-487-4545

P.O Box 15277 North Hollywood, CA 91615

PUBLISHER

Larry Lemieux

publisher@servomagazine.com ASSOCIATE PUBLISHER/

Jeff Eckert Tom Carroll Gordon McComb David Geer Pete Miles R Steven Rainwater Michael Simpson Kevin Berry Fred Eady Dennis Hong Karl Muecke Robert Mayo Jesse Hurdus Brad Pullins Steve Judd Peter Smith Alan Federman Ron Hackett Dan Danknick Chris Olin Ray Billings Jay Johnson Bryce Woolley Evan Woolley Heather Dewey-Hagborg

CIRCULATION DIRECTOR

Tracy Kerley

subscribe@servomagazine.com MARKETING COORDINATOR

Brian Kirkpatrick

WEB CONTENT/STORE

Michael Kaudze

sales@servomagazine.com PRODUCTION/GRAPHICS

Shannon Lemieux Michele Durant

ADMINISTRATIVE ASSISTANT

Debbie Stauffacher

Copyright 2007 by

T & L Publications, Inc.

All Rights Reserved

All advertising is subject to publisher’s approval.

We are not responsible for mistakes, misprints,

or typographical errors SERVO Magazine

assumes no responsibility for the availability or condition of advertised items or for the honesty

of the advertiser.The publisher makes no claims

for the legality of any item advertised in SERVO.

This is the sole responsibility of the advertiser Advertisers and their agencies agree to indemnify and protect the publisher from any and all claims, action, or expense arising from

advertising placed in SERVO Please send all

editorial correspondence, UPS, overnight mail,

and artwork to: 430 Princeland Court,

Academic vs Commercial Robotics

I had the fortune of participating in

the 7th IEEE International Symposium on

Computational Intelligence in Robotics

and Automation (CIRA) conference, held

this summer in Jacksonville, FL Academic

researchers from Asia, Africa, Europe, and

the Americas presented their work on

topics ranging from robot manipulators

and cooperative robotics, to robot vision

As expected, the event was a

concentrated refresher on the leading

edge of robotics research An unexpected

take-away from my conversations with the

researchers at the conference was the

differences between academic and

commercial robotics, and the merit of

each perspective

The most obvious differences are a

result of the reward systems applied to

each group Academic roboticists are

rewarded for tackling intellectually

challenging problems that can result in

academic publications, with little or no

regard for short-term commercial

significance As a result, academic

roboticists tend to focused on niche areas

that they can explore and perhaps claim

for their own — whether it’s a novel

algorithm, manipulator design, or robot

configuration Intellectual property,

which provides leverage for academic

promotion and securing government- and

industry-sponsored grants, is guarded

until publication I liken the academic

model to an inverted funnel, in which

researchers drill down on an intellectually

stimulating and novel domain, and only

later broaden their approach to consider

the broader applications of their work,

including commercialization

Commercial roboticists, in contrast,

are entrepreneurs They are rewarded by

the marketplace for recognizing and

understanding a problem and thenidentifying a commercially viable technicalsolution Their approach can be modeled

as a funnel in that broad issues areconsidered first, and then choices arenarrowed by issues such as intellectualproperty protection, cost of manufacture,competition, availability of distributionchannels, support requirements, liability,and likelihood of follow-on products thatwill insure growth of a company

Commercial roboticists tend tofocus on problem areas that are dull,dangerous, and dirty and know thatsuperior technology is necessary butinsufficient for commercial success

Intellectual property is either kept astrade secrets or protected with patents,trademarks, and copyrights

Despite these and otherdifferences, academic and commercialrobotics are intertwined in a symbioticrelationship The best marketingstrategy and business model willeventually fail if the underlyingtechnology doesn’t adequately solve thebuyer’s problems Conversely, academicresearchers increasingly rely on fundingfrom private firms and corporations tosupplement their typically limitedacademic funding

Academic departments are alsoaware of the pressure to develop roboticapplications that eventually see the light

of day, as opposed to a short-livedappearance on the Web or as areference in an academic journal

Academic programs in robotics arejudged in part on the placement success

of their graduates in industry, whichadds pressure on programs to producegraduates with knowledge and skillsapplicable to commercial products

The take-away for you is that, as anindependent robotics innovator, you can

Mind / Iron

by Bryan Bergeron, Editor Œ

Mind/Iron Continued

Dear SERVO:

I read Pete Miles’ reply in the July

SERVO about the problems of using

infrared beacons in the house and wanting

to offer another approach

I use RobotBASIC to develop a

complete house navigation simulation

using “infrared beacons.” It works

perfectly, mainly because I used a very

different approach (Note: If you are not

familiar with RobotBASIC, please see my

article in the June issue)

The assumption in the simulation was

that I had 10 or so beacons (each with a

unique ID code) that could be turned on by

remote control (BlueTooth, for example)

The program used a computer

science data structure called a graph thatessentially provided a “map” of the house

When the robot was requested from anygiven room, the program checked the mapand turned on the beacons one at a time,

in the proper order, allowing the robot tomove (using the shortest path) fromwhere it was to where it wanted to go

The simulation involved a complexfloor plan and demonstrated how simplenavigation can be if only somemanufacturer would provide a beacondetector, and several beacons (that perhapslook like smoke detectors) each with theirown ID code that could be set with DIPswitches Since beacons could be placedthroughout the house (probably primarilyover doorways) they do not need to beextremely powerful typically, they wouldonly need to be detectable from 15-20feet In the rare cases where that was notenough, two beacons could be used to getthrough a room or a hallway

The point is, I believe infraredbeacons can provide an extremely easy-to-use, economical way for hobbyists

to navagate a complex environment Mysimulation would even go around objectsfound to be blocking the path and thenresume movement toward the beacon

Of course, most microcontroller-based

robots would not have the memory or mathability to deal with the graph data but thenew version of RobotBASIC (2.0) provides abuilt-in protocol that allows all the standardsimulation commands to control a real robotover a wireless link This means even smallrobots now have the ability to be controlled

by a complete, powerful language giving newlife to the goal of AI in small machines.Hmmm looks like I rambled some,but I am very proud of the capablities ofRobotBASIC and I think it opens manydoors to more intelligent control thathave been overlooked because of previouslimitations imposed by the required use ofsimple microcontrollers

John Blankenship

Dear SERVO:

This correction refers to Tom Carroll’sarticle on robotic arms The followingstatements are incorrect: “the advantage

of using R/C servos is the positionalfeedback;” “potentiometric feedback, as inR/C servos allow the controlling computer

to know where each joint is positioned.”There are no feedback mechanismsbuilt into any standard servo today, with the exception of the AX-12+ and a fewspecialized servos used in Biped type-robots

Alex Dirks, CrustCrawler

benefit from aspects of both

commercial and academic approaches

Before beginning your next robotics

project, consider how you’d approach

design and development from a

commercial and then academic

perspective Certainly, you want to

learn something from everything you

do, but if you’re looking at making

your innovation applicable to a larger

audience, then you have to consider

the business issues in component

selection, design, potential sales

volume, and follow-on products

When considering component

selection, it helps to imagine that

you’re building, say, 5,000 robots

Suddenly, a few cents for a component

or the cost of an added square

centimeter of pc board or aluminum

chassis makes a big difference in overall

cost Sensors and other components

that are just good enough for your

application may not be pushing the

technology envelope, but customers

won’t care if the robots work as

advertised Conversely, a new

microprocessor might be more

expensive to install initially, but

incorporating it in your design will give

you a chance to learn about the device,

and may provide a growth path for

add-ons The bottom line is that, with a

combined approach, your robotics

innovations can be both technologically

and financially sound SV

SERVO 09.2007 7

Research Seeks Human-Like

Movement

Robots are not generally known

for smooth movement and gentle

touch, but Oussama Khatib, a

computer science professor at

Stanford (www.stanford.edu) and a

participant in the Honda Humanoid

Robot Project, is trying to fix that

Conducting a study of several

college students and a visiting Chinese

tai chi master, the professor noticed

that humans instinctively move so as

to minimize effort and discomfort

(such as how students throw their

beer cans and pizza boxes on the floor

rather than getting up and walking

over to the trash can) So, rather than

generating a slew of complicated

trajectory-computation algorithms, he

is taking an approach based on simply

minimizing the energy the robot uses

for a particular task

Smooth movements should use

less energy than jerky ones, so

the concept sounds valid (as long

as the robot doesn’t decide to just

switch himself off) So far, “StanBot”

is a prototype that exists only as

a computer simulation But in a year

or so, Khatib hopes to incorporate

his concepts into one of Honda’s

ASIMO units Eventually, with this

type of programming, it should be

able to iron clothes and pick up thetrash for us

Machine Has Adaptable Gait

Last year, RunBot — developed at

the University of Göttingen (www.

bccn-goettingen.de), the University

of Glasgow (www.gla.ac.uk), and the University of Stirling (www.stir.

ac.uk) — gained renown for becoming

the world’s fastest two-legged robot(relative to its size) It moves along at3.5 leg lengths per second, whichtranslates into 0.8 m/s or roughly 1.8 mph The trick is that it walks using just a few sensors and detects only when a foot touches the groundand when a leg swings forward, mimicking the way control reflexeswork in humans

Originally, RunBot just trottedaround in a circle, on a flat surface

(A couple amusing mpeg videos

are accessible at www.cn.stir.ac.uk/

~tgeng/research.html) Now the 30

cm device has been fitted with aninfrared eye that detects a slope in its path and automatically adjusts itgait Like a human, it leans forwardslightly to adjust for an uphill gradeand takes shorter steps In its firstattempt to climb a hill, it fell over backwards, but the control circuitsmanaged to learn from the experienceand did well after that

Upgraded UAV Deployed

On a more menacing note, the USAir Force Air Combat Command

(www.acc.af.mil) recently announced

that a new UAV, the MQ-1 Reaper (as

in grim), is ready for deployment inAfghanistan “soon.” Larger (3,700 lbs,

66 foot wingspan) and more powerfulthan the better known MQ-1 Predator,

it can fly at 300 mph, reach altitudes

of 50,000 feet, and carry payloads up

to 3,750 lbs

The $69.1 million craft is designed

to attack time-sensitive targets and hitthem primarily with gravity bombs, but

it can also be fitted with Hellfire siles and some other things that couldspoil your day A secondary mission is

mis-to perform intelligence, surveillance,and reconnaissance duties, using sensors to send real-time data to com-manders and intelligence specialists.The MQ-9 isn’t truly robotic at thispoint, as it requires a crew of two

Oussama Khatib and a mechanical

pal Photo courtesy of Stanford University.

RunBot has been upgraded to deal with sloped paths Photo courtesy

of Bernstein Center for Computational Neuroscience.

The MQ-9 Reaper UAV is ready for action Photo courtesy of USAF Air Combat Command.

by Jeff Eckert

R o b y t e s

(pilot and sensor operator) But at

least they sit at a remotely located

ground control station

Eurobot Passes Tests

Under development since 2003 by

the European Space Agency (ESA,

www.esa.int), the multijointed

(seven joints plus one end effector

and camera per arm), three-armed

Eurobot has undergone operational

testing in the Neutral Buoyancy Facility

of the European Astronaut Centre

in Cologne

In the trials, the wet model

— which is similar in size and

configuration to the planned space

flight model — was checked for its

ability to move and manipulate

objects, provide coordinated

multi-arm movement, and recognize

obscured targets By all accounts, the

tests went very well

According to Philippe

Schoonejans, the Eurobot project

manager, “Not only has it been

demonstrated that Eurobot can

walk around an orbital station

autonomously and safely, using no

more than the existing EVA handrails,

it is also becoming clear that Eurobot

can really help the astronauts And

in the next phase, we also plan

to demonstrate its use on a

to be an available bachelor The creatures have been around for 200million years, so it would seem that they have this breeding thingunder control Nevertheless, Victoria

University (www.vuw.ac.nz)

post-doctoral student Jennifer Moore(actually a Michigan native) wants to

understand how males establishdominance and attract females Enter

“Robo-Ollie,” made from a cast of arecently departed real lizard known

a female signal He was then reprogrammed to open his mouth andgape, which apparently irritates othermales to no end Among real tuatura,this degenerates into puffing (whichOllie hasn’t learned yet) and all-outwarfare in which the reptiles oftenlose their tails

About 75 percent of the malesturn out to be lonely losers, and theother 25 percent get to produce all ofthe young As one might expect, thewinners tend to be the largest individuals, some of which can be

90 years old Moore suspects thatsmell could play a role in the matingprocess, so research continues SV

R o b y t e s

Eurobot is lowered into the Neutral

Buoyancy Facility for operational

concept testing Photo courtesy of ESA.

A robotic version of this tuatara will mimic the real thing to help researchers.

Photo courtesy of Ohio University Research (www.research.ohiou.edu).

SERVO 09.2007 9

HobbyEngineering Hobby Engineering

The technology builder's source for kits, components, supplies, tools, books and education.

Robot Kits For All Skill Levels

Motors, Frame Components and Scratch Builder Supplies ICs, Transistors, Project Kits

BEAM Kits and Components

Books and Educational K

Most orders ship the day received! World-wide shipping Convenient payment options.

Order by Internet, phone, fax or mail.

www.HobbyEngineering.com

1-866-ROBOT-50

1-866-762-6850 1-650-552-9925 1-650-259-9590 (fax) sales@HobbyEngineering.com

180 El Camino Real Millbrae, CA 94030

Visit our store near SFO!

The 2007 Unmanned

Vehicle Show

The Unmanned Systems North

America arguably hosts the largest

robot, drone, and UV hardware

showcase and information exchange

in one place at one time This

year’s extravaganza “transformed” the

Washington Convention Center in DC,

August 6th through the 9th The event,

growing year after year for 35 years,

was the biggest thus far, according to

an AUVSI media release

This year’s event — sponsored byHoneywell, Northrop Grumman, AAICorp., the Air Force Research Lab atTyndall AFB, EADS North America, theIsrael Aerospace Industries LTD,NovAtel, and SpaceAge Control —hosted more than 200,000 square feet

of research and military robot vehicles

The UVs conquered the ConventionCenter, making it their own for fourdays of presentations, speeches, techni-cal sessions, and up-close examinations

Airborne, water-treading, andland-roving robots shared the stage

with representatives from the ArmySpecial Ops, Naval Meteorology and Oceanography Command, TheDefense Science and TechnologyAgency in Singapore, and other offices,commands, and branches

of foreign governments, and industryleaders and decision makers, according

to AUVSI

Attendees were filled to the brim with new technology briefingsand operational reports and live

UV exhibits They were offered aglimpse into the future in the form

of a live operation show at WebsterField near Patuxent River Naval Air Station in St Inigoes, MD whereUVs trampled the ground, took flight,and flaunted their larger-than-lifecapabilities

From Beyond the Shadows, it Sees!

UVs making a showing at the

2007 AUVSI event (present in 2006, aswell) included AAI Corporation’s

Contact the author at geercom@alltel.net

by David Geer

AUVSI UV Sampling

The Association for Unmanned Vehicle Systems International (AUVSI) — the largest non-profit globally that is dedicated to unmanned systems — hosted its annual Symposium in August

2007 AUVSI members trek the world making their way to this event, packing their UV wares

and showing them for government organizations, industry, and academia.

AUVSI Symposium 2006 show room with UVs on display.

Shadow 200 Tactical Unmanned

Aircraft Systems (TUAS) — an

intelligence, surveillance, and

recon-naissance UAV

During its average six hour journey,

the 11 foot plane (14 foot wing

span) collects target data via its

EO/IR (Electro-Optical/Infrared Remote

Sensing) payload The EO/IR data is

transmitted back to a US Army brigade

tactical operations center (ground

control) within 125 kilometers range

via wireless technology

The UAV takes flight from a

hydraulic launcher and trailer The

TUAS has been in active duty in Iraq

since 2003

Yes, it’s a Hauling

MULE

Lockheed Martin’s Multifunction

Utility, Logistics, and Equipment

(MULE) robotic vehicle uses a 6 x 6

independently articulated suspension

and in-hub wheel motors for each of six

wheels The MULE is the most mobile

vehicle in otherwise impassable terrain,

largely due to its unduplicated

suspen-sion system and independent wheel

configuration

The MULE vehicle is a hefty (2-1/2

ton) Unmanned Ground Vehicle (UGV)

designed to aid foot soldiers so they

don’t have to carry heavy packs and

artillery on their backs

The MULE will include a mobile

carriage atop a single chassis, an

autonomous navigation system, a

remote control unit, and three

varia-tions on the payload configuration,

depending on each MULE’s application

and mission

The Autonomous Navigation

System (ANS) gives the MULE the

capa-bility to follow other MULEs in a pack

or to move in a partially autonomous

fashion with the guidance of the

remote control unit

The MULE can be transferred

by air in a sling beneath a rotorcraft

(helicopter)

The MULE is configurable in a

number of ways including a

line-of-sight, gun mount assault

configuration, an anti-tank version, and

a countermine version

The assault configuration willinclude a rapid-fire suppressive weapon mount This Armed Robotic Vehicle (ARV)-Assault-Light (ARV-A-L) configuration MULE uses an integratedweapons and reconnaissance, surveil-lance, and target acquisition (RSTA)

package to support the dismountedinfantry’s efforts to locate and destroyenemy platforms and positions.The countermine MULE configura-tion will provide the capability to find,mark, and neutralize anti-tank minesusing a mine detection package.The MULE can carry (obviously,

SERVO 09.2007 11

The Lockheed Martin MULE Unmanned Ground Vehicle appeared at the AUVSI Symposium 2006 with sensors mounted atop

(The 2007 photos were not available at press time.)

The Shadow is AAI Corp’s UAV with sensing for intelligence, surveillance, and recon missions The shadow (blue) is seen aboard a field hydraulic

launching unit, which sends it airborne.

given the acronym) equipment and

supply payloads sufficient for two

squads of infantry men and women It

can carry up to 2,400 lbs of rucksacks

and equipment for foot soldiers and follow those soldiers into the mostdifficult terrain

The MULE can also transport

evac and rescue operations; it isextendable via reconfigurable side rails

to support a number of other mountsand applications

“Heaven’s on Fire?”

The Fire Scout UAV from NorthropGrumman is an unmanned intelligenceand targeting helicopter with missileweapons mounts

Payloads include an EO/IR/LD(electro-optical/infrared/laser designa-tor) BRITE Star II sensing, a UHF/VHFradio communications relay, a COBRAmine detector, and an airborne com-munications package The BRITE Star IIsystem — which offers optics, infrared,and laser rangefinder sensing — comeswith the basic model Fire Scout

The flying ‘heli-robot’ can belaunched and flown up to 110 nauticalmiles from the launch site The UAV iscontrolled by tactical control system(TCS) software from land or sea-goingvessels The Fire Scout can remain airborne for eight-hour missions TheUAV is equipped with a four-bladerotor system and state-of-the-art airfoilblades, which help it to achieve itslonger flights

The UAV is the first airborne robot

to land vertically on a sea-borne vesselwithout a human pilot onboard

The UAV can also use its VerticalTake Off and Landing TacticalUnmanned Aerial Vehicle (VTUAV)technology to help extend communica-tions range and capabilities It does this in tandem with ground control stations, other air vehicles, data links,remote data terminals, and tacticalcommunications hardware, according

to a source from Northrop Grumman.Basically, it helps create a communica-tions relay between these systems andthe robotic air vehicles using them in away not dissimilar to nodes that extendthe signal on a wireless network

A VTUAV system will consist of air vehicles, ground control stations(GCSs), modular mission payloads(MMPs), data links, remote data terminals (RDTs), launch and recoveryequipment, and tactical communica-tions equipment

GEERHEAD

The Northrop Grumman Fire Scout appears with four-blade rotors,

weapons mounts, and sensor module.

The American Dynamics BattleHog

UAV on display at the 2006 AUVSI with

nose, wings, wheels and landing gear,

sensors (black), and special high torque

aerial lift for flight enclosed in the

center of the vehicle

The BattleHog100x has a Turbine

Powerplant in the rear center of the

main trunk or shaft of the vehicle

Forward from that is the HTAL vertical

lift and control system Each wing, leftand right, houses fuel tanks and serviceaccess for maintenance

The main housing holds the centerfuel tank Each wing has a payloadpylon for rocket launchers Ruggedizedavionics, front and center, help fly the plane A modular mission payloadbay forward from that house the sensing module

BATTLE HOG!

Calling Hogs Into

Battle

The BattleHog100x Advanced

VTOL (Vertical Take-off and Landing)

Fixed-Wing UAS (Unmanned Aircraft

System) from American Dynamics

Corporation appeared at the 2006

AUVSI symposium

The land or sea operations UV

flight system is an all-operations

platform for intelligence gathering,

area surveillance, recon, and targeting

(ISR&T) field missions, making it

attractive to all military cores When

speaking of the system, we’re talking

about three BattleHog UAVs plus

ground control and data link

communi-cations technologies

The BattleHog gets lift and even

navigation from its patented High

Torque Aerial Lift (HTAL) technology

Through the HTAL, the BattleHog

can go straight up like a helicopter,

hover in a stationary position, and flyhorizontally (in fixed wing flight)

The craft can be fitted with AGM-114K Hellfire missiles, Hydra-70rockets, and an M134 Minigun (7.62 mm shells)

The basic payloads includeEO/IR/LD sensing modules TheBattleHog can also be outfitted with aminiSAR (Synthetic Aperture Radar)system, and environmental sensorssuch as biological, chemical, and meteorological apparatus Payloadscan be reconfigured in the fieldbetween operations

BattleHog communications nologies include a CDL STANAG 7085line-of-sight data link for forward link (uplink) command and control,telemetry, flight imagery data, and payload image transmission return link(a SATCOM satellite communicationsdata link is optional to ensure communications beyond line-of-sight)

tech-Why are so many companies

suddenly building unmanned vehicles?Simple: congress has mandated thatone third of all military vehicles berobotic by 2015 Not only is that good for soldiers but it is certainlygood for established — and emergent

Perform proportional speed, direction, and steering with

only two Radio/Control channels for vehicles using twoseparate brush-type electric motors mounted right and leftwith our mixing RDFR dual speed control Used in manysuccessful competitive robots Single joystick operation: upgoes straight ahead, down is reverse Pure right or left twirlsvehicle as motors turn opposite directions In between stickpositions completely proportional Plugs in like a servo toyour Futaba, JR, Hitec, or similar radio Compatible with gyrosteering stabilization Various volt and amp sizes available.The RDFR47E 55V 75A per motor unit pictured above.www.vantec.com

STEER WINNING ROBOTS

WITHOUT SERVOS!

Order at (888) 929-5055

To do this, we will examine

gradient descent learning

using the delta rule and will

teach a three layer network using

a technique called backpropagation

And finally, we will implement

this program on a PIC16F877A

microcontroller with a serial

connec-tion and LEDs

as points in two-dimensional space Forexample, the inputs to a logical ANDgate could be plotted as the x,y sets:

(1,1), (1,0), (0,1), (0,0) We want theneuron to classify these points based

on whether they are true or falseaccording to AND logic If we plot thefour points, we see that we can easilydraw a single line across the graphbetween the points that should outputtrue and the points that should outputfalse (Figure 1) This means the prob-lem has a linearly separable solution;

we can draw one straight line betweenclassifications in the solution space.The XOR logic function is aclassic example of a problem which

is not linearly separable and, whichconsequently, the perceptron isincapable of solving An XOR gatehas the following truth table:

FIGURE 1 Linear separation of

the AND function.

FIGURE 2 The XOR function is linearly

inseparable.

In Figure 2, you can see that it is

impossible to draw a single line

separating the true from the false

output classifications for this problem

In order to solve the XOR function, we

need multiple layers of neurons, and

in order to teach multiple layers of

neurons, we need backpropagation

Gradient Descent

Learning

Before we delve into

backpropa-gation, we are going to look

at two techniques which have

allowed neural networks to develop

finely-tuned error measurement:

continuous activation functions and

the delta rule Together these

techniques allow for an intelligent

learning technique called gradient

descent learning

With our perceptron, we used a

linear, hard-limiting activation function

This means that we picked a strict

threshold and decided every output

above the threshold would output high

and every output below the threshold

would output low With multi-layered

neural networks, we are going to

use a form of sigmoidal (s-shaped),

non-linear activation called the logistic

function:

activation = 1.0 / (1.0 +

exp( - input sum))

In other words, each individual

neuron is activated to 1 / 1 + the

exponential function of the negative

sum of all its inputs For example, let’s

look at a neuron with two input

weights: w0 with a value of 3 and w1

with a value of 65 If both weights are

connected to input values of 1,

the neuron activation will be computed

If both inputs were activated to

values of 0 instead of 1, the activationequations would look like this:

input sum = (0 * 3) + (0 * 65) = 0 activation = 1.0 / (1.0 + exp ( 0 ))

= 0.5

As you can see, the logistic activation function normalizes valuesaround a center point of 0.5

Now that we have a continuouslearning function, we can think ofneural network error as a curve intwo-dimensional space and use thedelta rule to minimize error duringeach learning iteration We don’tneed to get too far into the mathe-matical details here, just imagine that the error curve is made up ofpoints from every possible configura-tion of network weights, and gradientdescent is the slope of different portions of this curve With eachlearning iteration, we want to minimize our error by changing individual neuron weights in the mostbeneficial direction, and we can dothis by using the delta rule

Simply stated, the delta rule

choos-es the direction of traversal on theerror curve which most rapidly reducesour error The delta rule formula is:

change in weight = learning constant * (desired output – actual output) * f(x) * (1-f(x))

where f(x) is the logistic activationfunction described above

Feedforward

Now that we have an activationfunction and a learning function, weare ready to assemble our network Inorder to solve the XOR problem, weneed two inputs, one output, and fourhidden neurons plus one bias neuron

on both the input and hidden layers

The network representation is shown

in Figure 3

From the diagram, you can seehow network layers of nodes and connections easily translate into arrays

of activation and weight values in our

C program

These networks are called feedforward because activation flowsthrough in a forward direction frominputs to hidden and finally output.Feedforward networks can have anynumber of inputs, outputs, and hiddenneurons, but for the XOR example, this

is all we need

We calculate the output of the network by feeding activation frominput to output For example, if westart with input 0 = 1 and input 1 = 0,

we would begin by calculating the activations of each hidden neuron as

we did above, but this time we willinclude bias neurons:

hidden neuron 0 input sum = (input 0 * weight 0) + (input 1 * weight 1) + (bias 0 * weight 8) activation = 1.0 / (1.0 + exp ( - input sum))

We add up our final output tion in the same way:

activa-output activation input sum = (hidden 0 * weight 0) + (hidden 1 * weight 1) + (hidden

2 * weight 2) + (hidden 3 * weight 3) + (bias 1 * weight 4) activation = 1.0 / (1.0 + exp ( - input sum))

Propagate Back

Once we know our output activation, we can compare it to ourdesired output and adjust the weights

of the network toward this output.Continuing our XOR example, if weinput (1, 0) we would like an output

of 1 If we get an output of 43, weneed to adjust the individual weights

to make this happen We don’t want

to adjust them too much though or

we will make it impossible to get

an output of 0 when we have an input of (1,1), so we proceed to tweak the weights ever so slightlyusing the delta rule

We begin by adjusting the weightsconnected to the output neuron based

SERVO 09.2007 15

DIFFERENT BITS

DIFFERENT BITS

on how far off the output was from

what we wanted Then we propagate

this error back to the hidden neurons

and adjust the weights between the

inputs and hidden neurons based on

how much each hidden connection

was responsible for the incorrect

answer

Let’s walk through one example

from each layer, and then we will

see what this looks like in terms of

microcontroller C code

Our first calculation is called delta

output (delta just means change in

math-speak):

delta output = output activation *

(1.0 – output activation) * (desired

output – output activation)

Next, we calculate delta weight

for each weight connected to the

output neuron:

delta weight = hidden activation

* delta output

Finally, we adjust the value of

the weight in question:

output weight = output weight + (learning constant * delta weight)

We do this for each weight connected to the output neuron Then

we have to adjust the weights of theconnections between the input andhidden neurons This is a little bit trickier because it is more difficult tofigure out how much each weight is toblame for our error

For each hidden neuron, we late an error sum based on the deltaoutput we computed earlier, and theweight of that neuron’s connection tothe output neuron:

calcu-error sum = output weight *

delta output

Next, we compute a delta hiddenfor each hidden neuron just like ourdelta output above:

delta hidden = (hidden activation

* ( 1.0 – hidden activation))

* error sum

Finally, for each connection feeding into the hidden neuron, we calculate a delta weight and adjust thatconnection’s weight value:

delta weight = input value

* delta hidden hidden weight = hidden weight + (learning constant * delta weight)

Microcontroller Implementation

a Circuit

That is really all there is to it.Constant tweaking over a few thousand iterations and you haveevolved yourself an XOR gate! The CCS (Custom Computer Services;

www.ccsinfo.com) C implementation

is pretty straightforward and similar

to the perceptron implementation from last time The trickiest part iskeeping all the array indexes straightand having a clear picture of the network layout

FIGURE 3 XOR feedforward

neural network architecture.

Figure 4 explains how the various

array indexes map back to the network

architecture from Figure 3 I will briefly

explain the circuit layout and the code

flow, as well as parts of the program

that might be confusing

Wire up the PIC chip and MAX233

chip (see the July column for more

detailed instructions and illustrations)

Five of the LEDs are going to form a

level meter showing us intuitively

how well the neural network is

learning I used different color LEDS

and ordered them top to bottom from

hottest to cool (red, orange, yellow,

green, white)

The other three LEDs show us the

status of the inputs and output of our

aspiring XOR gate I used two white

LEDs for the inputs and a green LED for

the output

Connect the cathode (-) of each

LED through a 220 ohm resistor to

ground and connect the anodes (+) to

the following microcontroller pins:

Pin LED

D2 Error meter — red

D3 Error meter — orange

C4 Error meter — yellow

C5 Error meter — green

D4 Error meter — whiteB3 Input 0 — whiteB4 Input 1 — whiteB5 Output — green

When you are finished, your board should look like Figure 5

bread-b Code

The code is broken down into five functions plus main (download acopy from the SERVO Magazine web-

site at www.servomagazine.com):

• get_activation receives a neuron’s

input sum and returns the activationvalue passed through the logistic function

• output_error receives the current

root mean square error and sets the LEDs in the error level meteraccordingly

• output receives the current input

values and sets the input and output LEDs

FIGURE 4 Array maps.

FIGURE 5.

Breadboard setup.

DIFFERENT BITS

SERVO 09.2007 17

• feedforward receives the current

input values and calculates each

neuron’s current activation

•train receives the current input values

and the desired output value It calls

feedforward to activate the network

and get the output_activation It then

measures the actual output against the

desired output and backpropagates

the error, adjusting each weight value

accordingly

• main initializes the microcontroller

settings and performs a couple quick

tests to ensure that the serial portand LEDs are working It initializesrandom hidden and output weightsand iterates through a training loop8,000 times or until the root meansquare error is less than 0.35 (it usually takes about 400 iterations)

Every 50 iterations, it prints the countand current error value When theloop completes, it prints how manyiterations it took, the final error value,and loops once through an LED display sequence demonstrating what

it has learned

After programming, you should

see the following behavior from your circuit:

1) The LEDS perform their startup blinkand the “hi!” message prints out theserial port

2) Initial weight values print out theserial port, I/O lights blink and the topLED on the error meter lights up

3) Every 50 iterations, a count anderror measurement print out the serial port

4) The error level starts to fall lower;the level meter changes accordingly

5) After 400-8,000 iterations, the network has learned the XOR function (have patience!) The lastLED in the error level meter lights

up and the program performs its ending dance

Final Thoughts

This is a simple circuit to displayneural network principles, but thesame ideas and basic code apply tomore complex problem sets.Experiment with the code and circuitand try to create your own input/output mappings Think about whatyou would like to teach a robot tolearn, and how you could modify thiscode so that your circuit could learnfrom experience and feedback from itsenvironment Have fun! SV

• Eight light emitting diodes (LEDs)

• One pushbutton (for reset)

• RS-232 level shifter (MAX233 orsimilar)

• 1 µF capacitor (if using MAX233)

• Serial cable with receive pinavailable for breadboard use

SERVO 09.2007 19

Q. The other day, I stumbled

across a mini Sumo kit from

Solarbotics called a Sumovore

I was wondering if this is a good robot

kit for beginners It looks really cool

— Jackie Butler Dallas, TX

A. You can’t go wrong with the

Sumovore kit from Solarbotics

(www.solarbotics.com) This is

an excellent robot kit Many people

think that BEAM (Biology, Electronics,

Aesthetics, and Mechanics) robots are

too dumb to make an effective robot,

especially a mini Sumo robot I would

have to say they are dead wrong This

robot can track another robot better

than any of the other mini Sumo

kits currently available on the market,

in my opinion

Though the basic kit is not

programmable, it will stay on the mini

Sumo ring all day long, and will search

out and push any opponent on the

ring This basic kit will produce a very

competitive robot for mini Sumo

contests It is a little lighter than the

maximum weight for most events, but

a few lead weights can easily fix that

If you want to program the

Sumovore robot, Solarbotics sells

add-on boards for the BASIC Stamp, Atmel

Mega8L, and PIC16F877

microcon-trollers With these microcontrollers,

you will be able to add just about any

motion control feature you want

Keep in mind, this is a kit that

requires a lot of soldering Depending

on your soldering skills, it could takefrom a couple hours to a full day toassemble If you are not comfortablewith soldering, get some help By thetime you are done assembling this kit,you will be an expert at soldering

Q. I found an old pistol grip

soldering iron in a junk pile atschool It gets really hot andmelts solder fast, but it is really hard touse to solder electronic parts The tipcan be taken off with a screwdriver

Are there any tips that come to a smaller point for this soldering iron?

— Adam Salie

A.There is probably a reason that

that soldering iron was in the junk pile These pistol gripsoldering irons do put out a lot of heat, but they are not very good at soldering modern electronic circuits Inthe old days when vacuum tubes were common, they were great tools since many of the electronic components were soldered wire-to-wire, component-to-component, and

on the phenolic circuit boards, the component spacing was far and wide

These soldering irons do work well

at soldering large diameter wirestogether, and it will even solder smallbrass parts together (if you use soldering flux) It is a good tool to havearound in case you need a lot of heat

But for circuit boards? It is too bulky,and the soldering iron tip is too big

to solder a single integrated circuitleg/trace

Yes, the tips are replaceable, but not with the fine tips needed forroutine soldering of circuits I wouldrecommend that you get a basic pencil style soldering iron that is atleast 25 watts in power and has areplaceable fine tip for soldering yourelectronic circuits

Q. A friend of mine gave me a

huge box of old electroniccomponents, like resistors,capacitors, transistors, ICs, switches,LED displays, etc Some of theRadioShack receipts in the box arefrom 1976 So my question is, do elec-tronic components go bad with age?

— Bob Cook

A.From my experience, electronic

parts generally don’t go badwhen they are sitting on theshelf not doing anything Your biggestproblem will probably be some corro-sion on the lead coated componentleads/wires I personally have compo-nents that are over 30 years old, andthey still work like a charm, and I usethem all the time Mix and match the old with the new You might find itdifficult to find a data sheet for someparts that have long become obsolete,however

With the electronic world movingtowards 3.3 volt electronics, thoseolder components may become

Tap into the sum of all human knowledge and get your questions answered here!

From software algorithms to material selection, Mr Roboto strives to meet you where you are — and what more would you expect from a complex service droid?

valuable I know a guy that collects

vacuum tubes and builds custom

vacuum tube amplifiers for people’s

radios and musical instruments He

swears that the sound from vacuum

tube amps is better than what you can

get from modern digital equipment

Q.Where is a good place to buy

cheap electric gear motors?

— Mike Coleman

A.There are a lot of good places to

buy cheap electric gear motors,

like surplus houses and eBay I

generally don’t want to recommend

these places since inventory changes

without notice (i.e., the next time you

want the same motor, you may never

find it again), and quality is unknown

until you have it I’m not saying these are

bad places to get parts (since many of

the readers of SERVO Magazine buy

from these resources and appear to be

very happy with them) I prefer to

recom-mend places where you know you willalways be able to get the same parts

With that said, Lynxmotion

(www.lynxmotion.com), Jameco Electronics (www.jameco.com), Solarbotics (www.solarbotics.com), and the Robot Store (www.robot

store.com) are very good places to get

low cost motors If you want some really powerful motors, check out NPC

Robotics (www.npcrobotics.com).

Q. I am curious to hear your

opinion on timing belt drives

vs sprockets and chains for amiddleweight combat robot Rightnow, I am in the information-gatheringphase so I can build a winning robot

— Lynn Hickman

A. Timing belts (or otherwise

known as synchronous belts) andsprockets and chains will workjust fine for your robot Performancewise, you really are not going to notice

a difference between the two systems,

unless you under-size them Timing beltsystems are lighter and quieter thansprockets and chains, are not as readilyavailable as sprockets and chains andare usually a little more expensive.Finally, they usually require a little more precision in the overall robotdrivetrain construction If you use a belt tensioning pulley, then the precisionrequirements are removed Chain andsprocket systems can tolerate a looser fitand still function fully without slipping.Choosing one drive system overanother drive system is really more of apreference than anything else Though

in your area of design, where weight isparamount, the lightness of timingbelts may be well warranted

Sprockets and chains and timingbelts and pulley components can beobtained at Stock Drive Products

(www.sdp-si.com) They have

documents on their website thatexplain how to properly size your components and how to calculate thecenter-to-center distances between thedifferent drive shafts SV

.EW

+656*

6SHHGVHF 7RUTXHR]LQ 6WHHO*HDUV

+656*

6SHHGVHF 7RUTXHR]LQ 6WHHO*HDUV

+657*

6SHHGVHF 7RUTXHR]LQ 7LWDQLXP

$//63(&,),&$7,216$792/76

No Bamboo Required

The animal kingdom has

been brought to life with

Robopanda™ from Wowwee

Robotics™ Robopanda is

a fun-loving and curious

robotic friend who is

animated, talkative, playful,

and 100% interactive

You can interact with

Robopanda in a natural way

by touching or activating

his many sensors located

throughout his body

Standing approximately 19 inches tall, Robopanda is

equipped with lifelike behaviors and motions like that of a

genuine panda Robopanda comes with activity and story

cartridges and is equipped with internal dual-axis

accelerometer tilt sensors If left unattended for a couple

minutes, Robopanda will go into ‘Sleep Mode.’ Robopanda

also features volume control and safety touch sensors

Simply insert six ‘C’ batteries (not included) and four ‘AA’

batteries (not included) and Robopanda is fully functional

Robopanda (ages eight and up) comes equipped with

two “activity and story” cartridges and will be available

this fall for an approximate retail value of $169.99

Additional cartridges are sold separately and will be

new high power dual 25A

motor controller It accepts

battery voltages from 6V to

24V and will handle peak

currents of 50A per motor for several seconds When

paired with suitable high current motors and batteries,Sabertooth 2X25 is capable of driving around robotsweighing several hundred pounds

The Sabertooth 2X25 allows you to control twomotors with analog voltage, radio control, and serialmodes, making it suitable for a wide variety of projects Abuilt-in 5V BEC can provide power to an R/C receiver and

a selectable lithium mode will allow you to safely use R/CLiPo battery packs

As with Dimension Engineering’s other motor drivers,the product’s options are set with DIP switches and wiringconnections are made with screw terminals, making iteasy to reconfigure and move from project to project

Sabertooth’s custom-designed synchronous regenerative H-bridge topology returns the motor’s storedinductive energy to the battery in every switching cycle.This technique results in motors running cooler andextends battery life It also provides more responsive control — allowing you to make instant stops and reverses

A heatsink comes preinstalled and the unit has electronic thermal and overcurrent protection for maximum durability

The product retails for $124.99 at DimensionEngineering’s website, as well as other fine robotics stores.For further information, please contact:

TReX Dual-Motor Controller

Pololu introduces its new TReX Motor Controller — a versatile

Dual-DC motor controller suited formixed autonomous and radiocontrol of small- and medium-sized robots Three independentinterfaces are offered: radio control(RC) servo pulse interface; analog voltage;

and asynchronous serial The serial interfacecan switch instantly with one of the other two interfaces,allowing mixed autonomous and remote control For example, a robot could be configured to runautonomously most of the time, but a human operatorcould override the autonomous function if the robot getsstuck or into a dangerous situation If the serial mode is selected as the primary interface, high-resolutionmeasurements of all five RC inputs are made available tothe autonomous robot controller, allowing for complex

New Products

CONSUMER ROBOTS

MOTOR CONTROLLERS

Website: www.dimension engineering.com

Dimension Engineering

Website: www.wowwee.com

WowWee Robotics

and unlimited mixing of operator control and sensor input.

The TReX motor controller operates from 6V to 16V,

and the two primary outputs provide bidirectional control

with peak currents of 30A and continuous currents of

13A, while a unidirectional auxiliary output delivers over

15A (continuous) A fourth control channel for invertible

robots allows improved control if the robot does get

turned up-side down, and the fifth control channel

determines which interface controls the motors The unit

measures approximately 1.9” x 1.9” x 0.8”, and it is

available in fully assembled or kit form starting at $99.95

at www.pololu.com/products/pololu/0777/.

For further information, please contact:

RobotBASIC

RobotBASIC is a powerful computer language that

allows you to learn how to program a robot utilizing a

simulation that includes bumpers and infrared sensors foravoiding collisions, ranging sensors to detect distantobjects, and ground sensors that can detect lines on the floor and drop-offs such as stairways It also has anelectronic compass, a GPS system, and a camera that can detect the colors seen by the robot and navigationalbeacons mounted within the simulated environment.Since all the sensors are integrated into the language,users can concentrate their efforts on learning to programthe robot to do interesting things instead of spending timedownloading code, adjusting sensors, charging batteries,and other things that keep most hobbyists from doingwhat they really want to do — program intelligence intotheir machine

In the new release (2.0), RobotBASIC provides support for real-world control via parallel and serial ports,USB, and a built-in robot protocol for BlueTooth The newprotocol is easy to use yet flexible enough to allow you

to build the target robot using a wide variety of motors,sensors, and other hardware

Visit the RobotBASIC website to download a free copy (This is NOT a crippled copy or one that will time-out.)

For further information, please contact:

Fax: 702•262•6894 Email: www@pololu.com Website: www.pololu.com

· 2 Serial Ports including Bi-Directional USB

· The Wiring Programming Language The Wiring language provides a simplified subset of

C or C++ that hides more advanced concepts like

classes, objects, pointers (while still making them

accessible for advanced users) You get the power

of C or C++ with the ease of a language like Basic.

Programs execute at full C++ speed on the board.

• Dual quadrature encoder support

• Programming cable included with kit

• No additional hardware needed

• Works with BASCOM and AvrDude programming software

Ideal for controlling your small robot With a Microcontroller

and onboard motor controllers, you get all the electronics

that you need (except sensors) on one board.

Kit $37.95 / Assembled $41.95

Programmable Robot Kits

INEX MicroCamp Mega8

· Atmel ATMega8

· Dual DC motor drivers

· 2 Buttons, 2 LEDs

· Serial port

· 5-Analog ports for sensors

· +5V switching power supply

As no soldering is involved and the parts are fully reusable, you can build and rebuild programmable

robots as many times as you like.

$89.95

Also Available:

Electronic Components Servos Motors Hardware Wheels & Tires and More!

More New Products on the way!

Know of any robot competitions I’ve missed? Is your

local school or robot group planning a contest? Send an

email to steve@ncc.com and tell me about it Be sure to

include the date and location of your contest If you have a

website with contest info, send along the URL as well, so we

can tell everyone else about it

For last-minute updates and changes, you can always

find the most recent version of the Robot Competition FAQ

15 BotTrot 4Bottle Race

Robot Magazine and the Robot Dreams website

have combined forces to create a robot contestthat you can enter from home Build the course,record a video of your robot completing it, and submit the video by the contest date The rules are minimal so almost any robot can enter whether

it has wheels, legs, or tracks; whether it'sautonomous or remote-control

a window and identifying two targets while avoidingrandomly placed coat hangers The outdoor missionrequires the robots to fly through obstacles and overtwo 1.2 x 1.5 meter placards within a 1 km radius

www.mav07.org

21-23 Robothon

Seattle Center, Seattle, WA

Lots of events including Robo-Magellan,Micromouse, line-following, line maze solving, the

Walking Robot Contest, mini Sumo, and 3 kg Sumo

www.robothon.org

Oc to ber

12-13 Cal Games

Woodside High School, Woodside, CA

This year’s event will be a recreation of the FIRSTRobotics 2007 Rack and Roll contest

http://intronics.bogorodsk.ru

19-21 Elevator:2010 Climber Competition

Event Center, Greater Salt Lake City, UT

Autonomous climber robots must ascend a scalemodel of a space elevator using power beamedfrom the base

www.elevator2010.org

21 Chibotica

Donald E Stephens Convention Center, Rosemont, IL

Chibotica returns with even more events: maze solving, line-following, mini Sumo, Robo-One, a robottalent show, and remote-controlled vehicle combat

www.chibots.org

26-28 Critter Crunch

Hyatt Regency Tech Center, Denver, CO

Held in conjunction with MileHiCon See robotcombat by the folks who invented robot combatcompetitions

www.milehicon.org

26-28 Korea Intelligent Robot Contest

Pohang Indoor Gymnasium, Pohang City, Korea

Several events are planned for autonomous robotsincluding Micromouse, MIROzSOT games, andintelligent robot demonstrations

http://irc.piro.re.kr

Send updates, new listings, corrections, complaints, and suggestions to: steve@ncc.com or FAX 972-404-0269

Featured This Month

30 Results — Jun 11th - Jul 15th

33 Upcoming — Sept & Oct

One discussion I hear a lot

is the “gloves/no gloves”

debate Most feel that gloveslimit your feeling for what’shappening, are likely to snagand pull your hand INTOthe blade, and don’t provide any protectionanyway I must confess, Iwear gloves when cuttingmetal that has sharp edges

or burrs, and when cutting

really splintery wood I make sure

to have safety guards in place, usepush sticks if possible, THINKabout where my hands will go ifsnagged or slipped, and thenthink again if I could do withoutthem Metal gets really hot on atable saw, especially if using acomposite blade, so I often wind

up wearing them

Another tip you might not get

in shop class: clean out the sawdust before cutting metal! Ihave burned up (literally, not electrically) two table saws whensparks set them on fire due tobuilt-up wood residue

PARTICIPATI N

Table Saw Safety

● by Kevin Berry

SERVO 09.2007 27

Club Corner: Ohio Robotics Club Hosts Insect Event

● by Chris Olin

The Ohio Robotics Club (ORC) in

association with the Robot

Fighting League (RFL), Hobbytown

USA, Microbotparts, and Dimension

Engineering held ORC’s fourth

robotic combat event in the

Cleveland, OH area

The event — House of Robotic

Destruction, Spring 2007 (HORD) —

drew 14 robot teams with a total of

27 robots from the 150 g (Flea

Weight), 1 lb (Ant Weight), and 3 lb

(Beetle Weight) classes This event

featured many teams from Ohio,

Pennsylvania, and Michigan In

addition, Team Hockey Runner

Robotics flew in with five robots

from Brentwood, TN; and Team

Apocalypse came all the way from

Mahwah, NJ to enter their one robot

The event was held in the parking lot

of the Strongsville Hobbytown USA,

which is owned by Steve Broz

Action started withfive flea weight robotsfighting a fierce roundrobin tournament Mr

Bigglesworth and BuzzingFlea tied for first place

Meanwhile in AntWeight action, 11 robotsbattled through a doubleelimination bracket Dr

Super Brain’s LabAssistant rose to the topwinning five straightmatches Cupbot came in a close second and Criminal Mastermindtook third place

Rounding out the event were 11Beetle Weight robots fighting apitched battle through a double elimination bracket One FierceJavelin came back from the loserbracket to claim first prize Can ofNuns came in second losing only to

Javelin and The Box finished in third.Prizes were provided for the toptwo finishers in each weight class

by microbotparts.com and Dimen

sionEngineering.com.

Look for future ORC event dates

at www.ohiorobotclub.org SV

All remote controlled vehicles

require some form of onboard

power There are various methods of

providing that power, but for combat

robots the needs are fairly specific

You need:

1) The appropriate voltage to operate

the motors and/or accessories

that you use

2) Enough amperage to power all systems at once

3) Sufficient capacity to last throughout

an entire match, with some reserve

4) To fit within the size and weight constraints available

5) Possession the physical strength to withstand the shock loads of combat

Another game I play with myself

(and my helpers, often stray children)

is “where will it go?” When the saw

kicks back or bites too well, and the

cut object goes flying (and it will!), I

like to know that only solidly

immov-able, undentable things are in the

way Second choice is a wide open

space, like across the yard

A special problem is when thesafety guard has to come off, likewhen cutting a dado (groove) Thisincreases the chances of a kickback

or runaway, and exposes the blade

Even when by myself, I run through

my checklist out loud “Where will

my hand go when I slip?” “Wherewill it go when it gets tossed?”

“What happens at the end of the cutwhen it falls off the table?”

If you Google images of “tablesaw injury” you might see why I’msuch a fanatic about this topic If youlook at my hands, you’ll KNOW why!(I have eight fingers, two thumbs; alloriginal factory attachments withoutsurgical intervention) SV

Left to right: Evan Gandola, North Omlsted,

OH, driver of Criminal Mastermind; David Winek, Brentwood, TN, driver of Cupbot; Chris Olin, Perry Township, OH, driver of

Dr Super Brain’s Lab Assistant.

BATTERY PACK CONSRUCTI N

● by Ray Billings

FIGURE 1

Obviously, some thought has to

go into your power needs! By far, the

most common method used in

combat robots is multi-cell packs of

Nickel Cadmium or Nickel Metal

Hydride batteries You can purchase

ready made NiCd and NiMH packs

from a few different sources, but as

in most things, not all packs are

created equal

Sometimes due to an unusualshape needed to fit your design, or

an unusual voltage requirement, an

off-the-shelf pack may not fit your

needs Whether it’s to fit an unusualshape, or to save some money bydoing it yourself, you may want tocreate your own packs

The first requirement is to decide

on which particular cell you aregoing to use This is not as easy as

it sounds, as there are literally hundreds of cells on the market, withvast differences in performance

Typically, you want a cell with atleast a 10C discharge rate Thismeans that for a 3,000 mAh ratedcell, you should be able to draw atleast ten times that — 30 amps —without causing damage

Avoid cells that enthusiasticallyclaim “high discharge rates” withoutsupplying numbers to verify whatthey mean by “high.” Also, there aremany cells on the market that weredesigned to be used in power tools,and come with tabs pre-welded toone end (or both ends) of the cell

Although the tabs may havebeen adequate in making power toolpacks, they will not be adequate in acombat pack The tab itself is usuallyvery thin, and cannot handle theamperage a combat pack has to supply And, they have a tendency topop off during combat, which is bad

If you do come across cells that youwant to use that have tabs already

on them, you can carefully removethem before assembly

For this article, I am buildingsome packs from 3,500 mAh NiMHsub-c cells These are 10C dischargerated cells, without tabs, and werereasonably priced at about $50 for aset of 20 cells To be sure, there aremuch better cells out there thanthese, but for the price I felt theywould work well for my needs

Other supplies needed for packconstruction are battery bars or braidfor making connections, some form

of high strength glue (I prefer theglue used in tennis shoe repair,although I have seen some buildersuse two-sided tape), wire leads,shrink wrap, and cell spacers, ifdesired Many of the specific packconstruction materials were pur-

chased from www.battlepack.com.

You will need a fairly high wattagesoldering setup to solder quickly without damaging the cells Picturedare a 260 watt gun, a 500 watt iron for major work, and a heat gun forshrinking the wrap on the pack

Pack construction can be completed with or without the separation rings These keep the cellsfrom contacting each other, andallow air flow for cooling Although Inormally prefer to use them, they do

SERVO 09.2007 29

make the pack physically larger and

because I need these to fit into a

smaller area, I won’t be using them

First, glue the cells together in

the arrangement that you need I will

be making 24 volt packs in the

standard 4 x 5 (20 cells total)

config-uration Here the cells are glued side

by side with the polarity of the cells

opposite of the adjacent cells

Once you have all the cells glued

together in a flat configuration, let

the glue set before proceeding

Glue four rows of five cells

together into a block, again mindful

of keeping the polarity of all cells

opposite of adjacent cells I use

rub-ber bands to hold the packs together

until the glue sets

Create a small puddle of solder

in the center of each cell Make sure

to let them cool afterwards!

(Continuing to solder on the cell

while it is hot can damage them.)

The solder puddle should be

centered and approximately one

quarter the diameter of the cell

I prefer using flexible braid

instead of bars for pack construction

It tends to hold up better to hard

impacts, and is easier to solder than

solid bars You can find braided

grounding cable from many online

surplus places for reasonable prices

Here, I have cut the braid into piecesfor connectors About 1.25 inches or

so are good for the sub C sized cells

— they need to be long enough to

go from the center of one cell to the center of the next cell As withthe cells, tin the connecting braidwith solder

Lay the braid across the cells youwant to connect, and heat the solder

so that the puddle on the cell pletely melts into the braid Be care-ful not to hold the gun or iron on thecell for too long, as you can damagethe cell This is why a high wattagegun is essential: massive heat trans-fer lets you perform this step quickly

com-The completed solder jointshould look like one entirely meltedpuddle on each end of the braid

Notice in the photographs the pattern of connectors for the bottom

of a standard 24V pack

You can also see the top view ofthe same pack Be mindful of thepattern of connectors during construction You will need to con-nect all cells in series for the pack

Note the two cells on the left with noconnector This will be where thewire leads are soldered on

Gauge your wire leads ately for the discharge amperage ofthe packs I am using 10 gauge

appropri-silicone jacketed wire for these

The shrink wrap that is used onthe outside of the pack is easy topuncture, and some protection fromabrasion on the connectors is a goodidea You can use thin foam padding(many commercial packs are madethis way) or, in my case, I have foundthat 10 mil pipe tape (found at HomeDepot) works well and is inexpensive.Cut the shrink wrap so that itoverhangs the pack about 1/2 inch

FIGURE 10 FIGURE 11 FIGURE 12

US Events

ROBOlympics/RoboGames 2007

was presented by ComBots inSan Francisco, CA, on 6/15/2007

Results are as follows:

● Super Heavyweights — 1st: Ziggy,

CM Robotics; 2nd: The Judge,

Mechanicus; 3rd: Vladiator, Moon

● Heavyweights — 1st: Sewer Snake,

PlumbCrazy; 2nd: Last Rites,

Hardcore Robotics; 3rd: Original

● Lightweights — 1st: Touro Light,

RioBotz; 2nd: K2, Velocity; 3rd: Texas

Heat, CM Robotics

● Featherweights — 1st: Totally

Offensive, Mad Overlord; 2nd:

Proteus, Titan; 3rd: Whammo,

Sandman

● Hobbyweights — 1st: Surgical

Strike, Rolling Thunder; 2nd:

AlphaQ Jr., RoboTrauma; 3rd: Bullet,Target Practice

● Beetleweights — 1st: Dark

Pummeler, Dark Forces; 2nd: Itsa?,Bad Bot; 3rd: Gutter Monkey, Death

By Monkeys

● Antweights — 1st: Dark Pounder,

Dark Forces; 2nd: Dark Blade, DarkForces; 3rd: Revert

● Auto Antweights — 1st: Thinkling,

Tentacle Combat Robotics; 2nd:

Spider, Terror Turtle; 3rd: Chopper,Terror Turtle

● Fairyweights — 1st: Micro Drive,

Misfit; 2nd: Dark Bullit, Dark Forces;

3rd: Skeet Skeet, Ghetto Logic

House Of Slackers II was

present-ed by Team Slackers in GlenRock, NJ, on 6/30/2007 Results are as follows:

● Antweights — 1st: Box #5, Danger

Zone; 2nd: Yelo, Pinq; 3rd: Entropy,Danger Zone

● Beetles — 1st: D12, Headbangers;

2nd: Aggravator, Dreadfully WickedRobotics

● Hobbies — 1st: Upchuck, Dreadfully

Wicked Robotics; 2nd: Battle Brick,Interloper; 3rd: Evader, RC Cars

Osceola Robot Battles was presented by Daytona AreaRobot Tournament in Kissimmee, FL,

on 7/14/2007 The event was held

at JR Hobbies

Results were not available atpress time

EVENTS

RESULTS — June 11th - July 15th

or so on each end This will give

enough room so that it will fit

perfectly after heat shrinking Take

your time with the heat gun —

over-heating will cause the wrap to crack

Properly shrunk to the pack, theshrink will be tight and curve slightlyaround the ends

Four packs, combat ready! Newcells, especially NiMH, may requireseveral charge-discharge cyclesbefore reaching full capacity SV

Te technology of the Third

Millennium runs, frankly, rather

computer heavy As a user rather

than a programmer, I tend to resort

to other forms of “Aided Design”

than the “Computer” style In thelingo of today’s bot making, the “C”

in “CAD” means “cardboard” to me

Bandied about are also “WAD”(wood), “PAD” (posterboard ), “FAD”(foamboard), and even “BVD” (brainvisualized, not underwear)

******* Aided Design

TECHNICAL KN WLEDGE

● by Kevin Berry

SERVO 09.2007 31

I tend to start with

sketches on Microsoft

PowerPoint If things look

complex, I’ll model with

cardboard or wood, then

proceed to aluminum or

sheet metal before cutting

expensive titanium or

com-posites My beetleweight

Fir Darrig went through

many iterations It started as a BVD

to Fabrication bot, with 1/8”

aluminum free form bent in a vise

Later, I transferred the shell to paper,

onto posterboard, then to sheet

metal To bend the titanium, I made

a wood model which was also the

“shoe” for a homemade brake

Ultimately, this process failed me

when I tried to cold bend the

titanium rather than applying heat,

resulting in a big crack I wound up

running the sheet metal prototype

with titanium appliqué armor

My next attempt at using wood

for a design medium came when

I attempted to rebuild my

Lightweight, Chupacabra, after

its shredding by Killerbotic’s 2EZ

I ripped fir boards down to

1/2” square to simulate aluminum

tubing With the help of drywall

screws, I modeled the frame around

the components such as motors

and batteries The model went

to my sponsor’s welding shop,

where I learned another lesson

about modeling

In an attempt to make a low

clearance bot, I’d cut the pieces very

closely to the maximum possible

dimensions Unfortunately, I failed to

indicate the “critical dimensions” to

the fabricator, so it came back

unusable for my parts Slight

tolerance buildups, differences in

material dimensions, and frankly, just

some sloppy welding resulted in a

$200 frame I couldn’t use Lesson

learned: Build it yourself, or at least

be there when it’s built!

For the next two bots I built,

I skipped the modeling effort

Our highly successful antweight,

Babe The Blue Bot, moved from

version III to version IV using a

direct BVD (actually a dinnertimeconversation from the kids, afterthe spanking defeat of III) to a 3D PowerPoint visualization, to fabrication To size the pieces, I laidall components out on the rawbaseplate material, drew aroundthem with a Sharpie, then cut Thisversion worked very well, and itwent from concept to arena with

no modifications

(L-R) Fir Darrig, wood form, titanium blank partially bent.

Powerpoint “3-D CAD” of Babe IV.

Side view of wood form being used as a bending brake shoe.

John Henry, parts mounted directly from

the rough layout.

Chupacaba: High fidelity wood models don’t necessarily mean an accurate result.

Frame built directly around the core bot,

So, when it came time for

a beetleweight version of Babe, I

went straight from BVD to layout

without bothering to even make aPowerPoint version

To see what else is being done

around the builder community, I put out a call and received lots

of examples

No one technique works for anybuilder Whether you are a high-endCAD designer or a low-tech poster-board and masking tape type, what’simportant is to spend the right balance of time between design,modeling, and building Too much ofany one wastes precious time

A final closing thought The hardest step is between the “virtual”bots in our heads and the “hard” bots

in the box However it works best foryou, get out there and build! SV

This free and easy-to-use program is

perfect for whipping up a quick

electrical schematic Whether you are

designing wiring for your own

scratch-built robot or hacking an off-the-shelf

electronic gizmo, this is a painless way

to produce an as-built diagram before

you start modifying or to completely

plan your next cunning device

I’ve found it very useful to createas-built diagrams of as much of my

bot’s wiring as possible and keep a

copy with my toolkit when traveling

to an event Repairing damage ormaking modifications in a hurry isdefinitely easier when you have anas-built to refer to This simple ideacan prevent a wiring mistake thatmight “toast” your robot!

I downloaded the setup program from the ExpressPCB web-site, installed it on my PC, and built

my first simple circuit schematic inless than 15 minutes (Figure 1) Thesetup program actually installs twoapplications: ExpressSCH for creat-

Team Mad Cow’s Morpheus is another CAD to fab bot.

Team Kick Me applies Cardboard Aided Design techniques on Kick-Me-Mo.

Finger Tech Robotics also uses cardboard

modeling on Swiss Chef.

Team V, in the April ‘07 issue of SERVO, showed their progression from paper to wood to metal.

FIGURE 1

SOFTWARE REVIEW — ExpressSCH v5.6

● by Jay Johnson

SERVO 09.2007 33

ing schematics and ExpressPCB for

creating printed circuit board

dia-grams that can be sent to ExpressPCB

for actual production of your

custom-designed printed circuit boards

ExpressSCH includes the basic

drawing tools you would expect plus

an extensive library of pre-built

components Once you’ve snapped a

component to a wire, the two parts

will stay snapped together as you

move either of them around the

diagram One of the features I liked

best was the ability to create my own

custom components on-the-fly (like

the SPNO solenoid in Figure 1) and

save them to the library for future

re-use I can also share these

compo-nents with other ExpressSCH users

This program was actually created

as a pre-processor for creating complex circuit board diagrams inExpressPCB, but I can see no reasonnot to use it to satisfy my simpler needfor basic circuit and wiring schematics

And if any of my projects evolve torequire a custom PCB, I’ll already havethe tools I need!

There were a few features thatcould be improved:

1) Component alignment tools Theuser can snap the components to agrid, but this can be tedious

2) Multiple “Undo.” The current version limits the user to undoing asingle previous command

3) Colorful components! In the current

version of ExpressSCH, to paraphraseHenry Ford, “The customer can haveany color component he wants, solong as it’s black!” Actually, the userhas some control over the generalcolor scheme, but all the componentsand wires have to be the same color

ExpressSCH is an easy-to-use toolthat ought to be in every robot-builder’stoolkit — and best of all, it is FREE! SV

Jay and Valerie Johnson are Big Bang Robotics and have been designing, building, and fighting combat robots since 2001.

US Events

Robothon Robot Combat 2007

will be presented by Western

Allied Robotics in Seattle, WA on

9/22/2007 at the Seattle Center's

Center House Classes will be 3 lb

and 12 lb bots, using a double

elimination or round robin format,

following RFL Rules No ICE or open

flames Entry Fee: $40 for first 12 lb

robot $25 for First 3 lb robot

Additional robots are half price Entry

fee discount for helping with arena

setup and take-down Special entry

fee considerations for builders who

are under 18 Arena: 12 x 12 with

18” x 18” pit surrounded by 2” high

wall making it very difficult to

accidentally drive into it Visit www.

westernalliedrobotics.com.

Franklin Institute Robotic Conflict

will be presented by North East

Robotics Club in Philadelphia, PA on

9/17 Visit www.nerc.us.

2007 HalloweenRobot Terrorwill be presented byCalifornia InsectBots in Gilroy, CA on 10/27/2007

This is open to Flea weights, Antweights, and Beetle weights Therewill also be a bot costume contest —that’s right — you have to put a costume on your fighting bot Thecostume contest will take place during

a break in fighting and the audiencewill decide the winner There will beprizes for the 1st, 2nd, and 3rd placebot costume winners Held at GilroyHobby Weigh-in starts at 10:00 AMand fighting starts around Noon Theentry fee will be $20 per fighting botwith prizes for 1st , 2nd, and 3rd place

in each fighting weight class For fight

rules, go to www.calbugs.com.

UK Events

Robots Live! Hemel Hempstead,9/1-2/2007 Event Location:Sportspace Hemel Hempstead, ParkRoad, Hemel Hempstead Visit

www.robotslive.co.uk.

UK Antweight Championships,9/8/2007 in Luton Bedfordshire.Event will host competitions in the Flea (25-75g) division, Ant (75-150g),and SuperAnt (150-250g) division.Competition will be double elimination

Roaming Robots Winter TourRound 1 — Doncaster,9/15/2007 at The Dome, DoncasterLeisure Park, Bawtry Road,Doncaster, South Yorkshire Visit

www.roamingrobots.co.uk.

Roaming Robots Winter TourRound 2 — Portsmouth, 10/6-7/2007 at the Mountbatten Centre,Alexandra Park, Twyford Avenue,

Portsmouth Visit www.roaming

PP:KLWH ZDWHUFOHDU

/('9

PFG $% 



7KH9HOOHPDQ3HUVRQDO6FRSH

LVQRWDJUDSKLFDOPXOWLPHWHU EXWDFRPSOHWHSRUWDEOH RVFLOORVFRSHDWWKHVL]HDQG FRVWRIDJRRGPXOWLPHWHU +36 

7KHURERWIURJPRYHVIRUZDUG ZKHQLWGHWHFWVVRXQGDQG UHSHDWV VWDUWPRYHIRUZDUG ! VWRS!OHIWWXUQ!VWRS!ULJKW WXUQ!VWRS 65 

6ROGHULQJ

UHTXLUHG

7KRXVDQGV PRUHLWHPVRQ RXUZHEVLWH

Ask for our FREE 96 page catalog

VISIT OUR ONLINE STORE AT

www.allelectronics.com

WALL TRANSFORMERS, ALARMS, FUSES, CABLE TIES, RELAYS, OPTO ELECTRONICS, KNOBS, VIDEO ACCESSORIES, SIRENS, SOLDER ACCESSORIES, MOTORS, DIODES, HEAT SINKS, CAPACITORS, CHOKES, TOOLS, FASTENERS, TERMINAL STRIPS, CRIMP CONNECTORS, L.E.D.S., DISPLAYS, FANS, BREAD- BOARDS, RESISTORS, SOLAR CELLS, BUZZERS, BATTERIES, MAGNETS, CAMERAS, DC-DC CONVERTERS, HEADPHONES, LAMPS, PANEL METERS, SWITCHES, SPEAKERS, PELTIER DEVICES, and much more

What the Heck is

BashBall?

BashBall combines the visceral

excitement of one-on-one robot

combat with the strategy of team play

in a fast-paced ball game requiring

speed, skill, and brute force

Two teams of remotely-operated

robots vie for points by scoring goals,

ejecting opponents from the arena,

disabling opponents, or all three in any

combination BashBall is played in an

enclosed arena with transparent walls

featuring pneumatically operated pits,

flame jets, and “death blossom”

pistons that pop up from the floor

Scoring goals in this environment is not

easy, and every match is sure to feature

plenty of exciting action!

Why Invent Another

Robotic Sport?

Since the introduction of fighting

robots in 1994 at the original “RobotWars” in San Francisco, CA, combatamong remote controlled robots hasfollowed the same basic format:

one-on-one combat on a flat floor

There have been some variations alongthe way, but in the end the winner isusually decided by single combat, withthe results based on damage caused

The natural evolution of fightingrobots following this format has produced two dominant designs: verydestructive robots with kinetic energy(KE) weapons, and low, wedgy bricks designed to survive theseweapons With few exceptions,

“robot fights” today follow one of thefollowing scenarios:

• Two highly destructive KE robotsclash briefly in a hail of shrapnel, reducing both to ineffective pushy botsfor the remainder of the tournament

• A low, wedgy brick wedges a highly

destructive KE robot around the arenafor three minutes

• Two low, wedgy bricks wedge eachother around the arena for three minutes

• A novel design is reduced to a pile ofscrap metal by a KE weapon in the first

60 seconds (often less)

• A novel design is dominated by alow, wedgy brick for three minutes

There are exceptions and variations

of course, but the general direction isclear — if you don’t want to make a KEweapon or a low, heavily armored brick,you can expect to have two losses and

go home, often with your robot inshards While this format continues toappeal to some builders, there are manyothers who want something “different.”

As with other fields of endeavor, inrobotic sports “you get what you meas-ure.” To have a different outcome, you

need to change the game BashBall is a

new game combining elements of

robot fighting, an exciting and dynamic

playing field, and strategic team play in

a fast-paced, fun ball game Robots can

and will be damaged or disabled on the

playing field, but the game is not about

damage, it’s about scoring points

Earlier, experimental versions of

BashBall were played at Botbash events

in 2002, 2003, and 2004 The current

version draws on lessons learned at

those events Video of the 2004 version

of the game is available on YouTube at

www.youtube.com/watch?v=yJCAZ

Cx_iKw Additional videos can be

found at the Botbash Channel at

www.youtube.com/botbash.

BashBall 2007

The Arena

BashBall is played in the Mobile

Automated Robot Combat Arena(MARCA) First introduced in 2002, theMARCA was designed to support newforms of extreme robotic sports and behighly mobile The MARCA provides acompletely self-contained mobilearena; mounted on a trailer, the 30’ by16’ playing surface has hydraulicallyoperated “wings” that fold up for travel The outer arena screens hangfrom a track for use and fit betweenthe folded floor sections for travel

The arena consists of a 16’ x 30’

flat playing surface with a goal opening

at each end The arena is surrounded byrails, above which are angled sections

of metal mesh which allow the ball toroll back onto the playing surface

The arena is equipped with severalobstacles: a gap in the rail throughwhich robots can be ejected, pneumati-cally operated pits, a pendulum “sweep-er” wielding a spinning metal bar, flame

jets, and “death blossom” pneumaticrams that pop up from the floor

The Game

The object of the game is to accumulate the highest possible scoreduring regulation play In the case of atie, a “sudden death” overtime periodwill be played with the first team toscore winning the match

The game is played in four quarters of five minutes each; with a

10 minute break between quarters and

a five-minute Sudden Death overtime,

if necessary

The game begins with each team’srobots positioned behind the startingline at their respective ends of thearena Play is started with the ball inthe ceiling-mounted center-court Tee.The ball is released onto the arena floorwhen a robot hits a release bar project-ing from the bottom of the Tee

The ball remains in play until a goal is scored, the ball leaves thearena, or the ball ends up in anunplayable position

Following a goal, the MARCAautomatically shoots the ball back intothe arena The ball cannons are aimed

so that the ball will usually land back in

BASHBALL!

MARCA 2007 Floor Plan.

South Carolina State Fair, October 10-21, 2007 BotBash and Mechwars Robot

Combat will be presenting five shows a day in the MARCA at the fair The balance

of the time is available for BashBall teams to come test their mettle For further

information, contact info@botbash.com

The Botbash team is working on an event schedule for 2008, and will be

making announcements as the dates are finalized

UPCOMING EVENTS

the Tee It must then be released by a

robot hitting the release lever

A team may not control the ball

for more than 20 seconds without

scoring a goal If the shot clock expires,

the ball must be surrendered, within

five seconds If the ball is not

surren-dered play is stopped and a face-off

occurs, similar to a face-off in hockey

In the event that a robot is ejected

from the arena while carrying the ball,

a timeout will occur and the ball will be

placed in the Tee Should a robot

become stranded or disabled while

carrying the ball and be unable to

release it, a timeout will occur and the

ball will be placed in the Tee

There are four ways to score

points:

• (5 points) — ACE Ball score with ball

on the “serve,” originating from ball Tee

• (3 points) — Ball score from play

• (3 points) — Throw opponent out of

arena (10 point bonus for ejecting ALL opponents)

• (1 point) – Disable opponent

In addition, a team can win thematch during Sudden Death Overtime

by scoring with the ball

Teams and Robots

A team will consist of from three

to five people and from three to fiverobots Robots must weigh at least 30lbs but no more than 60 lbs A teammay bring up to five robots total andmust select at least three, but no morethan four to play in each quarter Thetotal weight of the robots selectedmust not exceed 150 lbs Robots must

be built according to the RobotFighting League “Sportsman Class”

rules, but can weigh up to 60 lbs ratherthan the usual 30 lbs for that class

Strategy

The game is designed to

encour-age team play, by rewarding teamsthat can work as a group to scoregoals While beating up opponents can

be part of an effective strategy, a teamwill need to score points while defend-ing their own goal to be successful

The Rules

Detailed rules for BashBall androbot construction can be found at

ht tp://botbash.architeuthis-dux org/botbash-ball-rules-2007.pdf or

ht tp://botbash.architeuthis-dux org/Botbash-RFLTechRegs-2007- 06.pdf SV

Botbash LLC, headquartered inTuscon, AZ, is one of the pioneers ofrobotic sports Originally formed in

1997 to present fighting robots at

a science fiction convention in Phoenix, AZ, Botbash grew to be one

of the longest-lived robotic sports organizations, celebrating its 10thanniversary this year

ABOUT BOTBASH

SERVO 09.2007 37

ROBOCUP 2007

DARwIn’s First S ccer Tournament

America’s First Entry to the Humanoid

Division of RoboCup

by Dennis HONG, Karl MUECKE, Robert MAYO, Jesse HURDUS, and Brad PULLINS

RoMeLa: Robotics & Mechanisms Laboratory, Virginia Tech

W e are Virginia Tech’s RoboCup team “SPRInt” (Soccer Playing Robots with Intelligence) and we

just came back from RoboCup 2007.As you may remember, we were the only team that passed the qualifications from the United Sates in the humanoid division in the RoboCup history and our robot “DARwIn” (Dynamic Anthropomorphic Robot with Intelligence) was featured recently on the cover of SERVO Magazine (Vol 4, No 12) A three-part technical series about its development was featured, as well Nearly 300 teams from 33 countries geared up to compete at RoboCup 2007 in Atlanta

at the Georgia Institute of Technology July 3-10 and DARwIn fought hard against other teams from all around the world We were not able to advance to the finals, but we had a strong showing at the event.

At least we were the audience’s favorite (underdog) team — especially with DARwIn’s acrobatic moves!

We played one of the most entertaining games in RoboCup history.This is our story.

FRIDAY (Day 0)

After a long drive from Blacksburg, we arrived at Atlanta (Georgia Tech) late at night Among VT’s RoboCup team, only members Karl Muecke, Robert Mayo, Jesse Hurdus, Brad Pullins, and advisor Dr Hong could attend the competition due to a limited travel budget We brought DARwIn 2A (goal keeper) and DARwIn 2B (striker) to be theactual players for the competition, but we also brought DARwIn 1 just for exhibition We pretty much stayed up all night tweaking the walking gaits, calibrating the vision system, and putting the final touches to the robots The campus

was filled with fellow robot enthusiasts from all around the world, and you could feel the excitement in the air It was going to be a veryinteresting week

Autonomous humanoid robot DARwIn 2B playing soccer in the lab.

Members of TEAM: SPRInt from Virginia Tech.

SERVO 09.2007 39

SATURDAY (Day 1)

The main competition venue was at the CRC(Campus Recreation Center) at Georgia Tech, andtoday we were able to enter the soccer field forthe first time Surprisingly, the actual soccer fieldcarpet texture was drastically different from what

we had been using for testing, thus the dynamicwalking gait algorithms we had developed werenot able to handle it in a stable manner Also, thelighting condition was so different from what

we had been using, our IEEE1394 cameras werehaving a difficult time seeing the field and the ball We stayed at the CRC until it closed at 10 PM,trying to adjust things to make them work right

It looked like there was still lots of work to do toperfect DARwIn’s soccer playing skills ThoughDARwIn performed well in our lab in a controlledenvironment, we realized the importance of making the system robust enough to handle different environments

be quite a challenge for us to bringhome the trophy, but we were ready totry our best as first time rookies

MONDAY (Day 3)

Today was the last practice

day before the actual

competi-tion tomorrow Dr Hong has

been getting a lot of email and

phone calls from the many fans

from VT and other universities

cheering for us It looks like we

have a big following after our

cult status YouTube video on the

Internet (www.youtube.com/

watch?v=klBjbDVftJw) and the

articles in SERVO and July’s issue

of Popular Science magazine.

Though we have tweaked most of the system to work well, we discovered that

one of the computers (the one in DARwIn 2B) suddenly stopped working! This

presented a big problem as our first competition was the next day at 2 PM It

was going to be another long night ahead Dr Hong took us to a nice

restaurant for a good meal since we had been living on fast food to save time

for working (He was more worried about our health than that of the robots.)

We took DARwIn 1 to the

restaurant with us and he

was quite popular there

After a quick nap, we went

back to the dorm and took

over the lounge area to set

up our workshop and testing

field Dr Hong stayed up all

night with us, helping out on

the last minute

troubleshoot-ing Hopefully, we had fixed

everything so DARwIn could

play a good game against the

other robots the next day

CRC at Georgia Tech’s campus.

DARwIn had too much to drink

Taking a break after

a very long day of testing and debugging DARwIn 1 is having a good time with us.

TUESDAY (Day 4)

GAME 1

Today marked our very first game! We went up

against Team TKU from TanKang University, a RoboCup

veteran since 2000 The situation for DARwIn didn’t look

too good because we had been struggling to fix the

problem of DARwIn 2B’s PC/104 computer all night, but

could not find why it was not communicating with the

rate gyros and motors through its serial port After much

discussion, we decided to swap the computers between

DARwIn 2B and DARwIn 2A, thus having the

non-functioning DARwIn 2A as the goal keeper (essentially

making it a “brick” in front of the goal), hoping that it

would simply block the shots from our opponents by just

sitting there After a long and difficult “surgery,” we

finally made DARwIn 2B, functional again Due to the

difference in texture of the playing field carpet, we were not able to use our analytically generated, awesome gaits, butrather manually recorded some hand-tweaked gaits on the spot so that it could at least walk on the field and play thegame The new walking gait was not as fast or impressive as what we had before, but it hopefully would do the job (Atleast it looked like we had the strongest kick in the entire humanoid league!)

Thirty minutes before the start of the game, we realized that the lighting conditions on the actual field differed again from what we used to calibrate the computer vision cameras and that DARwIn could not see the ball! Last-minute

tweaking did not help much but the game had to begin What a disaster!

The robots were in position You could feel the tension inthe air Striker 2B was having a difficult time locating the ball andwhen it did, it dribbled it towards the goal and it missed It didn’t really kick the ball, but pushed it with its toe insteadbecause it couldn’t accurately see the location of the ball anddidn’t know when to give the full kick The robots from TKU tooksmall but quick steps, intercepting the ball and kicking it towardsour goal Unfortunately, our goalkeeper — though it had animpressive pose — was essentially a sitting duck and could notblock the ball Score 1:0 You could actually see from its bodylanguage that 2B was struggling to recognize its surroundingsand the ball The light was reflecting off from every direction,confusing him

The second half started and striker 2B got tripped by theTeam TKU robot and fell face forward You could see and hear the impact it took from the fall Even with the protectiveguards on its chest, the inertial force from the mass of the cameras mounted on the head caused them to shift The panand tilt motors used to orient the camera did not work anymore! While striker 2B was trying to get up, Team TKU scoredanother goal DARwIn 2B tried to run back to block the kick, but fell forward again, and completely lost its head (camera).DARwIn was now playing blind! DARwIn was walking around in circles, headless, trying to understand why everything waspitch black The game ended,

with a score of 2:0 With all

that went wrong, at least we

were lucky that only two goals

were scored

After the game, we tried to

fix the pan and tilt unit, but

decided to just bolt the camera

directly to the chest DARwIn 2B

no longer had a neck and would

need to move its upper body to

look around It was sad to see

DARwIn fall and break its neck,

but it could have been worse

The huge RoboCup event at CRC.

Strategy meeting before the game.

DARwIn 2A’s impressive pose (but does not move ).

A 4 AM repair job in the dorm.

SERVO 09. 2007 37

ROBOCUP 2007< /b>

DARwIn’s... the cover of SERVO Magazine (Vol 4, No 12) A three-part technical series about its development was featured, as well Nearly 300 teams from 33 countries geared up to compete at RoboCup 2007 in Atlanta...

SERVO 09. 2007 39

SATURDAY (Day 1)

The main competition venue was