Intelligent Vehicle Technology And Trends Episode 1 Part 7 ppsx

SafeTRAC uses a video camera to watch the road ahead, track the road and vehicle position in the lane, monitor for weaving and lane drifts, and alert the driver before a road or lane dep

SafeTRAC uses a video camera to watch the road ahead, track the road and vehicle position in the lane, monitor for weaving and lane drifts, and alert the driver before a road or lane departure The camera is mounted in a center position on the windshield interior Lane detection software tracks both lane markings and subtle features such as the road edge and oil strips in the lane center area Assistware has optimized the lane-tracking algorithms so that the system performs well in a wide variety of lighting, environmental, and pavement conditions

SafeTRAC is unique in that the system provides a continuous indication to the driver of vehicle position within the lane, via a simple graphical display Per Figure 6.2, lane position is shown as a vertical “dash” character moving between two verti-cal “dash” lane boundary characters An audible alarm is sounded if the vehicle begins to depart the road or cross into another lane without the turn signal acti-vated Seat vibrators can also be activated from the system output Sensitivity is adjustable, and the system automatically disables when road features are inadequate for lane detection

The system also continuously tracks the driver’s relative accuracy over time in maintaining lane position to provide an “alertness feedback score” (shown as “86”

in Figure 6.2) In this way, erratic or degraded driving can be detected even if lane departures are not occurring This feedback can help the driver realize his or her level of fatigue may be more than they thought, and data such as this can also be logged for fleet managers to review as an indication of needed driver training

Iteris LDWS [8–10] The Iteris Autovue system, shown in Figure 6.3, is the market leader in LDWS Originally introduced in Europe in 2000, over 8,000 units have since been sold there and sales are averaging 4,000 systems annually Iteris estimates that the systems have logged one billion kilometers thus far in Europe alone A modified system entered the European motorcoach market in 2004 In the United States, Autovue is now available as a factory option from several truckmakers and over 600 units have been sold In the auto market, Iteris is also the supplier of the

Figure 6.2 SafeTRAC camera (left) and combination processing and driver interface unit (right).

(Source: AssistWare Technology.)

LDWS introduced by Infiniti and Citroën Autovue is sold as an integrated system installed by the manufacturer

Autovue detects lane boundaries through video-based image processing as well The system works in full daylight and at night with headlights on, as well as any weather in which lane markings are visible This includes heavy fog, as the viewing proximity is very close to the front of the vehicle A “virtual rumble strip” warning

is provided to the driver if a lane departure is imminent, using the left or right audio speakers to indicate the direction of the lane departure An exception is the Euro-pean motorcoach version, which provides directional warnings via seat vibration Iteris notes that the system promotes the use of turn signals when changing lanes and conditions drivers to have a keen sense of “lane position awareness” and remain in the lane center

In surveys conducted with over 200 truck drivers in the United States and Europe who have used Autovue, 75% or more of the drivers drove regularly with the system enabled and believed the warnings came at the right time, the system was valuable even with occasional false alarms, and the system could prevent crashes

MobilEye LDWS [11, 12] MobilEye has pioneered the development of application-specific integrated circuits for driver assistance with their EyeQ™ system-on-a-chip (SoC) The company’s vision-sensing approach based on the EyeQ™ provides lane departure warning as well as forward collision warning (see Chapter 7) The system became available to automotive and truck fleets as an aftermarket product in 2004 The system also mimics the rumble strip sound as a way of alerting the driver to a lane departure

MobilEye’s algorithmic approach fits a three-parameter road model that accounts for lateral position, slope, and curvature The curvature parameter is used for increasing the warning reliability on curved roads and for estimating time to lane crossing In addition the system retains multiple lane models (such

as urban roads, merging lanes, or exit lanes) so that it can switch between them

Figure 6.3 Iteris’s first generation LDWS Used in the United States commercial market starting in

2002 (Source: Iteris, Inc.)

instantaneously to find the best match for the conditions During heavy rain, the visual interference caused by raindrops and windshield wiper motion are pro-cessed so that lane detection is not impeded

Toyota Rearview System [13, 14] The Toyota LDWS on the market in Japan takes

an innovative approach in using the rearview camera for double duty Rearview cameras are primarily intended to assist the driver by providing an image of the area behind the vehicle on the navigation screen during parking maneuvers The Toyota system, developed cooperatively with supplier Aisin, uses the same camera to look

at lane markings immediately behind the vehicle while on the highway to realize the LDWS function

6.1.3 LDWS Evaluations

Researchers in Europe and the United States have conducted evaluations of the behavioral effects, driver acceptance, and overall safety effectiveness of LDWS for heavy truck operations These types of field trials provide valuable insight into “real world” use of such systems—are they truly supporting the driver? Two projects are briefly reviewed here

LDWS Evaluations Conducted by the Dutch Ministry of Transport [15] The Dutch Ministry of Transport sponsored field trials of LDWS during 2002 and 2003 The trials were conducted by a team of researchers led by TNO and focused on professional drivers operating heavy-duty trucks and long-distance buses

The research was organized into six major work packages:

• Analysis of the driving task and the role of LDWS;

• Behavioral effects of LDWS;

• Expert opinion of traffic flow effects of lateral driver support systems;

• Acceptance of LDWS;

• Infrastructural consequences of LDWS;

• Relation of LDWS to the use of narrow road lanes

The test fleet consisted of 35 trucks and five motorcoaches Five of the trucks had data recorders to collect detailed information Several different LDWS, typical

of those offered commercially, were used

Overall, the effects of LDWS on traffic safety were seen to be positive The results indicated that, with all trucks in the Netherlands equipped, approximately 10% of injury crashes involving heavy vehicles could be prevented With respect to traffic flow, LDWS are not expected to have either a positive or negative influence, other than the reduction in congestion due to fewer truck crashes

The Dutch government has investigated reconfiguring existing roadways into narrower lanes to create additional lanes as a way of reducing congestion There-fore, LDWS were evaluated for their ability to help drivers maintain correct lane position in narrow lanes Truck driving simulator experiments were conducted in which the lane widths on the virtual road varied from 3.5m down to 2.9m A dis-traction task was intentionally introduced to stress the lane-keeping task of the

driver The study results showed that the LDWS system improved lane-keeping, particularly for the narrower lane widths At the same time, however, drivers reported driving to require more effort when using the LDWS

LDWS enjoyed a high degree of user acceptance among drivers who used the systems during the on-road evaluation This was also the opinion of managers at the transport companies involved, a key point given that company management must see benefits to make decisions to buy such systems A total of 75% of the drivers had positive opinions of LDWS, and over 50% stated that they would prefer to drive with such a system installed in their vehicle However, 21% of drivers stated that they would prefer vehicles without LDWS

On the positive side, drivers noted fewer “startle” reactions during lane departure events by using LDWS, as they were advised earlier in the event and therefore could respond more gracefully; similarly, reaction times to take cor-rective action were reduced Interestingly, 60% of drivers concluded that the system caused them to pay more attention to the driving task Increased comfort levels were noted, as well

On the down side, 20% of the drivers felt that the system could cause a startle response that might be worse than crossing the lane line; this factor was seen as potentially being related to the loudness of the audible warning Further, drivers perceived 25% of the warnings as needless (i.e., false alarms) Obviously, this would argue for the provision of sensitivity and volume adjustments being available

to the user, at the level of either the fleet or individual driver

Mack Trucks/U.S DOT LDWS Field Operational Test [16] The U.S DOT Federal Motor Carrier Safety Administration, as part of the IVI program, partnered with Mack Trucks and McKenzie Tank Lines to evaluate the SafeTRAC LDWS on a fleet

of approximately 20 tractor-tankers The test, which is ongoing, involves extensive data collection on each of the trucks To provide a basis for comparison, lane detection and data collection is active on all trucks, but the driver warning is disabled for a subset of the fleet For the trucks with activated driver interfaces, driver acceptance is being evaluated through the use of surveys

The onboard measurement system provides the following critical information:

• Vehicle state;

• Driving behavior;

• Roadway alignment and lane markings;

• Presence of precipitation;

• LDWS operational status;

• LDWS alerts;

• Lateral velocity and lateral acceleration during lane departure events;

• Surrounding traffic;

• Location

When a lane departure event occurs, data from the previous 30 seconds and the following 30 seconds is captured in one-second intervals The data acquisition sys-tem logs several different types of lane departures: less than 10 inches, 10–18 inches,

and greater than 18 inches When the field testing is completed in 2005, over one million miles will have been logged by the test fleet The data will be analyzed to assess the overall safety benefit of the system, as well as any negative impacts

As one might expect, RDWS are similar to LDWS in providing lane tracking How-ever, there are some important differences and specialized applications in RDWS that are reviewed here

6.2.1 Curve Speed Warning

Curve speed warning systems advise drivers when their speed is too high for an upcoming curve These systems are currently in developmental stages and have not yet been introduced commercially; however, this application is expected to be brought to market in the near term

The Digital Map Approach [3, 17] Here, the premise is simple: Digital maps produced for use in onboard navigation systems could contain sufficient road geometry information to enable a safe speed estimate to be generated for upcoming curves in typical road conditions When a vehicle is approaching such

a curve, an onboard processor compares this estimate with the actual vehicle speed If the threshold speed is exceeded, a warning is issued to the driver or speed is automatically reduced While some road geometry is available in current digital maps, it is generally agreed that enhanced next generation maps are needed for curve speed warning to be sufficiently reliable

Curve speed warning, as well as vehicle control techniques to reduce speed auto-matically, have been addressed by projects in the United States and Europe examin-ing the application of digital maps to driver assistance in general Ford and GM separately prototyped curve speed warning approaches for both warning and vehi-cle control in the United States EDMap project, and BMW did the same within the European ActMAP project BMW’s approach provides a good example of an advanced system implementation BMW uses an active accelerator to provide feed-back to the driver in a manner that provides both warning and a form of control Whenever the current speed is deemed too high for the road conditions, the accelera-tor pedal presents a slight but insistent feeling of resistance to indicate that the driver should slow down Through its respective position, the active accelerator pedal also

“suggests” the right speed to the driver However, the system cannot be all-know-ing, so this feedback always remains a suggestion for the driver to accept at his or her discretion In this way, the driver remains in the loop and the amount of feed-back is based on the deviation between expected and actual behavior

In addition to curve radius and curve angle, a fully informed curve speed warn-ing system would also incorporate parameters such as surface quality, street width, number of lanes, shoulders, visibility (daytime), weather (for determining friction), and driving style of the driver As curve speed warning systems evolve, they can take advantage of such data from other sensor systems to enhance the safe speed calculation

Infrastructure-Oriented Curve Speed Warning [18] In Japan, AHSRA has pursued

an infrastructure-centered approach to a curve speed warning that focuses on particular road sections known to be hazardous While static roadside signs can

be placed to provide general warnings to all drivers, these are not deemed to be sufficiently effective—drivers who are going too fast for the curve need to receive a direct warning Prior to the curve, therefore, speed detectors and road-vehicle communications equipment are installed so as to warn drivers if their speed is too high

This system approach is being evaluated at several sites On National Highway

25 in Naga Prefecture, the “Omega Curve” is infamous for its steep downgrade and length, which tends to cause vehicles to accelerate to unsafe speeds Another site, National Road 246 in Kanagawa Prefecture, is a two-lane bidirectional road in which one section has two consecutive curves on a downslope, such that the second curve is not in view and tends to surprise drivers Lane departures caused by exces-sive speed for the situation have resulted in serious head-on collisions in this section Testing is also under way on the Tomei Expressway and at several metropolitan freeway interchanges with complex and sharp flyover ramps

6.2.2 U.S DOT Road Departure Warning Field Operational Testing [19–22]

In 2001, U.S DOT partnered with the University of Michigan Transportation Research Institute, Visteon Corporation, Navteq, and Assistware in a field operational test project focused on RDWS The project defines and evaluates a system which warns drivers when they are about to drift off the road and crash into an obstacle, as well as when they are traveling too fast for an upcoming curve As shown in Figure 6.4, technologies include a vision- and radar-based

New high-definition maps and GPS Short-rangeradarsensor

Camera-based lane detection

Short-range radar sensor

Long-range radar sensor

Long-range radarsensor

Integrated human system interface

Road departure crash warning system:

Combines curve speeds and lateral drifts warnings into one uniform function

Figure 6.4 Road departure crash warning system under evaluation by the U.S DOT.

(Source: Visteon.)

lateral drift warning system and a map-based curve speed warning system A photo of the radar sensors, which are installed on each side of the vehicle, is shown in Figure 6.5

The lateral drift warning subsystem takes road detection a step further than a typical LDWS Using machine vision, it assesses the existence and width of the road shoulder Furthermore, forward- and side-looking radar detects the presence of any obstacles on the shoulder (such as parked cars) or the roadside (such as poles or guardrails) Armed with this information, the driver warning modality can be more situation-aware In the case of no shoulder or an obstructed shoulder, the warning would be at its most urgent level; conversely, when the shoulder is broad and unob-structed, the driver might only receive an advisory message Audio, visual, and seat vibration warnings are used to present the various warning levels

For the region ahead and nearby the vehicle, data collected includes the following:

• Upcoming road curvature;

• Lane width;

• Number of lanes;

• Paved shoulder width;

• Boundary marker types;

• Any temporary roadside objects (such as parked vehicles);

• Permanent roadside objects (such as bridge abutments)

At the heart of the system is a “situation awareness module,” which fuses data coming from the sensors to understand the situation and calculate available maneu-vering room

Side-looking radar

Forward-looking radar

Figure 6.5 Side- and forward-looking radar units installed on a Nissan Altima for the road

departure crash warning system (Source: UMTRI; photo: Shekinah Errington.)

System development is complete and field data collection began with a test fleet

of 14 Nissan Altima’s in early 2004 Plans called for 78 people to drive the vehicles over a 10-month period Data collection will be completed in early 2005

6.3 Lane Keeping Assist Systems (LKA)

6.3.1 System Approaches

Lane-keeping systems are intended as convenience products by reducing the driver’s need to make the frequent minute steering corrections that are a normal part of driv-ing The lane detection function is handled as described in section 6.1 and active steering input is added by the LKA system For automotive products, the paradigm

is one of shared control, whereas in specialty applications such as transit buses, full steering control is sometimes provided

Automotive LKA Systems For automotive implementations of LKA, automatic steering torque is provided by a motor integrated with the vehicle steering system Future systems will likely use steer-by-wire to “actuate” steering Per Figure 6.6, torque increases as the vehicle nears a lane edge to create a “driving in a bathtub” type sensation for the driver Surprisingly, the delivered torque to adjust vehicle direction at highway speeds is quite small Therefore, the systems are easily override-able by even the weakest drivers—in fact, the systems were tested with specially selected “weak drivers” in Japan before being introduced to the market there!

Based on the shared control paradigm, automotive LKA require driver input to remain enabled Approximately 80% of control is provided by the system and 20%

by the driver, with the systems only operating on highways of rather modest curva-ture If the curvature limits are exceeded, the system disables automatically [39]

Max torque driver to feel

Lane marker

Assist torque

Lane position

Center

Assist torque

Figure 6.6 A relative indication of steering assist torque provided in steering assist systems.

(Source: Honda.)

BMW’s LKA system, called heading control, uses onboard sensors to analyze any crosswind, curves or ridges in the road, in addition to detecting lane edges [23]

It uses this information to calculate optimal steering behavior and define tolerance limits; should these be exceeded, the system applies force to the steering wheel to suggest corrections Drivers then decide to accept the recommendation or initiate another action, such as overtaking Heading control is currently under test by BMW and is expected to be introduced to the market soon

Lane-keeping performance for LKA systems is typically better than human driv-ers can maintain One study showed maximum lateral deviation within the lane at 0.2m for the automatic lane- keeping system, as compared to 0.4m for an experi-enced test driver

When lane-keeping support systems are discussed, questions frequently arise as

to the driver’s ability to remain alert Is the system providing too much assistance, such that the driver tunes out? This is one reason that system designers have adopted the 80/20 rule for first generation systems, so that driver input is required for the sys-tem to remain active Research on this topic is discussed in Chapter 12

Full Steering Support for Transit Bus Applications BRT systems rely on various methods to provide express service to travelers as compared to automobile travel, so

as to attract greater bus ridership One approach is to provide exclusive lanes for the buses In many cities, the creation of such lanes is a major challenge given the existing development and already crowded streets Real estate is at a premium, and even fractions of a meter in the width of a new lane can make or break the viability

of new bus service of this type When the pressure is on to make the lanes as narrow

as possible (only centimeters wider than the width of the bus itself), automatic steering assistance is called for

Given the limited lane-miles of such implementations, various approaches to lane-tracking can be used, including special infrastructure treatments The CIVIS system, developed by Irisbus, uses image processing to detect distinctive lines painted on the road surface (Figure 6.7), and a guidance module synthesizes this and

Figure 6.7 Specialized pavement markings used by the CIVIS system for lane-tracking.

(Source: Irisbus.)

other relevant inputs to generate steering commands (Figure 6.8) The Phileas sys-tem, developed by APTS, relies upon magnetic markers installed in the road surface

6.3.2 LKA Systems on the Market

Automotive Systems LKA systems were introduced initially by Nissan in Japan in

2001 [24] and are now available from all major automakers there The system philosophy is that steering assist is for the purpose of improved stability and reduced driver fatigue; it is not intended for autonomous driving The systems on Japanese vehicles typically operate only over 65 km/hr on roads with a radius of curvature of 1,000m or more However, the Honda system operates on road radii down to 230m, which essentially covers all Japanese highways

An example of the Nissan LKA driver interface on the market is shown in Figure 6.9 [38] This interface is placed within the instrument cluster and shows the activation of the LKS function by illuminating the “LANE” icon When the system

is actively tracking the lanes, an image is illuminated to iconically illustrate road lanes In this image, a vehicle icon is also illuminated to indicate that the ACC sys-tem is tracking a vehicle ahead

Similar systems are expected to be introduced on the European market within approximately three years, and North America not long afterwards The market pull in North America is seen as particularly strong, given the long, monotonous intercity trips undertaken by Americans for business and vacation travel on the nation’s interstate highway system

Vision module

Guidance module Special painted lines

Steering sensor Electric

actuator Camera

Figure 6.8 CIVIS lateral guidance approach (Source: Irisbus.)

Figure 6.9 Nissan LKS driver interface integrated into instrument panel (Source: Nissan.)