Advances in Analog Circuits Part 12 doc

Analog Circuit for Motion Detection Applied to Target Tracking System 319 L1 Time t Motion signal and VD and ID are decreased by MN2.. By using this model, it is able to track the targe

Analog Circuit for Motion Detection Applied to Target Tracking System 319

L1

Time t

Motion signal

and VD and ID are decreased by MN2 The current IC is 0 since the nMOS transistor MN4turns off when the target is not projected on PD2

The target moves toward the right side, and the target projected on PD2 Then, the voltage

VL2 becomes about VDD and IC is equal to ID since MN4 turns on IC is converted to the output

voltage VE by the integration circuit constructed with the capacitor CO and the nMOS transistor MN5 where the voltage VG2 is set to the constant value VE is proportional to the velocity of the target

In the case that the circuit is applied to the target tracking system, the voltage Vcenterdescribed in section 4 is generated by the PD located on the center of the array When the

target locates on the center of the input part, VE shows about 0 by the nMOS transistor MN6

CL PD1

Large monopolar cell L 1

Fig 2 Unit analog motion detection circuit

4 Target tracking model based on the biological vision system

Figure 3 shows the model for tracking the target based on the biological vision system The unit model EMD in Fig 1 are arrayed in one-dimensionally By using this model, it is able to track the target and capture the target in the center of the input parts In this section, I will describe the details of the model

The input part of the model is the photoreceptor P array P generates the signal which is proportional to light intensity The signal of P is input to each EMD EMDR generates the

signal VER when the target moves toward the right side EMDL generates the signal VELwhen the target moves toward the left side

I describe about the model in Fig 3 in the case that the target moves toward the right side

When the target moves toward the right side, VEL1 and VEL2 are not generated, and VER1 and

VER2 are sequentially generated The signal Vright is generated by summing VER1 and VER2 Vright

and Vleft are signals for controlling the motor M Since Vleft is generated by summing VEL1 and

VEL2, Vleft is not generated in this case Table 1 shows the method for controlling the motor In

this table, VDD means that the signal is generated and 0 means that the signal is not generated

When the target moves toward the right side, Vright is VDD and Vleft is 0 Then, the motor normally rotates for tracking the target The visual area (P array) turns to the target by the rotation of the motor When the target is captured on the center of the input array, PC located

on the center of the array generates the signal Vcenter Vright and Vleft are decreased by Vcenter

Then, Vright and Vleft become 0 and the motor stops The model repeats the tracking toward the right (rotation of the motor) and the capture of the target (stop of the motor) When the target moves toward the right side, the model can track the target well

When the target moves toward the left side, VER1 and VER2 are not generated, and VEL1 and

VEL2 are sequentially generated Then, Vleft is VDD and Vright is 0, and the motor rotates inversely for tracking the target When the target is captured on the center of the input

array, VPC is generated Vright and Vleft become 0 and the motor stops The model repeats the tracking toward the left (rotation of the motor) and the capture of the target (stop of the motor) When the target moves toward the left side, the model can track the target well

Analog Circuit for Motion Detection Applied to Target Tracking System 321

PL2PL3PL4

EMDL1EMDL2

VEL1

M : MotorFig 3 Model for tracking the target based on the biological vision system

Normal rotation (track toward the right side) Reverse rotation (track toward the left side)

Table 1 Method for controlling the motor

5 Test system for tracking the target using analog motion detection circuit

The test system for tracking the target was fabricated based on the model in Fig 3 Figure 4 shows the photograph of the fabricated test system for tracking the target It is able to track the target by arranging the unit circuits in Fig 2 in one-dimensionally The PD array fabricated on the printed board was placed on the rotating table which rotates with 360 degrees

I describe the test system for tracking the target in this section In the subsection 5.1, the measured results of the test circuit for motion detection are described The operation principle of the circuit for controlling the motor is also described in the subsection 5.2 The measured results of the test system are shown in subsection 5.3

5.1 Motion detection circuit

The test circuits of Fig 2 were fabricated on the printed board by using discrete MOS transistors (nMOS:2SK1398, pMOS:2SJ184, NEC) I measured the test circuit based on EMD

applied to the tracking system The supply voltage VDD was set to 5 V Vth, VG1 and VG2 were set to 1 V, 0.8 V and 2 V, respectively

The relationship between PD and the target (light) is shown in Fig 5(a) The light is provided

as the object The light was moved toward the right side, i.e., the light moved on PD1 and PD2

sequentially The output voltage VE was monitored by the oscilloscope The measured result

of the output voltage of the motion detection circuit is shown in Fig 5(b) When the light moved on PD2, VE showed about 4.3 V The test circuit could generate the motion signal Thus,

it is clarified from the results that the proposed circuit can operate normally

Analog CMOS circuit based on EMD

Motor driver (H bridge circuit)

Input part (PD array)

Motor

Power supply equipment

Rotating tableFig 4 Photograph of the fabricated test system for tracking the target

(a)

PD1 PD2Target

Analog Circuit for Motion Detection Applied to Target Tracking System 323

The motor rotates normally when the switches SW1 and SW4 turn on and SW2 and SW3 turn off, as shown in Fig 6(a) When the SW1 and SW4 turn off and SW2 and SW3 turn on, as shown in Fig 6(b), the motor rotates inversely The motor stops when all switches turn off

or turn on, as shown in Figs 6(c) and (d)

To realize the condition table 1, Vright controls SW1 and SW4 And Vleft controls SW2 and SW3

When Vright is about VDD and Vleft is 0, SW1 and SW4 turn on and the motor rotates normally

When Vleft is about VDD and Vright is 0, SW2 and SW3 turn on the motor rotates inversely

SW4 (OFF)

SW4 (ON)

M

SW1 (OFF)

SW2 (ON)

SW3 (ON)

SW4 (OFF)

SW2 (ON)

SW3 (ON)

SW4 (ON)

VDD

Stop

Fig 6 H bridge circuit (a) Normal rotation (b) Inverse rotation (c) Stop (d) Stop

5.3 Measured results of the test system

The fabricated test system for tracking the target in Fig 4 was measured Bias voltages set in subsection 5.1 were provided to the circuits based on EMD As the target, the light was projected on PD array

The measured results of the test system, when the target moves toward the left side, are shown

in Fig 7 The light was moved toward the left side until t=5 s from t=0 s At t=5 s, the light was stopped The system tracked the light, as shown in images at t=4 and 5 s At t=6 s, the motor of

the system stopped, and the system could capture the target on the center of the PD array

Fig 7 Measured results of the test system when the target moves toward the left side

Analog Circuit for Motion Detection Applied to Target Tracking System 325 The measured results of the test system, when the target moves toward the right side, are shown in Fig 8 The light was moved toward the right side until about 3 s The light was stopped at about 3 s The system tracked the light toward the right side, as shown in images

between t=0.5 s and t=3 s As shown in the image at t=4 s, the motor stopped and the system

could capture the target Thus, it was clarified from the results that the fabricated system can track the target and capture the target on the center of the PD array

Target (Stop) Motor (Stop)

Fig 8 Measured results of the test system when the target moves toward the right side

6 Conclusion

In this study, the simple analog CMOS motion detection circuit was proposed based on the biological vision system The simple circuits for motion detection were applied to the first stage of the target tracking system The test circuit for motion detection was fabricated on the printed board by using discrete MOS transistors The test system for tracking the target was fabricated by using the test circuit The test circuit could generate the motion signal for controlling the motor of the system The test system could track the target and capture the target on the center of the input part By using proposed basic circuits and system for tracking the target, we can expect to realize the novel visual sensor for robotics system, monitoring system and others

7 References

Asai, T.; Ohtani, M.; Yonezu, H & Ohshima, N (1999a) Analog MOS Circuit Systems

Performing the Visual Tracking with Bio-Inspired Simple Networks, Proc of the 7th

International Conf on Microelectronics for Neural Networks, Evolutionary & Fuzzy Systems, pp 240-246

Asai, T.; Ohtani, M & Yonezu, H (1999b) Analog MOS Circuits for Motion Detection Based

on Correlation Neural Networks, Jpn J Appl Phys., Vol.38, pp.2256-2261

Liu, S (2000) A Neuromorphic a VLSI Model of Global Motion Processing in the Fly, IEEE

Trans Circuits and Systems II, Vol 47, pp 1458-146

Liu, S & Viretta, A (2001) Fly-Like Visuomotor Responses of a Robot Using a VLSI

Motion-Sensitive Chips, Biological Cybernetics, Vol 85, pp 449-457

Mead, C (1989) Analog VLSI and neural systems, Addison Wesley, New York

Moini, A (1999) Vision Chips, Kluwer Academic, Norwell, MA

Nishio, K.; Yonezu, H.; Ohtani, M.; Yamada, H.; & Furukawa, Y (2003) Analog

Metal-Oxide-Semiconductor Integrated Circuits Implementation of Approach Detection

with Simple-Shape Recognition Based on Visual Systems of Lower Animals, Optical

Review, Vol 10, pp 96-105

Nishio, K.; Matsuzaka, K & Irie, N (2004) Analog CMOS Circuit Implementation of Motion

Detection with Wide Dynamic Range Based on Vertebrate Retina, Proc of 2004 IEEE

Conf on Cybernetics and Intelligent Systems, 2004

Nishio, K.; Matsuzaka, K & Yonezu, H (2007) Simple Analog Complementary Metal Oxide

Semiconductor Circuit for Generating Motion Signal, Optical Review, Vol 14, pp

282-289

Reichardt, W (1961) Principles of Sensory Communication, Wiley, New York

Yamada, H.; Miyashita, T.; Ohtani, M.; Nishio, K.; Yonezu, H.; & Furukawa, Y (2001) Signal

Formation of Image-Edge Motion Based on Biological Retinal Networks and

Implementation into an Analog Metal-Oxide-Silicon Circuit, Optical Review, Vol 8,

pp 336-342

Gessyca M., Tovar Nunez

Here I propose a sub-threshold CMOS circuit that changes its dynamical behavior; i.e.,oscillatory or stationary behaviors, around a given threshold temperature, aiming to thedevelopment of low-power and compact temperature switch on monolithic ICs Thethreshold temperature can be set to a desired value by adjusting an external bias voltage.The circuit consists of two pMOS differential pairs, small capacitors, current referencecircuits, and off-chip resistors with low temperature dependence The circuit operation wasfully investigated through theoretical analysis, extensive numerical simulations and circuitsimulations using the Simulation Program of Integrated Circuit Emphasis (SPICE) Moreover,

I experimentally demonstrate the operation of the proposed circuit using discrete MOSdevices

2 The model

The temperature sensor operation model is shown in Fig 1 The model consists of anonlinear neural oscillator that changes its state between oscillatory and stationary when itreceives an external perturbation (temperature) The key idea is the use of excitable circuitsthat are strongly inspired by the operation of biological neurons A temperature increasecauses a regular and reproducible increase in the frequency of the generation of pacemaker

potential in most Aplysia and Helix excitable neurons (Fletcher & Ram, 1990) Generation

of the activity pattern of the Br-type neuron located in the right parietal ganglion of Helix

pomatia is a temperature-dependent process The Br neuron shows its characteristic bursting

Analog Circuits Implementing a Critical Temperature Sensor Based on Excitable Neuron

Models

15

Fig 1 Critical temperature sensor operation model.

activity only between 12 and 30◦C Outside this range, the burst pattern disappears and theaction potentials become regular This means that excitable neurons can be used as sensors todetermine temperature ranges in a natural environment

There are many models of excitable neurons, but only a few of them have been implemented

on CMOS LSIs, e.g., silicon neurons that emulate cortical pyramidal neurons (Douglas etal., 1995), FitzHugh-Nagumo neurons with negative resistive circuits (Barranco et al., 1991),artificial neuron circuits based on by-products of conventional digital circuits (Ryckebusch etal., 1989) - (Meador & Cole, 1989), and ultralow-power sub-threshold neuron circuits (Asai etal., 2003) Our model is based on the Wilson-Cowan system (Wilson & Cowan, 1972) because

it is easy to both analyze theoretically and implement in sub-threshold CMOS circuits.The dynamics of the temperature sensor can be expressed as:

τ ˙u = − u+ exp(u/A)

whereτ represents the time constant, θ is an external input, and A is a constant proportional

to temperature The second term of the r.h.s of Eq.(1) represents the sigmoid function, amathematical function that produces an S-shaped (sigmoid) curve The sigmoid function can

be implemented in VLSIs by using differential-pair circuits, making this model suitable forimplementation in analog VLSIs

To analyze the system operation, it is necessary to calculate its nullclines Nullclines are curves

in the phase space where the differentials ˙u and ˙v are equal to zero The nullclines divide the

phase space into four regions In each region the vector field follows a specific direction

Along the curves the vector field is either completely horizontal or vertical; on the u nullcline the direction of the vector is vertical; and on the v nullcline, it is horizontal The u and v

nullclines indicating the direction of vector field in each region are shown in Fig 2

The trajectory of the system depends on the time constantτ, which modifies the velocity field

of u In Eq (1), if τ is large, the value of u decreases, and for small τ, u increases Figures 3(a)

and (b) show trajectories whenτ=1 andτ <<1 In the case whereτ <<1, the trajectory on

the u direction is much faster than that in the v, so only close to the u nullcline movements of

vectors in vertical direction are possible

0 0.2 0.4 0.6 0.8 1

nullcline u

u(V)

(b)Fig 3 Trajectory when a)τ=1 and b)τ <<1

Let us suppose thatθ is set at a certain value where the critical temperature (T c), which is

proportional to A is 27 ◦C The critical temperature represents the threshold temperature wedesire to measure Whenθ changes, the v nullcline changes to a point where the system will be

stable as long as the external temperature is higher than T c This is true because the system is

unstable only when the fixed point exists in a negative resistive region of the u nullcline The fixed point, defined by ˙u= ˙v=0 is represented in the phase space by the intersection of the u nullcline with the v nullcline At this point the trajectory stops because the vector field is zero, and the system is thus stable On the other hand, when the external temperature is below T c,the nullclines move, and this will correspond to a periodic solution to the system In the phasespace we can observe that the trajectory does not pass through the fixed point but describes aclosed orbit or limit cycle, indicating that the system is oscillatory Figure 4 shows exampleswhen the system is stable (a) and oscillatory (b) In (a) the external temperature is greaterthan the critical temperature, hence, the trajectory stops when it reaches the fixed point, andthe system is stable In (b), where the temperature changes below the critical temperature, thetrajectory avoids the fixed point, and the system becomes oscillatory

Deriving the nullclines equation ( ˙u=0) and equaling to zero, I calculated the local minimum

(u − , v − ) and local maximum (u+, v+), representing the intersection point of the nullclines

329

Analog Circuits Implementing a Critical

Temperature Sensor Based on Excitable Neuron Models

( ˙v=0) and remembering that A is proportional to temperature, I determined the relationship

betweenθ and the temperature, to be given by:

2.1 Stability of the Wilson-Cowan system

Wilson and Cowan (Wilson & Cowan, 1972) studied the properties of a nervous tissuemodeled by populations of oscillating cells composed of two types of interacting neurons:excitatory and inhibitory ones The Wilson-Cowan system has two types of temporalbehaviors, i.e steady state and limit cycle According with the stability analysis in (Wilson

& Cowan, 1972), the stability of the system can be controlled by the magnitude of the all theparameters Equations (1) and (2) are a simplified set representing the Wilson-Cowan system

Limit cycle ar ea

u nullcline

u (V)

Fig 5 a) u and v local maximum and local minimum b) Threshold values x and y showing

the area where the system is oscillatory

0 0.2 0.4 0.6 0.8 1

Fig 6 Nulclines and trajectories when a)θ=0.1 and b)θ=0.09

equations with and excitatory node u and an inhibitory node v The nullclines of this system,

which are pictured in Fig 2, are given by:

for the v nullcline ((Eq 2) = 0).

For an easy analysis, let us suppose that A is a constant In this case, there are some important

observations for the stability of the system

• There is a low threshold value ofθ bellow which the limit cycle activity can not occurs.

• There is a high threshold value ofθ above which the system saturates and the limit cycle

activity is extinguished

• Between these two values (x for the lower threshold and y for the higher threshold), the

system exhibit limit cycle oscillation

331

Analog Circuits Implementing a Critical

Temperature Sensor Based on Excitable Neuron Models

0 0.2 0.4 0.6 0.8 1

Fig 8 Nulclines and trajectories when a)θ=0.9 and b)θ=0.91

Let us suppose that the value of A is fixed to 0.03, in this cases, depending on the magnitude

of the parameterθ (that is the external input of the system) the Wilson-Cowan oscillator will

show different behaviors Figure 5(b) shows the area inside which the system exhibits a limit

cycle The threshold values x and y are shown in the figure.

The nullclines and trajectories for different values ofθ are shown in Figs 6 and 8 In Figure 6

(a),θ was set to 0.1, we can observe that the system is exhibiting limit cycle oscillations Thus,

for this case the system is unstable When the value ofθ is reduce to 0.09, as show in Fig 6

(b) It can be observed that the trajectory stops at the fixed point The fixed point in this area is

bI

Fig 10 Relation betweenθ ± and T c

an attractor, i.e a stable fixed point Thus, the system is stable Figure 7 show the position of

the v nullclines when θ=0.09 andθ=0.1 The other case (for a high threshold), is shown isFig 8 In figure 8 (a)θ is set to 0.9, at this point the system is oscillatory When θ is increased,

(θ=0.91) the system is stable

We could observed that depending on the parameterθ (external input) the stability of the

system can be controlled It is important to note that the stability also depends on the

magnitude of A, and that A is proportional to the temperature These observations are the

basis of the operation of the temperature sensor system.for example, by setting the value ofthe input θ, when the external temperature changes the system behavior also changes i.e.

stable and oscillatory

3 CMOS circuit

The critical temperature sensor circuit is shown in Fig 9 The sensor section consists of two

pMOS differential pairs (M1− M2 and M3− M4) operating in their sub-threshold region

333

Analog Circuits Implementing a Critical

Temperature Sensor Based on Excitable Neuron Models