Automatic Route Tracking of Smartcar

Automatic Route Tracking of Smartcar

Based on Fuzzy Control

Guiyou Chen and Zhenhua Li Ping Yang

School of Control Science and Engineering Key Laboratory of Complex System and Intelligent Sciences

Shandong University Institute of Automation, Chinese Academy of Sciences, Graduate University of Chinese Academy of Sciences

Shandong, Jinan, 250061,P.R. China Haidian district,Beijing,100080,P.R. China

chenguiyou@https://www.360docs.net/doc/753986696.html, yangping06@https://www.360docs.net/doc/753986696.html,

Abstract - The principle of fuzzy control and its application in automatic route tracking of smartcar are presented in the paper. The fuzzy controller is established to control the steering serv o motor of the smartcar. Simulation of the designed controller based on MATLAB fuzzy logical toolbox is proposed. And the fuzzy controller is realized using Freescale fuzzy inference machine. It is successfully applied in the automatic route tracking. The hardware design of the smartcar is introduced. Then, the process of establishing the fuzzy controller is described in detail, including the choice of fuzzy input and output v ariables, linguistic v alues, domain, input and output membership functions, rule base, fuzzification, rule inference, defuzzification. The validity of the designed controller is verified by MATLAB simulation and actual operating results.

Index Terms -fuzzy controller, automatic route tracking, smartcar, Freescale microcontroller.

I.I NTRODUCTION

‘Freescale cup’ smartcar contests have been successfully held in many countries. Many specialized knowledge are involved in the contest such as control science, pattern recognition science, sensor technology, automobile electronic, electricity, computer science, mechanics, and so on, as in [1].

A good long-term impetus to integrate the knowledge and practice abilities of the undergraduates has been brought out. And the contest can stimulate the development of automobile electronic technology. All the teams are required to use the same smartcar models provided by the organization committee. The microcontroller MC9S12DG128 is assigned as the core control unit, which is made in Freescale semiconductor. Participants are required to choose and design the control system, including acquisition and processing of sensor signals, control algorithm design and implementation, DC motor driver design, steering servo control, and so on. Upon completion of the smartcar control system design and debugging, the contest result largely depends on the speed of the car in the designated route. The smartcar’s speed is largely influenced by the smartcar’s route tracking time, as in [2].

Fuzzy control belongs to the intelligent control field. As the precise mathematical model of control object is not needed, fuzzy control has been widely used in industrial control area, as in [3].

The hardware design of the smartcar is introduced in this paper. The process of establishing the fuzzy controller is proposed in detail. The simulation of the designed controller based on MATLAB’s fuzzy logical toolbox is presented, which is used to verify the validity of the designed controller. The fuzzy controller is realized using Freescale fuzzy inference machine. It is used to control the smartcar’s steering servo motor. The practical operating result shows that the performance of the controller can meet the design requirements.

II.S YSTEM S TRUCTURE

The control system of smartcar is composed of 6 modules, power supply module, route information acquisition module, DC motor driver module, speed acquisition module, steering servo module, and microcontroller module. The control system structure is shown in Fig. 1.

Ni-cd batteries of 7.2V 2000mAh are used as power supply module to provide energy for the whole system. Regulators are used to transform 7.2V to 5V, 6V and 7.2V for the power supply of the rest modules. Route information is acquired from route information acquisition module. After being processed by the necessary hardware and software, the route information is input into the microcontroller. The microcontroller is responsible for dealing with route information processing, realizing the control algorithms and performing the system output control. The speed of the smartcar is adjusted through DC motor driver module. This module accepts speed control instructions and drives DC motor. Steering is controlled by steering servo motor module. Real-time speed is measured by speed acquisition module. It is used to realize precise closed loop control of the speed.

Fig. 1 System structure

The route information acquisition module is the core module to realize the automatic route tracking. Infrared light emitting diodes and photosensitive receiving triodes are used to acquire route information. Eleven pairs of optoelectronic sensors are placed in the front of smartcar in parallel to detect

Proceedings of the 7th

World Congress on Intelligent Control and Automation June 25 - 27, 2008, Chongqing, China

the route information. B lack track line is placed in the path. Reflection light intensity received by photosensitive receiving triodes is different when the triodes are over black or white track. The signals are amplified and filtered, and then are input into ADC module of microcontroller. According to the route information, the microcontroller calculates the deviating angle between the smartcar and the route track. The control algorithm module makes some calculation to get the system outputs, including object speed and servo steering angle, so achieves the target of automatic route tracking.

There are several kinds of controllers to complete the automatic route tracking of the smartcar, such as PID control, expert systems, fuzzy control, and so on, as in [1] and [3]. In the paper, a fuzzy controller based on Freescale fuzzy inference machine is used to control the steering of smartcar.

III. DESIGN OF F UZZY CONTROLLER

A. The basic principle of fuzzy control

A fuzzy control system is composed of assigned input, fuzzy controller, controlled object and feedback signal. A fuzzy controller is divided into four parts, as in [3, 4, 12, and 13]: fuzzification, rule base, rule inference and defuzzification. The structure of the smartcar’s fuzzy controller is shown in Fig. 2.

B. The basic requirements of fuzzy controller

The basic requirements of the fuzzy controller are,

1) When the smartcar is on straight track, steering servo motor should steer middle.

2) The smartcar should not run out the track.

3) When the smartcar is on the bending track, the angle of steering should increase with the increasing of the smartcar deviating angle. The steering angle output should be a bit larger than usual when speed is high.

C. The design of fuzzy controller

The fuzzy controller flow chart based on Freescale inference machine is shown in Fig.3, as in [5, 6, and 7].

Fig. 2 the smartcar’s fuzzy controller structure

Fig. 3 Fuzzy controller flow chart

1) Fuzzification

The process of fuzzification includes the determination of all system’s input and output variables and their domains, the number and values of each linguistic variable, the membership functions of each linguistic which decide every linguistic variable’s position and sharp, as in [4]. The fuzzification instruction of Freescale inference machine is MEM. The definition of system variables, their domains and the membership functions are described as following.

a) The system input are two variables: the first is the information acquired from route information acquisition module, which is named DepartureAng. The second is the real time speed acquired from speed acquisition module, which is named Speed.

b) The system output variable is named ServoAng. The output angle of steering servo motor ranges from -40 degree to 40 degree. The domain of ServoAng is quantified to 0~150, and its value is assigned as: Negative Large (NL), Negative Small (NS), Medium (Med), Positive Small (PS), Positive Large (PL).

c) The universe of discourse of DepartureAng is -45 degree to +45 degree. The 0 degree is supposed as the central position. The DepartureAng is quantified to domain [0,1,2,3,4,5,6,7,8,9,10], and its value is assigned as: Very Left (VL), Left (L), Middle (M), Right (R), Very Right (VR).

d) The domain of real time speed is 0~2500mm /s, and its value is assigned as in three class according to actual situation: Slow, Fast, Rapid.

e) The membership functions of linguistic variables are usually determined by triangle law, as in [3, 4, and 5]. The formula is presented as equation 1. The determination of membership function is shown in Fig. 4.

°°°

°°°?

-

<<≤

?≤

c a x

c x b b c x

c b

x a a b a x x A ;;0;;)(μ (1) (x A |ì1

input

Fig. 4 Determination of membership function

According to the practice experience, the definition of the membership functions of linguistic variables is shown in Fig.5.

(a) DepartureAng membership function

(b) Speed membership function

Fig.5 Membership functions of input linguistic variables

f) Membership function of output variable

Non-single-value function is used as membership function of ServoAng. While when writing microcontroller assembly program, single-value function is used, as in [4]. The definition of the function is shown in Fig. 6.

Fig. 6 Membership functions of ServoAng

2) Rule inference

a) Establishment of Rule Base

The basic rule of the inference is: IF X is A AND Y is B THEN Z is C, as in [4].

Where, ’X is A’ is antecedent 1, ’Y is B’ is antecedent 2, and ‘Z is C’ is consequent. Antecedents is connected by Symbol ‘AND’.

The rule base of smartcar control system is shown in Table I.

TABLE I R ULE B ASE

Antecedent1 DepartureAng

Antecedent 2 Speed

VL L M R VR

Slow NS NS Z PS PS Fast N B NS Z PS P B Rapid N B N B Z P B P B

follow: if DepartureAng is very left (VL) and speed is slow then ServoAng is Negative small (NS).

b) Inference

MAX-MIN law is used in fuzzy inference based on Freescale inference machine. The related instruction of Freescale inference machine is REV.

Fuzzy inference instruction sets up rule base, finds out the effective rules to figure out rule intensity, and outputs the fuzzy output.

3) Defuzzification

The results of fuzzy inference are fuzzy outputs. The crisp output can be got by defuzzification. In order to defuzzificate, the membership functions of outputs variables

are needed. Centroid law is usually used to realize defuzzication, as in [4, 5].

Formula of centroid law is shown in equation 2.

||===

n

n

1

1

i i i

i

i

F

F

S COG

(2)

In equation 2, F i is the result of inference. S i is the weight of output variables. COG is the crisp output of fuzzy controller.

IV. S IMULATION OF T HE D ESIGNED F UZZY C ONTROLLER Fuzzy logic toolbox of MATLAB software is used to simulate the designed fuzzy controller. In the MATLAB Fuzzy Inference System editor (FIS Editor), basic attributes are set as follows: ‘And’ method is set as prod; ‘Or’ method is set as probor; ‘Implication’ is set as prod; ‘Aggregation’ is set as max; ‘Defuzzification’ is set as centroid, as in [8,9,10,11].

I/O variables of fuzzy controller, as well as their membership functions are input into MATLA B by Membership Function Editor (MFE). The rule base shown in TABLE I is the input in Rule Editor (RE).

The surface view of the designed fuzzy controller is shown in Fig.7.

The output of fuzzy controller is checked through Rule Viewer as shown in Fig. 8.

When the Speed is set to 150cm/s, with the DepatureAng changing, ServoAng is changed at the same time. The results are shown in TABLE II.

If the real time speed keeps unchanged, the following conclusions can be drawn,

a) When DepartureAng is 5, which means smartcar is on the central of the track, ServoAng value is 75. That means steering is not needed.

b) When deviating from the track, the output of fuzzy controller changed synchronously. For example, if Departure=3, the output value will be 42.8.

c) With the smartcar’s deviation from central angle increasing, the output increases.

The characteristics of fuzzy controller above can satisfy the basic requirements of smartcar control system as described before.

When the DepatureAng is set to 3, with the Speed changing, ServoAng is changed at the same time. The results are shown in TABLE III.

It can be seen from TAB LE III that if the DepatureAng keeps unchanged, the output of fuzzy controlled increases with the increasing of smartcar’s real speed. This characteristic can satisfy the requirement when speed is high, even with the smaller deviation; output of the fuzzy controller is larger than

usual.

Fig.7 Surface viewer of the fuzzy controller

Fig. 8 Rule viewer of the output of fuzzy controller

TABLE II

O UTPUT O F F UZZY C ONTROLLER (1)

Inputs Output

DepatureAng Speed (cm/s) ServoAng

0 150 13.2 1 150 13.7 2 150 32.4 3 150 42.8 4 150 59.1 5 150

75

6 150 90.8

7 150 111

8 150 122

9 150 135 10 150

135

TABLE III

O UTPUT O F F UZZY C ONTROLLER (2)

Inputs Output DepatureAng Speed (cm/s) ServoAng 3 120 42.8 3 150 42.8

3 160 41.4

3 170 38.7

3 180 35.2

3 190 31.4

3 200 27.3

V. R EALIZATION OF

F UZZY C ONTROLLER

The designed fuzzy controller is based on Freescale

inference machine. The inference machine is supported by KB G software package, which makes the engineers design much easier, as in [4, 5]. Freescale inference machine is usually written with assembly language to ensure the desired performances. Freescale inference machine mainly includes

three parts. The definition of data structure and variables are the first part. The definition of input and output variables’ membership functions are the second part. Executing fuzzy

control inference is the third part. The flowchart of using Freescale inference machine is shown in Fig. 9.

Fig. 9 Flowchart of using Freescale inference machine

VI. C ONCLUSION

A fuzzy controller based on Freescale inference machine is designed in the paper. The validity of the designed

controller is verified by MATLAB simulation. Fuzzy

controller module is realized by using Freescale inference

machine instructions. The fuzzy control module is called by C

in main routine, which realizes the automatic route tracking

together with other control modules. Using the fuzzy controller, the smartcar system actually works well.

R EFERENCES

[1] Kaisheng Huang, Jin Huamin Jiang Dinan, O verview of Design for

Korea Intelligent Model Car Design Contest , Beijing: Tsinghua Freescale MCU/DSP Application Centre, 2004. (In Chinese).

[2] The Rules of 1st freescale cup undergraduates’ smartcar contest , Beijing: Tsinghua Freescale MCU/DSP Application Centre. 2006. (In Chinese). [3] Shiyong Li. Fuzzy Control Neurocontrol and Intelligent Cybernetics, Helongjiang: Haerbing industrial university publishing house, 1996. (In Chinese). [4] B eibei Shao , the O nline Development of the Embedded Application of

Microcontroller , Beijing: Tsinghua university publishing house, 2004. (In Chinese).

[5] S12XCPUV2 Reference Manual , V02, 2006, Freescale Inc.

[6] MC9S12DG128 Device User Guide , October 2002 Motorola Inc. [7] HCS12 CORE Datasheet , August, 2000, Motorola Inc.

[8] SHAO Gui_rong, “Matlab_based design and simulation of fuzzy control

system,” Journal of Chongqing Institute of Technology (Natural Science

Edition), vol. 21, no. 5, pp. 51-54+85, May 2007. (in Chinese).

[9] Wei wei, MATLAB Control Toolbox Handbook, B eijing: National Defense industrial publishing house, 2004. (in Chinese).

[10] Lotfi A, Zadeh, B erkeley, CA, “Fuzzy Logic Toolbox User’s Guide Version 2.1.1”, MathWorks, 2001.

[11] SIMULINK-Dynamic System Simulation for MATLAB, MathWorks,

1998. [12] Weiguo Zhang, The Theory and Application of Fuzzy Control, Xi'an:

Northwest industry university publishing house, 2000. (in Chinese). [13] Shuguang Liu, the Technology of Fuzzy Control. Beijing: China Textile

and apparel Press .2001. (In Chinese).