Vehicles Avoidance and Identify System

1
?????????????????????????? (Vehicles Avoidance
and Identify System)

• TRS

2
??????

• ??????????
• ?????????????? (Sensors)
• ??????????????????? (Avoid Obstacle System)
• ????????????? (Identify System)
• ????

3
?????????????????????? (Velocity Sensors)

• ????????????????????????????????????? (Inertia
  Navigation Systems) ???????? gyroscope
  ?????????????????? (acceleration)
  ????????????????? (x, y, z) ????????? integrate
  ??????????????????? ???????????????????????????
  (???? DMU-6x)
• ??????????????????????????? Optical Encoders
  ?????????????????????????????????????????????

4
???????????????????? (Heading Sensors)

• Gyroscopes provides angle relative to robot, but
  are not good when the robot is turning very
  slowly.
• Compasses
• Magnetic Compasses
• Halls effect Compasses
• Fluxgate Compasses (i.e. KVH C100)

5
?????????????????? (Distance Sensors)

• Ultrasonic or Sonar Sensors measures the time
  elapsed between the transmission of a signal and
  the receiving of an echo of transmitted signal
  (time of flight) to determine the distance to an
  obstacle.
• Laser Sensors An optical measurement sensor
  system with a useful range of 0.3 to 50 m. for
  most diffuse reflective surfaces. It operates by
  emitting a collimated laser beam that is
  reflected from the target surface and collected
  by the sensor. The sensor is suitable for a wide
  variety of distance measurement applications that
  demand accuracy and fast response times.

6
Laser Sensors
Comparison of commercially available scanning
laser rangefinders
7
????????????????????? (Position Sensors)

• Dead Reckoning a simple procedure for
  determining the robots position. The most
  simplistic implementation of dead reckoning is
  called odometry which involves optical encoders
  directly coupled to the motor armatures or wheel
  axles.
• GPS a satellite navigation system. The GPS
  receiver will be considered as a global
  navigation device for mobile robots.

8
Feature Extraction
Each fuzzy input is given by (e.g. sonar sensor)
Construct three membership functions for each
input
9
Input MFs
ZR MD
LG
10
MFs for Output 1 (w)
NB N Z P PB
11
MFs for Output 2 (v)
NB,N, P, PB
12
Rules for Output 1 (w)
13
Rules for Output 2 (v)
14
If-Then Rules
15
FIS 1
16
FIS 2
17
The output surface of a fuzzy system for angular
velocity and linear velocity
linear velocity
angular velocity
18
??????????????????
19
Avoid-Obstacle Behavior
--- Agent
Avoid-obstacle (Fuzzy)
S
Base Server
Avoid-obstacle (VFF)
S
20
Learning Fuzzy Rules (Cont.)

• Reinforcement Learning Gain

21
Learning Fuzzy Rules (Cont.)

• Integrate a Set of Fuzzy Rules

22
????????????????????

• ??????????
• Sonar ??? Laser
• ?????????????? fuzzy inputs
• Obstacle force from all obstacle distances
• Closest obstacle distance
• Closest obstacle direction

23
Learning Fuzzy Rules

• Fuzzify the sensory readings
• Obstacle force direction (q )
• Closest obstacle distance (d )
• Closest obstacle direction (q )
• Fuzzify the motor action
• Linear velocity (v)
• Angular velocity (w )
• Learning to built a set of fuzzy rules

force
min
min
24
Learning Fuzzy Rules
Fuzzify
Max(.)
Max(.)
Obst. Force (q )
Linear Velocity
force
Xin-1
Yin
Closest obst. (d )
min
Angular Velocity
Closest obst. (q )
min
Inputs
Outputs
Rules
Example of Fuzzy Rules
25
Example of fuzzy rules for avoid-obstacle
behavior
Fuzzy Control
26
Fuzzy Control
Adaptive fuzzy rules
Defuzzify
Fuzzify
Max(.)
Obst. Force (q )
Linear Velocity
Xin
force
Yin
Closest obst. (d )
min
Angular Velocity
Closest obst. (q )
min
Inputs
Outputs
Rules
27
Fuzzy Control
Adaptive fuzzy rules
Defuzzify
Fuzzify
Max(.)
Obst. Force (q )
Linear Velocity
Xin
force
Yin
Closest obst. (d )
min
Angular Velocity
Closest obst. (q )
min
Inputs
Outputs
Rules
28
????????????????
Ex1 ??????????????????????????????????????Ex2
?????????????????????????????????