Download presentation
Presentation is loading. Please wait.
Published byGloria Richardson Modified over 8 years ago
1
1 Algoritmos Genéticos aplicados em Machine Learning Controle de um Robo (em inglês)
2
2 Robot Control using Genetic Algorithms
3
3 Summary Introduction –Robot Control –Khepera Simulator Genetic Model for Path Planning –Chromosome Representation –Evaluation Function –Case Studies Conclusions
4
4 The Robot Controller Problem Given a robot and a description of an environment, provide commands (motor speeds) to the robot, in order to achieve a path between two specified locations, which is collision-free and satisfies certain optimisation criteria. (x i, y i ) (x f, y f )
5
5 Optimisation Criteria Robot should: –attempt near-optimal paths –avoid obstacles –perform straight motion Controller should be independent of: –the robot’s environment –target location
6
6 The Khepera Simulator Freeware mobile robot simulator (designed by Olivier Michel, University of Nice Sophia-Antipolis) User designed worlds Control algorithms can be written in C/C++ Robot’s position and angle reading 8 sensors (S0-S7): [0, 1023] 2 motors (M1, M2): [-10, +10]
7
7 Simulator Readings: sensors, position and angle S0-S7: [0, 1023] 1000 X Y Robot’s World angle of the robot with the world : [- , ] x y 0 obstacle not obstacle very detected closed
8
8 Control Mode To evolve the robot’s attitudes as it interacts with the environment Each robot action determines: –how well the controller performs with respect a given task; –the next input stimuli to the controller. learnThe controller should learn as the robot interacts with the environment
9
9 Controller Model GeneticAlgorithm evolves robot’s attitudes Sensors Position Robot’s Angle Goal Location Motor 2 Motor 1 KheperaSimulator
10
10 Proposed Model based on human behavior Obstacle detected IF Obstacle detected THEN Avoid collision, forget target ELSE Straight to the target according to the target direction END
11
11 Sensors Reading Simplification
12
12 Determining the Target Direction Direction =
13
13 Model ((S left > L) or (S right > L) or (S back > L)) IF ((S left > L) or (S right > L) or (S back > L)) THEN Obstacle detected, avoid collision, forget target Proximity-sensor = highest value (S left, S right, S back ) ELSE Obstacle not detected (collision-free), straight to the target Target direction = - END L=collision threshold=900
14
14 Genetic Algorithm Modelling Problem Chromosome Representation Evaluation Function Genetic Operators Techniques Parameters
15
15 Chromosome Representation Which speed should be imposed to each motor in each situation the robot is?
16
16 Evaluation Function Main objectives: –(V) speed: as high as possible –(D) straight motion: same motor speed for M1 e M2 –(A) action: reach a target and avoid obstacles Calculated based on the contribution for each gene [1,7], at each step.
17
17 Speed Normalised sum of the absolute value of the motors speeds; Vi increases as both speeds increase Whatever the robot does, it does quickly.
18
18 Straight Motion It favours high positive speeds to both motors When the robot is not oriented to the target (2,3,4), D=1 avoids contradictory learning
19
19 Action It considers the benefit of each gene regarding to: –obstacle avoidance –target closeness TPi = total of steps executed by attitude i AAi=action’s fitness at stept of attitude i
20
20 Action It considers the benefit of each gene regarding to: –obstacle avoidance –target closeness TPi = total of steps executed by attitude i AAi=action’s fitness at stept of attitude i Rates the distance variation to the target between two consecutive steps, and the maximum distance in one step, for collision free/front
21
21 Action It considers the benefit of each gene regarding to: –obstacle avoidance –target closeness TPi = total of steps executed by attitude i AAi=action’s fitness at stept of attitude i Rates the angle variation between two consecutive steps, and the maximum angle in one step, for collision free left, right, back
22
22 Action It considers the benefit of each gene regarding to: –obstacle avoidance –target closeness TPi = total of steps executed by attitude i AAi=action’s fitness at stept of attitude i Increases as the distance to the proximity-sensor increases in the step
23
23 Improving the Target Direction Model 4 possible target directions 0 /4 /2 3 /4 /2 -3 /4 - /4 - /2 8 possible target directions
24
24 Chromosome Representations
25
25 Genetic Algorithm Integer chromosome Population Size =100 Generations = 50 Crossover Rate = 80 % Mutation Rate = 4% Roulette Wheel Reproduction Elitism Linear scaling of fitness 300 Evaluation Steps for each chromosome Average of 25 Experiments
26
26 Genetic Algorithm Performance 7 Genes Chromosome
27
27 Genetic Algorithm Performance 7 Genes Chromosome
28
28 Genetic Algorithm Performance 11 Genes Chromosome
29
29 Genetic Algorithm Performance 11 Genes Chromosome
30
30 Paths Achieved in World 1 Case Study 1 7 Genes Chromosome11 Genes Chromosome
31
31 Paths Achieved in World 1 Case Study 2 7 Genes Chromosome11 Genes Chromosome
32
32 Speed Comparison 11 Genes Chromosome 7 Genes Chromosome
33
33 Paths Achieved in World 2 Case Study 1 7 Genes Chromosome11 Genes Chromosome
34
34 Paths Achieved in World 2 Case Study 2 7 Genes Chromosome11 Genes Chromosome
35
35 Speed Comparison 11 Genes Chromosome 7 Genes Chromosome
36
36 Speed Comparison (%) Case Study 1 Case Study 2 Case Study 3
37
37 Paths Achieved in World 3 Case Study 1 7 Genes Chromosome11 Genes Chromosome
38
38 Paths Achieved in World 3 Case Study 2 7 Genes Chromosome11 Genes Chromosome
39
39 Paths Achieved in World 3 Case Study 3 7 Genes Chromosome11 Genes Chromosome
40
40 Paths Achieved in World 3 Case Study 4 7 Genes Chromosome11 Genes Chromosome
41
41 Speed Comparison
42
42 Conclusions A simple GA was able to gradually evolve the robot control The robot achieved near optimal path towards the goal, avoiding obstacles Retraining is not necessary when the environment changes Controller improved performance with the 11 genes model The robot has no memory about previous unsuccessful paths and may get lost Other tasks can be included in the model (e.g. energy supply) Chromosome codification is limited for few robot’s situations
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.