This paper presents an autonomous guidance system of wheeled tractor-like-robot on slope terrain. A neural network (NN) vehicle model was developed to represent the input-output relationship of vehicle motion on slope. The model was trained using Back Propagation algorithm. A cost function was also designed. Genetic algorithms were used to search the optimal steering values for different combination of lateral and heading deviations. Based on those values of the optimal steering, a NN-based steering controller model was developed to generalize the optimal steering for different land-inclination. Autonomous travel tests were conducted with a prototype test tractor along predetermined rectangular path on 15 deg of angle sloping land. The tractor could precisely follow the path. The mean and standard deviation of offsets along four linear directions of the rectangular path were 0.058 m and 0.063 m respectively, which are insignificant for tractor motion on agricultural farms.
Neural network based steering controller for tractor-like robot
Self-Optimizing Mechatronic Systems: Design the Future, International Heinz Nixdorf Symposium, 7 ; 347-359
2008
13 Seiten, 7 Bilder, 1 Tabelle, 12 Quellen
Aufsatz (Konferenz)
Englisch
Feedforward Neural-Network Controller for Ship Steering
British Library Conference Proceedings | 1995
|