Neural network-based optimal control for autonomous mobile vehicle navigation

This work addresses the problem of generating autonomously an optimal control action sequence for an autonomous vehicle based on artificial neural networks (ANN). The principal objective is to autonomously design an optimal controller that steers the center of the vehicle through a number of via points in a particular order using a minimum amount of time. In general, the steering of robotic vehicles depends on the interactions between the vehicle and its supporting medium. Planning for the future encounters with the via points should be part of the current control decision, since the vehicle's position and orientation as it moves through one gate greatly alter the case of navigation through successive points. The proposed neural network-based intelligent controller learns to guide the vehicle through a set of points autonomously. The simulation results show the performance of the proposed approach for a simple case.

[1]  J. Canny,et al.  Nonholonomic Motion Planning , 1992 .

[2]  Vicente Mut,et al.  Algorithms for stable control of mobile robots with obstacle avoidance , 1999 .

[3]  Richard S. Sutton,et al.  A Menu of Designs for Reinforcement Learning Over Time , 1995 .

[4]  Richard S. Sutton,et al.  Challenging Control Problems , 1995 .

[5]  Karl Mathia,et al.  Asymptotic dynamic programming: preliminary concepts and results , 1997, Proceedings of International Conference on Neural Networks (ICNN'97).

[6]  Antonio Bicchi,et al.  Closed loop steering of unicycle like vehicles via Lyapunov techniques , 1995, IEEE Robotics Autom. Mag..

[7]  R. Murray,et al.  Exponential stabilization of driftless nonlinear control systems using homogeneous feedback , 1997, IEEE Trans. Autom. Control..

[8]  Paul J. Werbos,et al.  Consistency of HDP applied to a simple reinforcement learning problem , 1990, Neural Networks.

[9]  Ken-ichi Funahashi,et al.  On the approximate realization of continuous mappings by neural networks , 1989, Neural Networks.

[10]  Derong Liu,et al.  A self-learning ship steering controller based on adaptive Critic Designs , 1999 .

[11]  Peter Veelaert,et al.  Ultrasonic potential field sensor for obstacle avoidance , 1999, IEEE Trans. Robotics Autom..

[12]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[13]  Paul J. Werbos,et al.  Approximate dynamic programming for real-time control and neural modeling , 1992 .

[14]  George Cybenko,et al.  Approximation by superpositions of a sigmoidal function , 1992, Math. Control. Signals Syst..

[15]  Richard M. Murray,et al.  A motion planner for nonholonomic mobile robots , 1994, IEEE Trans. Robotics Autom..