In the article a new approach to a reactive navigation of a wheeled mobile robot (WMR), using a neural dynamic programming algorithm (NPD), is presented. A proposed discrete hierarchical control system consists of a trajectory generator and a tracking control system. In the trajectory generator we used a sensor-based approach to path design for the WMR in an unknown 2-D environment with static obstacles. The main part of the navigator is an action dependant heuristic dynamic programming algorithm (ADHDP), that generates control signals used to design a collision-free trajectory, that makes reaching a goal possible. ADHDP is the discrete algorithm of actor-critic architecture, that works on-line and does not require a preliminary learning or a controlled system knowledge. The tracking control system realises the generated trajectory, it consists of dual-heuristic dynamic programming (DHP) structure, PD controller and the supervisory term derived from the Lyapunov stability theorem. Computer simulations have been conducted to illustrate the performance of the algorithm. © 2012 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Hendzel, Z., & Szuster, M. (2012). Neural dynamic programming in reactive navigation of wheeled mobile robot. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7268 LNAI, pp. 450–457). Springer Verlag. https://doi.org/10.1007/978-3-642-29350-4_54
Mendeley helps you to discover research relevant for your work.