Planning collision-free paths in time-varying environments

The authors describe an intelligent controller and a corresponding planner. The controller has been described in detail elsewhere (Meng and Picton (IFAC Symposium SICICA ´92 Malaga, Spain, May, 1992)), and so only a brief summary is presented. An important feature of this work is that the controller is developed with an artificial neural network (ANN). A remarkable property of multilayer feedforward type of ANNs is the ability to approximate any arbitrary nonlinear function. This makes the ANN an obvious candidate to use for complex system modelling. In the following sections, an AGV (autonomous guided vehicle) navigation system structure and navigation process are introduced first. Then the ANN controller background and interpretation are presented. This is followed by the simulation results of obstacle avoidance with ANN controller. Finally, the path planner and simulation results are further described