Hierarchical Hybrid Symbolic Robot Motion Planning and Control

Abstract

AbstractThis paper addresses the motion planning problem using hybrid symbolic techniques. The proposed approach develops a unified hierarchical hybrid control framework using a bismulation‐based abstraction technique over the partitioned motion space that can be applied to autonomous aerial robots (3‐D symbolic motion planning) or ground vehicles (2‐D symbolic motion planning). The bisimulation relation between the abstracted model and the original continuous system guarantees that their behaviors are the same. This allows to design a discrete supervisor for the abstracted model, and then, the designed supervisor can be applied to the original system while the closed‐loop behavior does not change. To apply the discrete supervisor to the original continuous system, an interface layer is developed, which on the one hand translates discrete commands of the supervisor to a continuous form applicable to the continuous plant and on the other hand, abstracts the continuous signals of the continuous low layer to discrete symbols understandable by the supervisor. The proposed algorithm is verified through implementation of a hybrid symbolic algorithm for the formation control of unmanned aerial vehicles.