MDL-based posture stabilization for wheeled mobile robots with nonholonomic constraints

This paper addresses the problem of posture stabilization for a wheeled mobile robot with nonholonomic constraints. Due to the richness of mappings from sensor space to actuator signal space and the complexity of system and environment modeling, it becomes particularly difficult to select an appropriate control law for the posture stabilization problem. In this paper, we propose a novel approach based on Motion Description Language (MDL) for this problem. The method decomposes the control task into a finite collection of reactive behaviors and generates different control procedures through concatenation operation of predefined motion atoms. Each symbolic string defines an expected behavior while a complex control task can be achieved by combining several symbolic strings. As an illustration, parking problem of the wheeled mobile robot with nonholonomic constraint is studied and simulation results demonstrate the effectiveness of this method.