A Model Predictive Navigation Approach Considering Mobile Robot Shape and Dynamics

Abstract

Most mobile robot navigation approaches assume the robot being point-like or consider only its bounding circle while looking for a collision-free path to a given goal position. A well-known method called the Dynamic Window Approach (DWA) introduced an interesting idea for solving the navigation problem by local optimization in the control space of the robot. Some extensions of the original DWA method can also be found in the literature, which enable its applicability to holonomic and non-holonomic robots and ensure a global and safe solution to the navigation problem. The method described in this paper has also been motivated by the basic idea of dynamic window and contributes to the previous variants by taking the robot shape into consideration as well. A navigation function based model predictive control scheme is utilized to choose the appropriate control for a safe and successful navigation.