Computationally Efficient Path Planning for Wheeled Mobile Robots in Obstacle Dense Environments

Abstract

Smooth path generation for nonholonomic wheeled mobile robots (WMRs) has been researched significantly in the recent years. The nonholonomic constraints of WMRs impose difficulties for effective path planning; on top of this, the presence of static and/or dynamic obstacles in operation environments complicates the problem further. Alternative solutions have been proposed for WMR trajectory planning ranging from graph search methods [1, 2] to the application of potential function based techniques [4, 3, 5, 6]. Many of these methods assume full observability of the operational space [1, 3, 5] and some cannot provide dynamical path planning [3, 5], which is impractical for general applications of WMRs. Recently more advanced methods have been presented that offer better solutions to the path planning problem for obstacle cluttered environments [7, 8]. However, these methods utilize triangulation based mappings of the nearby environment by expensive sensory devices. This increases the computational cost of the planning algorithms, and necessitates more expensive electronics like laser scanners. A qualitative revision of the relation between the sensing abilities of wheeled mobile robots and the topology of their environment can be found in [9, 10].