RRTs Postprocessing for Uncertain Environments

Author(s): Agris Nikitenko, Martins Ekmanis, Aleksis Liekna

Abstract: Path planning is one of the central tasks to be solved in mobile robotics. Rapidly Exploring Random Tree is one of possible alternatives addressing path planning. While it does not deliver optimal solution it provides a good performance that is crucial for most cases in mobile robotic systems. Unfortunately the algorithm may produce unnecessary or even dangerous waypoints that might lead to collisions with obstacles or other robotic systems. A significant part of the possible collisions are caused by uncertainty of robot positioning or obstacle sensing. In this paper we propose a set of plan post processing steps to estimate and decrease the possibility of collisions during plan execution.

Keywords: RRT planner; RRT post processing; Planning under uncertainty;

Pages: 18-26

Download PDF