Probabilistic time-dependent models for mobile robot path planning in changing environments

Abstract

In the context of mobile robot path planning, a common strategy is to assume that the world is static and rely on heuristic approaches and obstacle avoidance to deal with the changes in the environment. When planning, not taking the potential changes of the environment into account usually leads to poor performances. In this paper we propose a probabilistic model that explicitly characterizes the traversability of the environment as a stochastic process. Furthermore, we present a path planning approach that exploits this traversability information to compute paths that minimize the expected travel time of the robot. Experimental results show that by explicitly modeling and reasoning about changes in the environment path planning performance can be improved.