A framework for planning motions in stochastic maps

In this paper we propose a framework for motion planning in stochastic map. Most of the recent planners are good enough to solve motion planning problems. However, they need a complete and accurate model of the environment and such an assumption may cause a collision in executing the results in a real world with its uncertainties. Considering uncertainties in the model of environment, we reformulate the path planning problem in a stochastic map and then propose a way to modify classical path planning methods in order to fit into this new framework. Our work shares ideas with previous work in this area but follows a different approach. In this framework, sensors and landmarks need to be taken into account. The core computations lie in the evaluation of the probability of collision of configurations with the map.