Motion Planning Using Fast Marching Squared Method

Abstract

Robotic motion planning have been, and still is, a very intense research field. Many problems have been already solved and even real-time, optimal motion planning algorithms have been proposed and successfully tested in real-world scenarios. However, other problems are not satisfactory solved yet and also new motion planning subproblems are appearing. In this chapter we detail our proposed solution for two of these problems with the same underlying method: non-holonomic planning and outdoor motion planning. The first is characterized by the fact that many vehicles cannot move in any direction at any time (car-like robots). Therefore, kinematic constrains need to be taken into account when planning a new path. Outoor motion planning focuses on the problem that has to be faced when a robot is going to work in scenarios with non-flat ground, with different floor types (grass, sand, etc.). In this case the path computed should take into account the capabilities of the robot to properly model the environment. In order to solve these problems we are using the Fast Marching Square method, which has proved to be robust and efficient in the recent past when applied to other robot motion planning subproblems.