A Path Planning Algorithm for an AUV Guided with Homotopy Classes

This document presents a demo storyboard for an accepted paper in the ICAPS 2011. The paper presents a method that builds a topological environment based on the workspace to compute homotopy classes, which topologically describe how paths go through the obstacles in the workspace. Then, the homotopy classes are sorted according to an heuristic estimation of their lower bound. Only those with smaller lower bound are used to guide a planner based on the Rapidly-exploring Random Tree (RRT), called Homotopic RRT (HRRT), to compute the path in the workspace. The demo consist of a short presentation where the theoretical concepts of our paper are explained and an execution of an application that applies the whole method in bitmap scenarios. Introduction This storyboard presents the practical part of a paper accepted in the ICAPS 2011 (Hernández, Carreras, and Ridao 2011). The paper presents an extension of the method to generate homotopy classes that can be followed in any 2D workspace. Two paths that share the start and the end points belong to the same homotopy class if one can be deformed into the other without encroaching any obstacle. The homotopy classes generated are sorted according their quality given by a lower bound estimator. In order to maximize the number of homotopy classes that can be explored, we propose a path planner based on the RRT, called Homotopic RRT (HRRT), to generate paths in the workspace. The HRRT algorithm starts looking for a path in the homotopy class that has a high probability of containing the optimal solution. Our method has been tested in several scenarios and this demo will show up all the execution steps during these tests. The demo consists of two steps. First, there will be a small presentation to introduce the goal of our research work, the theoretical concepts of homotopy classes and how are they generated in 2D workspaces. Then, there will be an application to run the tests proposed in the paper and see how our method overcomes each step of the process. The full paper describing this system appears in the ICAPS2011 Proceedings at pages 82-89. 1 α 2 β g