New technique to build probabilistic roadmap methods

In this work a new alternative is proposed to improve the quality of the roadmap obtained by probabilistic roadmap methods. This alternative is based on the use of some geometric features of the robot and the obstacles to move and rotate the robot in some way. We believe that narrow corridor problems can be solved taking advantage of the robot form and its geometric features. The main idea is to obtain information that allows the algorithm to guide the search of useful configurations, and to reduce the account of non-valid configurations to be calculated. In order to reach a better performance of these planners, we obtain a better representation of the connectivity of the configuration space. We develop this idea to solve the motion planning problem for a specific kind of robot, the "free flying objects" with six degrees of freedom. We show results that allow us to be sure that geometric features can be used to build heuristics which can find free configurations close to the obstacles and such heuristics can be used to solve the narrow corridor problems.