Environment representation using enclosed obstacles and minimum-turns path planning

Abstract

This paper presents a searching algorithm and a tree-type two-dimensional representation of the environment for solving the path-planning problem of a mobile robot navigating in a known space. We propose: (i) a search algorithm that chooses paths by the number of turns needed to reach the goal and, from these, the best one in terms of distance and/or rotation angle, and (ii) a tree-type representation of the environment that results from considering obstacles to be enclosed by their peripheral rectangle and from defining different types of obstacles and vertices. This representation does not expand the obstacles to guarantee a collision-free solution as conventional methods do- a local reactive planner is used instead. Our proposal relies on the following two empirical facts. (i) Distance is not a priority in places where solution paths are relatively short - time and energy must be considered instead. (ii) Calculations are simplified when all obstacles have the same structure. To simplify the path-searching process even more, the Searching Space was reduced to what we have called the INSIDE zone. This reduction resulted in considering fewer nodes when looking for a solution path. A comparison with other approaches and simulation results are also presented.