An Integrated Approach to Real-Time Mobile Robot Control in Partially Known Indoor Environments

In this paper we present a navigation method for mobile robots in partially known indoor environments based on integration of graph based search algorithms and dynamic window local obstacle avoidance method. With the attention on a dynamic environment three different graph based search algorithms for global geometrical path planning are considered and compared: A*, D* and focussed D* algorithm. The admissible robot trajectories are generated in the dynamic window local obstacle avoider module that ensures safe robot operation. A simple and efficient procedure to the selection of appropriate motion commands based upon alignment of acquired trajectories and global geometric path is proposed. The initial a priori knowledge is used about environment in the form of the occupancy grid map that is incrementally updated in runtime. The algorithms were verified both in simulation and experimentally on a Pioneer 2DX mobile robot using laser range sensor where a good correlation was proven.