Abstract
Mobile robots in living spaces for humans are required to recognize their surroundings well to avoid collisions with pedestrians. In this paper, we propose an autonomous robot which can move safely in such dynamic environments. The proposed robot can recognize surrounding pedestrians' and its own positions in the environment using distance information obtained by a laser range finder. Furthermore, positions of the pedestrians at the next time step are predicted by Kalman filter. Then, routing decision is made by using the potential field method whose repulsion field is decided with consideration for velocities of the pedestrians. Experimental results show that the proposed robot can move to its destination without a collision with pedestrians.
