Abstract
This paper proposes a safe path generation based on a gradient method, and obstacle avoidance in an occluded area which is invisible area for the mobile robot. The gradient method is most widely used method for path generation; however, it could be plausible that collision may occur with speedy invisible dynamic obstacles. Therefore, in this paper, occluded humans can be detected using sensor installed in the environment and posterior existence probabilities of humans are calculated to redefine the cost function of a gradient method. By using the redefined cost function, safe path taking into account dynamic obstacles′ posterior behaviors to the current states is generated. Experimental results with real robot show that the proposed scheme works successfully in the real environment.
