Abstract
An electric wheelchair for indoor use should automatically avoid obstacles in its path and stop at hallway intersections to await instructions from the user. Here we propose a path control method in which an environment map including indoor hallway intersection information is constructed and self-position is estimated in combination with a particle filter. In this way, an electric wheelchair is semi-autonomously controlled. Using data from a laser range finder, the center of an intersection is estimated from the collected intersection information by using our intersection recognition technique. Through use of the intersection map and intersection recognition technique, the initial position from the particle filter can be estimated; this has the benefit of using the intersection itself as the target. Also, there is no need to change the environment in advance by positioning artificial landmarks, thus allowing cumulative error to be removed. This proposed method is applied to a modified electric wheelchair and the effectiveness is examined experimentally.
