This paper describes an algorithm to reduce risks for personal mobility robots. The algorithm we propose is designed for mobility assistance robots that move around in everyday spaces such as pedestrian roads and shopping centers. Since such robots must be designed to remain safe while moving, we based our risk assessment on limiting their speed. Our algorithm allows for the completion of tasks safely based on the risk analysis, and can be used to calculate a speed limit for mobility assistance robots based on their environments. The algorithm is optimized for ensuring safe operation of practical life support robots. It is based on a 2D occupancy polar grid map updated in real time, with information on nearby obstacles. It can be adapted to various environments through the use of laser range finders, and is implemented in a low-cost microprocesser-based system. The implemented risk reducing algorithm has been evaluated through successive practicality tests at Robot Safety Center as well as public environments of the mobility robot special district at TSUKUBA, showing the validity of our approach.
