Reflective collision avoidance for mobile service robot in person coexistence environment

A person coexistence environment is always changing. For example, people in the environment move around, someone changes the arrangement of the furniture, and someone puts something on the floor. Tasks of a service robot to support people include many scenes that the robot should move around in this environment. Without sensing the objects in the environment, it is difficult for an autonomous robot to move in the person coexistence environment protecting a people's safety. Thus, an advanced ability to avoid collisions by adapting to the environment is required for a robot. From the viewpoint of robustness, a reflective collision avoidance algorithm has been used with predefined map-based path planning. However, because the algorithm is composed of simple operation rules, a robot depending on the reflective collision avoidance algorithm may often moves inefficiently. To ensure efficient collision avoidance, we propose a reflective collision avoidance technique that corrects direction and magnitude of robot velocity independently using 1- dimensional potential-like functions ruled by a pseudo distance that modifies its own distance in proportion to an angle from the robot traveling direction to the obstacle direction. We verify the effectiveness of the proposed technique for a robot developed to serve persons in a retail store.