Dynamic Social Zone based Mobile Robot Navigation for Human Comfortable Safety in Social Environments

Abstract

We propose an effective human comfortable safety framework enabling a mobile service robot to navigate safely and socially in social environments. The proposed framework takes human states (position, orientation, motion and hand poses) and social interaction information relative to the robot into account to model extended personal space and social interaction space, respectively, the combination of which results in a dynamic social zone (DSZ). The DSZ-based human comfortable safety framework is able to estimate an approaching goal pose of the robot for a human or a group of humans, thus allowing the robot to not only avoid but also to approach a human or a group of humans in a socially acceptable manner. The DSZ is incorporated into the robots motion planning system comprising the D* planner technique and dynamic window approach algorithm to generate motion control commands for the mobile robot. We verify the effectiveness of the proposed method through simulation and experimental results under the newly proposed human comfortable safety indices.