Social Cues in the Autonomous Navigation of Indoor Mobile Robots

Abstract

Robots are now operating in workspaces occupied by humans and the robots often need to avoid people while travelling to their desired goal. Humans follow some navigation conventions while walking, called as social cues, like avoiding an oncoming person from the left. It is important for the robots to display the same social cues to be socially accepted by the humans. Current works in social robot motion planning are limited to maintenance of a socially compliant distance. This paper exhibits a socialistic behavior of a robot avoiding an oncoming human by the preferred side. This paper extends the social force model to incorporate the social cues by adding new social forces. The paper also extends the geometric approach to incorporate the social cues by selecting the geometric gap as per the social preference. Finally, we propose a novel hybrid approach combining the social potential field and geometric method, wherein the preferred gap for navigation adds a new social force to the robot. The proposed approach maintains the proactive nature of the geometric approach as well as retains the reactive nature of the social force model. The hybrid method maintains a larger clearance and generates safer trajectories as compared to the baseline social potential field and follow the gap methods. The experiments are done with the Pioneer LX robot using vision cameras and a lidar for navigation operating at the Centre of Intelligent Robotics at IIIT Allahabad.