A human-friendly robot navigation algorithm using the risk-RRT approach

Abstract

According to the proxemics, it is very important for pedestrians to keep themselves out of the personal spaces of others. The encroachment of the personal space usually causes discomfort, anxiety, or anger. In the human robot coexisting environment, the mobile robots are also expected to navigate like human beings and obey the motion conventions of human beings. Inspired by that, a pedestrian discomfort function is proposed to represent the discomfort value of pedestrians. A Comfort and Collision Risk (CCR) map is proposed to unify the discomfort value of pedestrians and the collision risk between the robot and the static barriers in one map. A human-friendly robot navigation algorithm is proposed by using the CCR map and the Risk-based Rapidly exploring Random Tree (Risk-RRT) algorithm. A multi RGB-D cameras intelligent robot system is constructed in this paper for experiments. Experimental results in both simulations and realistic environments reveal that our proposed method can achieve a good performance in human robot coexisting environment.