A Game of Social Forces: Integrating Non-cooperative Game Theory with Social Force Model for a Socially-acceptable Mobile Robot Navigation

Abstract Robot integration in daily life demands research on both safety and social acceptance. Current methods focus on safety, but social factors are understudied. Moreover, existing studies lack deep analysis of human perception towards robot movement. Here, we present a novel navigation approach based on the combination of Game Theory and the Social Force Model (GTSFM) to bridge these gaps. We model navigation as a non-cooperative game to consider both pedestrians and robot as rational agents influencing each other’s choices. We evaluate the social acceptability of the GTSFM algorithm from both quantitative and qualitative perspectives. In both evaluations, the GTSFM is compared against two state-of-the-art algorithms: the social force model (SFM) and the optimal reciprocal collision avoidance (ORCA). According to the quantitative analysis performed in simulation, the GTSFM outperforms the SFM in all considered performance metrics and ensures higher performance than ORCA considering the smoothness of the trajectories and the proximity to pedestrians. The qualitative measurement is performed through a real-world experiment using a questionnaire administered to a pool of 76 participants. Our qualitative analysis revealed no statistically significant differences in performance between the algorithms tested. This lack of distinction may be due to unaccounted factors. The robot’s appearance and the limited velocity of the real robot could have obscured the distinction between the algorithms. These results represent a significant milestone in advancing the integration of robots into social environments also leave important hints for future research.