Crowd-Comfort Robot Navigation Among Dynamic Environment Based on Social-Stressed Deep Reinforcement Learning

Robot navigation in a dynamic environment is a challenging task because not only the safety but also the comfort of surrounding pedestrians shall be necessarily considered. This paper proposes the concept of social stress based on tension space of robot and human, which is an important part of Human-Robot interaction. Especially, the proposed approach develops crowd-comfort navigation by combining social stress indexes with a deep reinforcement learning framework and the value network. A set of typical simulation experiments show that our method improves the comfort of surrounding pedestrians effectively during the process of robot navigation. In addition, the fine-tuned technology proposed in this paper has also been proven to be suitable for different scenarios.