Human-inspired robot navigation in unknown dynamic environments

This paper proposes a novel algorithm for mobile robot navigation in completely unknown dynamic environments involving humans. Unlike existing navigation algorithms that assume full knowledge of the environment, i.e. position, direction and velocity of all obstacles, the proposed algorithm is only based on the local sensory information. The algorithm, inspired from pedestrian studies, has a proxemics-based local reactive planner that works on a cognitive representation of the environment. It enables the robot to mimic human-like responses in navigation. The efficiency of the algorithm is studied and analyzed through extensive quantitative simulations for different complexities of the environment and the results exhibit a good performance.