Real-time navigation in dynamic human environments using optimal reciprocal collision avoidance

In this paper, the navigation strategy that a service robot navigates in dynamic human environments only relying on its own sensors is studied. Because of the limitation of partially-observable first-person perspective, the uncertainties of robot localization and estimation of people's states are increased, which blocks the navigation decision for a service robot. To solve this problem, a local collision avoidance method based on optimal reciprocal collision avoidance (ORCA) is proposed. The states of multiple pedestrians are estimated by combining a variant of particle-PHD filter for multi-target tracking with constant velocity motion model. To reduce the uncertainties, an encircling-particles method is proposed to refine the true states of robot and pedestrians from the probabilistic particle distribution. The effectiveness of the proposed technique is demonstrated through experiments in real environments.