Directional motion of a self-steering active intruder in a dense crowd of cognitive active agents

The fast and efficient directed motion of particles through crowded environments is challenging problem. In this work, the surface-bound motion of an intruder in a crowd of identical active agents is studied by overdamped Langevin dynamics simulations. Both intruder and agents are modeled as intelligent active Brownian particles (iABPs) with visual perception and directional steering to avoid collisions - which implies non-reciprocal interactions between all particles. The reorientation of intruder and agents is limited by their maximal maneuverability, which controls the ability of an iABP to adjust its velocity direction. The simulation results show that the intruder's attempt to increase directional speed by steering around agents fails; in fact, this even reduces the directional speed. In contrast, the intruder has to be perceived by the agents so that they can move out of the way in time. The intruder speed and transverse diffusivity are determined as functions of several key control parameters, like maneuverability, vision angle, and agent density. Here, an important parameter is the uniformity of the agent distribution. It is shown that the agent's self-steering to avoid collision enhances hyperuniformity (class III), which facilitates an easier directional navigation of the intruder. Results are relevant, inter alia, for the motion of emergency personnel in semi-dense human crowds.