Collective Behavior in Active Swarms: Dynamics at the Edge of Disorder

We propose a minimal flocking model based on the inertial spin model for agents with vision-based steering interactions -- alignment, group cohesion and local collision avoidance. Numerical simulations in two-dimensions show that the local avoidance interaction induces a transition from polar order to disorder, triggered by the fast response of the flock to local changes in direction, thus acting as a source for emergent noise. Near this transition, vision-based cohesion interactions create a feedback loop between local density, alignment, and motion, leading to significant shape fluctuations of the flock, along with a change in the collective motion of the flock from ballistic to diffusive. In the overdamped regime, deep in the ordered phase, the vision-based cohesion interaction is shown to act similar to a surface tension, driving the flocks toward circular configurations. The universality class of the avoidance induced transition is different from the Vicsek-type transition. Notably, the emergence of large vortices and an exponential scaling of the correlation length, suggest some similarities to the Berezinskii-Kosterlitz-Thouless transition. The strong dependence of flock behavior on parameters like local avoidance can enable rapid responses to external cues, aiding predator evasion and foraging in biological systems.