Orthogonal Steering and Delay Organize the Phase Structure of Flocking

Beomseo Song

Keywords: collective behavior, delay reaction, anisotropic vision, biomechanics, ecology, global dynamics

TL;DR: Made by AIs

Abstract: Bird flocks transition among distinct dynamical phases as ecological context changes. We introduce \textbf{AVES} (Anisotropic Vision–Energy–Steering), a continuous–time mechanistic framework in which sensory inputs are filtered by anisotropic vision with occlusion, acted upon after a finite reaction delay, and converted into roll-limited turns under a curvature ceiling set by lift–gravity balance; speed adapts under an explicit energy budget. Steering cues are orthogonally projected onto the plane normal to heading so that only biomechanically feasible rotations are executed. We embed delay as a finite-dimensional augmented state, enabling continuous-time propagation with discrete, frame-rate observations and thus a well-posed likelihood for Kalman filtering and simulation-based inference. This delay-aware state-space representation makes the four dimensionless controls—reaction delay, bank factor (curvature budget), interaction range, and vision half-angle—statistically identifiable from trajectories and comparable across species and habitats because the parameters and indices are dimensionless and share a common observation model. Phases are diagnosed using a global alignment index, a milling index, connectivity-based fragmentation, and residence-time statistics for dwell and switching. The theory yields testable predictions—e.g., narrower vision or longer delay expand intermittent/fragmented regimes while stricter roll limits suppress milling—that map to standardized field measurements and translate into design rules for bio-inspired collectives (e.g., maintaining a low social-delay number and enforcing curvature caps to preserve alignment), with applications to bird-strike risk assessment, conservation monitoring, and resilient UAV swarms.

Submission Number: 106