Noise-robust navigation from an adaptive run-and-tumble policy
π arXiv:2608.27751 Β· π₯ PDF Β· 2026-08-27 Β· cond-mat.stat-mech
Authors: Aniruddha Datta [arXiv Β· scholar] , Shiladitya Banerjee [arXiv Β· scholar]
π Abstract
How do organisms navigate when the signals guiding them are noisy? Variance adaptation, the rescaling of sensitivity to noise, is common in sensory systems, but its role in navigation is unexplored. We introduce a minimal active Brownian particle whose run-and-tumble policy follows from an optimality principle. Variance adaptation emerges as part of this policy. Adaptation keeps chemotactic drift finite as noise grows, while a non-adaptive particle's collapses exponentially. Adaptation also carries a cost, degrading performance in quiet environments and requiring a tuned adaptation sensitivity.