Distributed Network Sensing as an Escape Strategy

Abstract

The response of a swarm of ants or birds, living tissue of cells, or a robotic swarm, to spatial signals from the environment as a whole is a fundamental challenge for these decentralized systems. A promising alternative is embodied distributed sensing and decision-making, so that local environmental cues directly shape coordinated reconfiguration, growth, or movement without centralized control. Fluidic networks offer a physical basis for such systems, yet existing implementations are predesigned and limited in their ability to process spatial information across extended systems. Here, we investigate how fluid transport networks enable environmental sensing and adaptive response using the slime mould Physarum polycephalum. By exposing Physarum vein networks to dynamic pH environments, we show that the organism's ability to respond to aversive stimuli is governed by its spatial extent and network connectivity. Above a certain size, organisms with network morphology reliably detect and escape harmful conditions by network reconfiguration, whereas disconnected or spatially confined organisms fail to respond. In environments with competing stimuli, network morphologies enable integration of spatial information and repositioning into favorable regions. We identify local contractions as a physical mechanism that drives fluid redistribution and converts local stimulation into system-level responses. Finally, we generate different morphologies using fluidic and electrochemical control, demonstrating large-scale adaptation to complex environments. These results establish distributed fluidic networks as a framework for integrated sensing and actuation, providing design principles for autonomous soft robotic systems in dynamic environments.