89-3 08:30 - 08:45 Emergent tunnel branching in Coptotermes formosanus (Shiraki) through group interaction and stigmergic signalling. Bardunias, PM*; Nilles, AQ; Greenberg, RG; Goldman, D; Petersen, KH; Cornell University; Cornell University; Freehold Township High School; Georgia Institute of Technology; Cornell University email@example.com
Tunneling termites work in rotations of individuals that take soil from the end of a tunnel while others wait in turn to excavate. The geometry of these branching tunnel networks is crucial for efficient search and transport. Branches could result from high-complexity cognitive processes or pheromone-mediated communication; however, evidence suggests that branching events are triggered by traffic congestion. Tunnels widen as individuals in queue to reach the front dig from the walls until separate tunnel fronts form. This is a memory-free process for individual termites, with collective memory stored stigmergically in the tunnel. Impressively, this produces branching patterns are optimal for searching for patchy resources that termites exploit. To demonstrate that such a mechanically-mediated, memory-free branching process is feasible, we introduce a robophysical model consisting of sensorless, vibrating bots that move slowly down a 15o incline and carve channels through sand. These simple bots showed branching patterns in the sand that were governed by the local interactions between bots in channels, the local interaction between bot heading and channel geometry, and stigmergic information transfer through the piling of sand on the walls of the channel. These findings led us to examine individual interactions and stigmergic signaling in live termites that led to branch formation by characterizing the dimensions of tunnels and critical angles of tunnel walls that elicit excavation. Based on these measures, groups of 50 workers were presented with preformed tunnels to determine the combinations of size and angle that most reliably led to branch formation. Understanding how solutions emerge to complex cognitive tasks from the interactions and indirect communication of groups of individuals may inform the construction of robot swarms.