Each biohybrid unit fuses a living muscle-cell actuator with a rigid chassis. Its heading is the weighted sum of three local rules computed only from nearby swarm-mates (a Reynolds flocking model), the same class of rule used to design decentralized swarm-robot controllers:
v' = v + dt·(w_coh·steer_to(centroid)
+ w_align·steer_to(avg_heading)
+ w_sep·steer_away(neighbors < r_sep))
v = clamp(v', v_max)
- Swarm size — number of active biohybrid units instanced into the arena.
- Bio-signal cohesion — weight w_coh; how strongly units chase the swarm centroid, modelling chemical/mechanical signalling strength between living actuator cells.
- Actuation speed — the muscle-cell contraction rate, capping each unit's top speed v_max.
- Separation radius — the minimum spacing units try to keep, preventing chassis collisions.
- Disturb Swarm — injects a random impulse into every unit's velocity, testing how fast the decentralized rules re-synchronize the group.
Real biohybrid swarms (e.g. cardiomyocyte-driven microrobots) use exactly this kind of local, sensor-free coordination for search-and-rescue or environmental-monitoring tasks where no central controller can reach every unit.