Characterization of the Design Space of Collective Braitenberg Vehicles

Large collectives of artificial agents are quickly becoming a reality at the micro-scale for healthcare and biological research, and at the macro-scale for personal care, transportation, and environmental monitoring. However, the design space of reactive collectives and the resulting emergent behaviors are not well understood, especially with respect to different sensing models. Our work presents a well-defined model and simulation for study of such collectives, extending the Braitenberg Vehicle model to multi-agent systems with on-board stimulus. We define omnidirectional and directional sensing and stimulus models, and examine the impact of the modelling choices. We characterize the resulting behaviors with respect to spatial and kinetic energy metrics over the collective, and identify several behaviors that are robust to changes in the sensor model and other parameters. Finally, we provide a demonstration of how this approach can be used for control of a swarm using a single controllable agent and global mode switching.