Biologically-inspired control for self-adaptive multiagent systems

Biological systems achieve robust and scalable group behaviors, such as flocking, through local interactions amongst vast numbers of unreliable agents. In these systems, each agent acts autonomously and interacts only with its neighbors, yet the global system exhibits coordinated behavior. Large-scale multi-agent systems, e.g. distributed robot systems, are similar to these biological systems, in that their overall tasks must be achieved by coordinating many independent agents. One important question to ask is: How can we program large numbers of agents to achieve collective tasks, and at the same time adapt to dynamic conditions like living systems? In this thesis, I develop a biologically-inspired control framework for multi-agent systems to achieve coordinated tasks in a scalable, robust, and analyzable manner. I focus on the type of tasks in which agents must use distributed sensing to solve collective tasks and to cope with changing environments. This task space can be formulated more generally as distributed constraint-maintenance on a networked multi-agent system such that it can capture various multi-agent tasks. I show how one can exploit the locality of this formulation to design nearest-neighbor agent control, based on simple sensing, actuation and local communication. I further theoretically determine several important properties of this decentralized control, such as convergence, scalability, and reactivity. Using these results, we can provide precise statements on how the approach scales and how quickly agents can adapt to perturbations. Practically, I demonstrate this approach with various modular robots, a type of robotic system composed of many connected and autonomous modules. With this approach, a diverse set of modular robot challenges can be similarly addressed, including environmentally-adaptive shape formation, grasping, and collective locomotion. Altogether, the thesis provides examples of how one can systematically design decentralized control for multi-agent tasks and autonomous robot control, and provides a deeper understanding of the contrast between centralized and decentralized algorithms.