Biologically-Inspired Control for Self-Adaptive Multiagent Systems

LINKS: Movies | Publications

Biological systems can achieve scalable and adaptive group behaviors, such as animal flocking, through local interactions amongst a vast network 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 networks of agents to achieve collective tasks, and at the same time adapt to dynamic conditions like living systems?

Our group has developed a biologically-inspired control framework for multi-agent networks to achieve coordinated tasks in a scalable, robust, and analyzable manner. We focus on a type of tasks, called distributed homeostasis, 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, and this approach allows us to capture various multi-agent scenarios and tasks. We show how one can exploit the locality of this formulation to design nearest-neighbor agent control, based on simple sensing, actuation and local communication.

The main contributions and directions of this project are:

• Simple Decentralized Algorithms for Various Self-Adaptive Tasks: We formulate this problem more generally as distributed constraint maintenance on a networked multiagent system. Such a formulation allows the framework to model and simplify tasks on a variety of distributed agent applications. It also allows us to exploit locality of the task to derive decentralized agent control that is based on a simple sensing and actuation strategy.
• Theoretical Analysis and Understanding of the Approach: We analyze important aspects of this decentralized approach, including convergence, scalability, and robustness. We also outline sufficient conditions that guarantee agent control laws' correctness for different networked multiagent systems. This analytical framework allows one to have a concrete understanding of the strengths, limitations, and scope of this class of decentralized approaches.
• Many Application Areas: This framework has wide applicability to autonomous systems that exploit sensor-actuator networks as underlying architectures, including modular robots, adaptive architecture,  multi-robot cooperation, and adaptive medical devices. We have designed and demonstrated a number of different hardware prototype systems, using both chain-style and link-style modular robots, based on this theoretical approach.