Distributed control and coordination of autonomous agents in a dynamic, reconfigurable system

Systems and techniques are described for distributed control and coordination of autonomous agents in a system. A protocol for distributed control allows autonomous agents to negotiate a group task for the agents, such as locomotion or self-reconfiguration in a robot application, and to synchronize individual agent actions to effect the group task. A protocol for adaptive communication allows autonomous agents, such as physically coupled self-sufficient robot modules, to continuously discover changes in their local topology. In general, in one implementation, the technique includes: discovering a local topology, which includes a type of a communication connection to autonomous agents in the reconfigurable network topology, and determining an action based upon a received control message and the determined local topology.