A hybrid, self-organizing controller for multi-agent motion planning in a stationary environment

This work explores the construction of a decentralized traffic controller for more than one agents sharing a workspace with stationary forbidden regions. The suggested controller is required to be complete and have a computational effort that linearly increase with the number of agents. The controller is also required to be self-organizing, therefore is able to generate the needed information that is required to execute a successful action. In addition to the above, the controller is required to be open to enable any agent to join or leave the group without the remaining agents having to adjust the manner in which they function. To meet these requirements a new definition of decentralization is suggested. This definition equates decentralization to self-organization. Hybrid systems, as well as the potential field approach for control action synthesis are used to convert the above definition into a workable framework for the generation of decentralized control. Theoretical development, as well as simulation results are provided.