Decentralized discrete-event modeling and control of task execution for robotic networks

This paper describes a decentralized approach to control the execution of a set of tasks for a network of heterogeneous robotic agents, i.e., agents with different sets of sensors and actuators. The proposed approach is based on a discrete-event modeling and control framework. The novelty is the full decoupling between the tasks modeling and their deployment. This implies an higher robustness to faults in the network of agents and a more systematic way for the task modeling regardless of the actual capability of the agents. A consensus-based mechanism is exploited in the control system to achieve coherence among the agents state and synchronize the agents action.