Multi-robot LTL Planning Under Uncertainty

Robot applications are increasingly based on teams of robots that collaborate to perform a desired mission. Such applications ask for decentralized techniques that allow for tractable automated planning. Another aspect that current robot applications must consider is partial knowledge about the environment in which the robots are operating and the uncertainty associated with the outcome of the robots’ actions. Current planning techniques used for teams of robots that perform complex missions do not systematically address these challenges: (1) they are either based on centralized solutions and hence not scalable, (2) they consider rather simple missions, such as A-to-B travel, (3) they do not work in partially known environments. We present a planning solution that decomposes the team of robots into subclasses, considers missions given in temporal logic, and at the same time works when only partial knowledge of the environment is available. We prove the correctness of the solution and evaluate its effectiveness on a set of realistic examples.