Project Website http://groups.csail.mit.edu/mers/robotic-networks.htm
Networks of unmanned air, space, and land vehicles are being created that will perform search and rescue in uncertain environments. Autonomy is key; the potential interactions between vehicles are too complex for programmers to predefine manually and the required response time is too short for operators to handle on the fly. This is similar to the preceding challenge of coordinating a system's internal network of devices, except that the devices are far more complex, highly autonomous, and mobile. Kirk extends model-based programming and execution to the coordination of these agile, cooperative systems.
Kirk demonstrates that model-based programming languages can manage a rich set of interactions automatically - planning, scheduling, state estimation, and control - and that representations for describing the semantics of programs can be used to describe complex, team behaviors. This research is enabling a new paradigm for cooperative air vehicles and Mars exploration. Kirk was demonstrated on the task of search and rescue on a hardware in the loop unmanned air vehicle test bed. In addition, Kirk was selected for evaluation by the Mars Science Laboratory Technology Acceptance Board at JPL. Research topics include incremental and optimal planning of missions, unified planning of vehicle activities and trajectories, and distributed, team-based planning and execution. Our future research will extend further the work on mission-directed trajectory planning and distributed cooperation.