2016-090 – Motion Planning to Enable Safe and Fast Flight of Autonomous Helicopters

This invention enables safe flight of autonomous helicopters and multi-rotors by efficiently calculating safe trajectories that respect helicopter constraints, obstacles, and no fly zones. Additionally the system actively steers the sensor mechanism in the direction of travel and focusing on relevant parts that need to be sensed such as landing zones that need to be mapped in detail. Current trajectory planning approaches cannot respect all helicopter relevant constraints, and generate high performance trajectories that enable safe flight since the planning problem becomes to difficult. The algorithms and software system in this invention disclosure can guarantee safety, run efficiently in real-time, and solve the challenging problems through an efffective decomposition into several components to enable real-time calculation. The components include an ensemble of motion planning algorithms, a safety executive, and a state machine for the different flight modes. Additionally, the sensor steering actively reaction to the information gathered form the sensor to scan only relevant scenes in the environment.The current main use for an autonomous full-scale helicopter. However, the system can be extended to be applicable to small-scale and large-scale aircraft. Additional potential applications include sense and avoid (avoiding other aircraft). Currently, the system requires significant computational resources, however, the CPU usage can be tuned to the fidelity and scale of the planning problem.The invention has been tested in field experiments on an autonomous helicopter in significant field trials and experimental data in simulation and on the real helicopter is available. Scott McEvoy smcevoy@andrew.cmu.edu 412-268-6053