Multi-agent collision avoidance with acceleration constraints

A switching control law is derived using optimal control techniques to guarantee separation between acceleration constrained vehicles.  The method augments existing multi-agent control algorithms with the capability to switch to provably safe collision avoidance maneuvers when required.  It is extended to the control of arbitrarily many vehicles, limited only by communication of vehicle states.  These methods are demonstrated in simulations where existing techniques are prohibitively slow or fail, and in quadrotor helicopter flight experiments.