Biomimetic Control of Flying Robots

Physical Sciences Inc. (PSI) demonstrated in Phase I that it is possible to design a Micro Air Vehicle (MAV) which is robust to collisions by using techniques similar to what insects use to recover from collisions while still in flight. Observational studies of insects colliding with objects (high speed films of instrumented insects) conducted at the Harvard Microrobotics Laboratory show that the organisms maneuver and recover from collisions reflexively, likely with little to no high level processing of sensory data. PSI demonstrated the feasibility of using Central Pattern Generator neural networks (CPG) in an MAV reflexive flight control system; a system that does not require a central processing unit, microcontroller, or finite state machine. In Phase II PSI will complete the design of the CPG-based reflexive flight controller and will demonstrate its performance on PSIs robust quad-rotor MAV demonstrator. The reflexive flight controller will be capable of performing inner-loop stabilization of the MAV while simultaneously avoiding obstacles and recovering from upsets and collisions.