A reflexive vehicle control architecture based on a neural model of the cockroach escape response

This paper presents a biologically inspired architecture for rapid real-time control of autonomous or semi-autonomous vehicles based on a neural model of the escape response of the American cockroach, Periplaneta americana. The architecture fuses exteroceptive and proprioceptive inputs in a manner similar to the insect to produce commands for collision avoidance and, in some cases, orientation for target strike. It functions as a reflexive subsystem that integrates smoothly with higher-level planning and behavioral control systems. The performance of the reflex is demonstrated in simulation and in hardware experiments on both air and ground vehicles, even in the presence of noisy, false or disruptive sensor data.