Sensorimotor control in insects and robots

Traditional control methods, when applied to sensorimotor control of robots, demand high sensor and actuator accuracy and depend on complex information processing. Biological systems, by contrast, typically depend far more on the physical matching of sensors and actuators to specific tasks and environments so that the appropriate behaviour emerges with relatively little information processing, and is relatively robust against error. This paper describes four mechanisms of sensorimotor control in the cricket that have been successfully utilised for cheap and simple robot control.