SPONBERG, S.N.; CHANG, C.; FULL, R.J.; Univ. of California, Berkeley; Univ. of California, Berkeley; Univ. of California, Berkeley: Sensory Independent, Feedforward Control of Locomotion in Cockroaches Running over Rough Surfaces

Kinematic analysis provides no evidence for feedback control when cockroaches run over rough terrain. A tripod gait is maintained and animals do not use a follow-the-leader stepping pattern. Spring-mass models of cockroach running self-stabilize when perturbed. Mechanical responses alone can mediate recovery from perturbations during high-speed running (~20 cm/s). Recovery from perturbations to a cockroach�s center of mass can begin within 10 msec, challenging the fastest neuro-muscular feedback loops. Yet, a feedforward model of control relying on a tuned mechanical system must be tested by direct measurement of motor output. We ran Death�s head cockroaches (Blaberus discoidalis) over a randomly rough terrain that imposed repeated perturbations. Open-loop, feedforward control would be evident if motor unit activation patterns remained unchanged in face of the perturbations. Alterations in activation patterns would support reliance on neural feedback. We simultaneously recorded body kinematics (500 fps, 4 views) and EMG activity (bipolar, silver wire, 10 kHz) resulting from Df motor neuron activity – the single, primary determinant of femoral extension during high speed running. We observed no significant difference in the EMG spikes/burst or in interspike interval between running on flat and rough terrains. Highly significant differences were evident when cockroaches were required to climb onto or off of the terrain. Our results support the hypothesis that a feedforward neural signal is the primary controller for high-speed running. Stride to stride sensory feedback must be layered on top of this controller to allow tasks such as navigation and maneuvering. Supported by ONR/DARPA. N00014-98-1-0747 and the Fannie and John Hertz Foundation.