Closing the loop in legged neuromechanics: An open-source computer vision controlled treadmill

Abstract

The neural circuitry that controls legged locomotion is exquisitely state dependent. The response to a perturbation at one phase of the locomotor cycle is very different from another. As such, tools that enable closed loop experiments in which freely moving animals can be precisely perturbed, either neurally or mechanically, will improve our ability to probe the locomotor control architecture. Because locomotion emerges from the coupled interaction of the nervous system, musculoskeletal system, and physics of the moving body, ideally an experimenter could simultaneously perturb the nervous system, mechanics of the moving animal, or both, at precise moments during well defined locomotor behaviour. Combining the new technique of optogenetics with closed loop tracking of state and fast mechanical perturbations would make this possible. Here we present a computer vision feedback controlled treadmill designed to enable such perturbations. We show that insects (cockroaches Blaberus discoidalis) and mice (Mus mus; C57BL/6) run freely, with preferred speeds in agreement with the literature, over many strides, on the system. We demonstrate the utility of the closed loop system by automatically gathering high speed video (HSV) data when the running animal meets specific operational criteria; here, running within a restricted range of speeds near the midline of the treadmill for a minimum of 2s. We quantify the mouse's gait from HSV data with a phase-based approach, confirming a normal trotting gait. These results demonstrate the system's potential for dissecting the neuromechanical basis of legged locomotion through closed-loop experimental design.