Some Results In Passive-Dynamic Walking

Human walking can be approximated as a mechanical process governed by Newton’s laws of motion, and not controlled. Tad McGeer first demonstrated, and we have confirmed, that a two-dimensional legged mechanism with four moving parts can exhibit stable, human-like walking on a range of shallow slopes with no actuation and no control (energy lost in friction and collisions is recovered from gravity). More recently, we have found a simple walking mechanism that also balances from side to side. There is much that might be understood about walking by considering it as a natural motion of a simple uncontrolled and unpowered dynamical system, or a passive-dynamic system. As seen from a control perspective, our work largely involves investigation of control parameters which are physical properties rather than the traditional active-control parameters (such as feedback gains, neural net parameters, genetic algorithm reward schemes, etc.). We are testing the hypothesis that human walking is based on an uncontrolled mechanical process by designing, building, and studying uncontrolled or minimallycontrolled walking devices and seeing how well they mimic human motion.