Adaptive walk control for hexapod robot based on iterative model-checking and Timekeeper control

Abstract

This paper proposes an adaptive walk control for a hexapod robot which is controlled on the basis of a gait control, Follow-the-Contact-Point (FCP) gait control. The FCP gait control is a decentralized and event-driven controller and decides the order and timing of the legs to contact on the ground. Thanks to that, The robot is capable to walk over various uneven terrain. However, adaptive and real-time adjustment of the timing is required to keep walking performance, such as deadlock-free and not to drag the feet on the ground. Therefore, in this paper, such the adjustment method for the FCP gait control is addressed. First, the behavior of each leg of the robot is modeled by a timed-automaton, and specifications that the robot should satisfy to keep walking performance are expressed by computational tree logic. Next, upper/lower boundaries of time which is elapsed in each control state of the FCP gait control are derived based on an algorithm which is utilizing model checking. Then, the adaptive method “Timekeeper control” to guarantee the time boundaries in real-time is proposed. Finally, the validity of the proposed method is verified through experiments in which a real hexapod robot walks on uneven terrain.