Intelligent Control of High-Speed Turning in a Quadruped

Abstract

Understanding and implementing the control mechanisms that animals use to robustly negotiate a variety of terrains at high speed remains an unsolved problem. Previous research has resulted in control of quadruped running over a range of low speeds or narrowly around a single high speed. Control over a range of both low and high speeds is difficult because a quadruped system is significantly more responsive at high speeds than at low speeds, and because the proportional-derivative style controllers used by many of the previous researchers are only effective locally around the single speed and turning rate at which the controller was tuned. This work presents a fuzzy control strategy that manages the complex coupling between the multiple system inputs and outputs to successfully execute high-speed turns over a range of speeds and turning rates. The resulting control system stabilizes a 3D quadruped trot up to 4 m/s and turning up to 30 deg/s, on a quadruped system with articulated legs and practical leg mass properties in a simulation environment with realistic friction coefficients and system losses.