Safe Trajectory Generation for Wheel-Leg Hybrid Mechanism Using Discrete Mechanics and Optimal Control

Abstract

The wheel-legged robot inherits the merit of both the wheeled robot and the legged robot, which can not only adapt to the complex terrain but also maintain the driving efficiency on the flat road. This article presents an optimization-based approach that leverage ideas from computational geometric mechanics to generate safe and high-quality wheel-leg hybrid motions among obstacles. The formulation of the proposed motion optimization problem incorporates the Lagrange–d’Alembert principle as the robot’s dynamic constraints and an efficient closed-form formulation of collision-free constraints. By discretizing the variational mechanics principle directly, rather than its corresponding forced Euler–Lagrange equation, the continuous trajectory optimization problem is transformed into a nonlinear programming (NLP) problem. Numerical simulations and several real-world experiments are conducted on a wheel-legged robot to demonstrate the effectiveness of the proposed trajectory generation approach.