Agile Maneuvers in Legged Robots: a Predictive Control Approach

TL;DR

This paper introduces a hybrid predictive control method enabling legged robots to perform agile maneuvers in real-time by considering full-body dynamics and actuation limits, improving agility and responsiveness.

Contribution

It presents the first predictive controller that handles actuation limits, generates agile maneuvers, and executes optimal feedback policies without a separate whole-body controller.

Findings

Enables ANYmal robots to perform agile maneuvers in realistic scenarios.

Converges within a few milliseconds due to a feasibility-driven approach.

Successfully tracks local feedback policies to control angular momentum.

Abstract

Planning and execution of agile locomotion maneuvers have been a longstanding challenge in legged robotics. It requires to derive motion plans and local feedback policies in real-time to handle the nonholonomy of the kinetic momenta. To achieve so, we propose a hybrid predictive controller that considers the robot's actuation limits and full-body dynamics. It combines the feedback policies with tactile information to locally predict future actions. It converges within a few milliseconds thanks to a feasibility-driven approach. Our predictive controller enables ANYmal robots to generate agile maneuvers in realistic scenarios. A crucial element is to track the local feedback policies as, in contrast to whole-body control, they achieve the desired angular momentum. To the best of our knowledge, our predictive controller is the first to handle actuation limits, generate agile locomotion maneuvers, and execute optimal feedback policies for low level torque control without the use of a separate whole-body controller.