Research Analyzer
← Back ICRA 2026

Robust Bipedal Walking with Closed-Loop MPC: Adios Stabilizers

Antonin Dallard, Mehdi Benallegue, Nicola Scianca, Fumio Kanehiro, Abderrahmane Kheddar

PDF

AI summary

Key figure (auto-extracted from paper)
Eliminates the need for external stabilizer correction policies in bipedal walking by embedding contact force dynamics directly into a closed-loop model predictive controller.
Bipedal locomotion Model predictive control Humanoid robots Zero moment point Stabilizer-free control Force control

Problem

Standard LIPM-based walking controllers rely on short-sighted, hard-to-tune external stabilizers to compensate for modeling errors and disturbances, which disconnects the high-level planner from the robot's actual state and can cause instability.

Approach

The authors reformulate the inherently stable MPC to explicitly model ZMP and contact force dynamics, creating a closed-loop system that directly controls the center of mass, velocity, and zero moment point while dynamically replanning footsteps.

Key results

  • Eliminates external stabilizer correction policies for balance maintenance
  • Enables closed-loop control of CoM, velocity, and ZMP within the MPC
  • Demonstrates robust walking under sudden pushes and on compliant terrain
  • Successfully deployed across five distinct humanoid robot platforms

Why it matters

Simplifies and stabilizes humanoid locomotion control, making robust bipedal walking easier to implement and more portable across diverse hardware without tedious parameter tuning.

Abstract

In this article, we propose a novel walking control scheme based on the dynamics of the linear inverted pendulum (LIP) model. Pattern generation incorporates a model of contact forces, enabling closed-loop control of the humanoid robot’s state, including the center-of-mass position, velocity, and zero moment point. No additional control policies are required to maintain static and dynamic balance. Our approach also includes dynamic replanning of step locations and timings, thus preserving the LIP’s boundedness condition. We validated this controller on five differ- ent humanoid robots, testing its robustness through various distur- bances, including sudden pushes during walking and static phases. In addition, our controller demonstrated effective locomotion over uneven and compliant terrain. Both simulation and experimental results confirm the effectiveness and robustness of this controller.

Index terms

Humanoid and Bipedal Locomotion Legged Robots Force Control Humanoid Robots

Related papers