Humanoid locomotion requires deciding where to step and how to distribute contact forces. However, a disturbance can invalidate the nominal plan, requiring coordinated revision under constraints that switch with every footfall. To solve this online, the proposed approach employs the single-rigid-body model, which retains attitude dynamics and the full contact wrench. An iterative linear-quadratic-regulator MPC (iLQR-MPC) then jointly optimizes contact forces and footstep locations in a single receding-horizon problem, without a separate step planner. Specifically, contact and motion constraints are incorporated into the backward pass through augmented-Lagrangian and relaxed-barrier penalties. A whole-body controller subsequently maps the plan to joint torques at 500 Hz while preserving contact feasibility. The model-based formulation makes disturbance responses interpretable through explicit costs and constraints, and requires no platform-specific policy training. In MuJoCo simulations, the iLQR-MPC replans at 100 Hz (mean solve time 8.9 ms); the Unitree G1 sustains steady-state walking and recovers from impulsive 245 N forward and 175 N lateral pushes.
Figures & tables
Fig. 1: Overall control architecture. A reference-planning module supplies the CoM reference, nominal footholds, and contact schedule. The reduced-model iLQR-MPC computes the contact wrench and footstep placements online from centroidal state feedback, with contact and motion constraints entering the cost J as augmented-Lagrangian and relaxed-barrier penalties. The whole-body controller maps this plan to joint torques via task-priority projection, a dynamics QP, and inverse dynamics, using joint state feedback, with the QP enforcing the floating-base block of the full whole-body dynamics and contact feasibility.
Fig. 2: MPC horizon prediction of the CoM state. (a)–(c) CoM position; (d)–(f) planar velocity and yaw rate. The reference xref (light red, dashed) and prediction xpred (blue) are shown; RMSE is annotated per panel.
State
RMSE
State
RMSE
Roll (rad)
0.038
ωx (rad/s)
0.214
Pitch (rad)
0.016
ωy (rad/s)
0.114
Yaw (rad)
0.029
ωz (rad/s)
0.175
X (m)
0.231
vx (m/s)
0.078
Y (m)
0.064
vy (m/s)
0.267
Z (m)
0.004
vz (m/s)
0.064
TABLE I: Per-State Tracking RMSE During 40 s Forward Walking at 0.1m/s
Fig. 3: Forward ( +x ) push recovery at 245N at 15.0s during 0.1m/s walking. Left: walking sequence; (c) iLQR solve-time distribution. Right: (a) base orientation; (b) swing-leg vertical contact force; (d) angular velocity.
Direction
Type
Max Force (N)
+x (forward)
sagittal
245
−x (backward)
sagittal
205
+y (right-to-left)
lateral
80
−y (left-to-right)
lateral
175
TABLE II: Maximum Recoverable Push Force During Forward Walking at 0.1m/s
Fig. 4: Left-to-right ( −y ) push recovery at 175N at 15.0s during 0.1m/s walking. Left: walking sequence; (d) iLQR solve-time distribution. Right: (a) CoM lateral position; (b) base orientation; (c) swing-leg vertical contact force; (e) angular velocity.
Module
Time (ms)
Share (%)
iLQR inner loop (per iteration, N=10 )
Dynamics linearization
0.083
3.7
Cost quadratization
1.204
53.8
Backward pass & gain computation
0.274
12.2
Forward rollout
0.680
30.3
iLQR iteration total
2.24
100
TABLE III: Computation Time per Iteration and per iLQR-MPC Solve
This paper presents a multi-phase whole-body model predictive control (MPC) approach for bipedal walking, combining a detailed whole-body model in the near horizon with a simplified single-rigid-body model in the later prediction steps. This reduces computational complexity while retaining prediction capabilities. The resulting nonlinear optimal control problem is solved entirely within the general-purpose, off-the-shelf nonlinear MPC framework acados, using sequential quadratic programming (SQP). Given a contact schedule and a target walking speed, the controller optimizes joint torques without depending on preselected footstep locations. The controller is validated in MuJoCo simulation on the 18-DoF bipedal robot HyPer-2.
Franek Stark, Felix Wiebe, Shubham Vyas +2
Robotics Innovation Center at the German Research Center for Artificial Intelligence (DFKI), Bremen, Germany · University Bremen, Germany
This paper presents dynamics-relaxed model predictive control (DR-MPC), a novel MPC formulation for legged locomotion, and a tailored interior-point method (IPM) solver. The formulation combines online optimization feasibility by construction with a contact-aware input parameterization. DR-MPC moves the dynamics equality and affine input constraints into quadratic penalties and retains only nonempty box constraints. The resulting box-constrained quadratic program (QP) has a block-arrow Hessian that enables the state and affine-output directions to be eliminated through a Schur complement. The solver factors only the reduced control system after swing-force elimination and contact-aligned move blocking. For the evaluated implementations using the same DR-MPC formulation, our method achieves median end-to-end MPC speedups of 16.0× over HPIPM and 4.4× over OSQP, with comparable locomotion performance in simulation. DR-MPC achieves a median onboard MPC end-to-end time of 4.4 ms and is validated on a Unitree Go1 quadruped. Open-source code will be made available after publication.
Run Wang, Alapati Tuerxun, Shuo Liu +4
Department of Automation, Tsinghua University, Beijing, China · Department of Mechanical Engineering, Boston University, Brookline, MA, USA · School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore +1
In this paper, we propose a cost-matching approach for optimal humanoid locomotion within a Model Predictive Control (MPC)-based Reinforcement Learning (RL) framework. A parameterized MPC formulation with centroidal dynamics is trained to approximate the action-value function obtained from high-fidelity closed-loop data. Specifically, the MPC cost-to-go is evaluated along recorded state-action trajectories, and the parameters are updated to minimize the discrepancy between MPC-predicted values and measured returns. This formulation enables efficient gradient-based learning while avoiding the computational burden of repeatedly solving the MPC problem during training. The proposed method is validated in simulation using a commercial humanoid platform. Results demonstrate improved locomotion performance and robustness to model mismatch and external disturbances compared with manually tuned baselines.
Wenqi Cai, Kyriakos G. Vamvoudakis, Sébastien Gros +1
Electrical Engineering Program, New York University Abu Dhabi (NYUAD), 129188, UAE. · School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United States. · Department of Engineering Cybernetics, Norwegian University of Science and Technology (NTNU), Trondheim, Norway.