Abstract:This paper proposes a novel control framework for agile and robust bipedal locomotion, addressing model discrepancies between full-body and reduced-order models. Specifically, assumptions such as constant centroidal inertia have introduced significant challenges and limitations in locomotion tasks. To enhance the agility and versatility of full-body humanoid robots, we formalize a Model Predictive Control (MPC) problem that accounts for the variable centroidal inertia of humanoid robots within a convex optimization framework, ensuring computational efficiency for real-time operations. In the proposed formulation, we incorporate a centroidal inertia network designed to predict the variable centroidal inertia over the MPC horizon, taking into account the swing foot trajectories -- an aspect often overlooked in ROM-based MPC frameworks. By integrating the MPC-based contact wrench planning with our low-level whole-body controller, we significantly improve the locomotion performance, achieving stable walking at higher velocities that are not attainable with the baseline method. The effectiveness of our proposed framework is validated through high-fidelity simulations using our full-body bipedal humanoid robot DRACO 3, demonstrating dynamic behaviors.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to achieve high - speed and agile bipedal walking in whole - body robots. Specifically, the paper proposes a new control framework in view of the differences between the whole - body model and the simplified model, and in particular, addresses the impact of the change in the centroid inertia due to leg swinging on the walking stability. ### Problem Background - **Importance of High - Speed and Agile Walking**: For humanoid robots, the ability to walk at high speed and with agility is required to perform human - level tasks in dynamic environments. - **Limitations of Existing Methods**: - **Reduced - Order Models (ROMs)**: Such as the Linear Inverted Pendulum (LIP) model and the Single Rigid Body (SRB) model. Although they simplify the control problem, they overlook crucial factors in robot dynamics, especially the change in angular momentum. - **Constant Centroid Inertia Assumption**: Existing control methods usually assume that the centroid inertia is constant, which can lead to inaccurate models in cases such as leg swinging. ### Method Proposed in the Paper - **Variable Inertia Model Predictive Control (VI - MPC)**: - **Core Idea**: Improve control accuracy by introducing variable centroid inertia, especially during leg swinging. - **Specific Implementation**: - **Centroidal Inertia Network**: Use a neural network to predict the change in centroid inertia over time. - **Convex Optimization Framework**: Transform the variable - inertia problem into a convex optimization problem to ensure real - time computational efficiency. - **Low - Level Whole - Body Controller (WBC)**: Combine with MPC to achieve multi - task tracking, including centroid position, whole - body posture, and contact force. ### Main Contributions 1. **More Accurate Angular Dynamics**: By introducing variable centroid inertia and whole - body orientation (WBO), the accuracy of capturing angular dynamics is improved. 2. **Stable Computation**: Combine offline and online processes to ensure the stability and efficiency of computation and reduce the risk of getting trapped in local minima during the optimization process. 3. **Dynamic Walking Performance**: Demonstrate higher walking speed, longer stride length, and more agile turning behavior, effectively bridging the gap between the whole - body model and the simplified model. ### Experimental Verification - **Simulation Results**: The effectiveness of the proposed method is verified through high - fidelity simulation, especially in terms of high - speed walking and dynamic behavior. - **Performance Indicators**: Including MPC solution time, centroid velocity tracking, etc., demonstrating the real - time and accuracy of the method. In conclusion, by introducing variable centroid inertia, this paper addresses the shortcomings of existing methods in high - speed and agile walking and provides a new solution for the control of whole - body robots.

Variable Inertia Model Predictive Control for Fast Bipedal Maneuvers

Reduced Model Predictive Control Toward Highly Dynamic Quadruped Locomotion

Robust Locomotion Exploiting Multiple Balance Strategies: An Observer-Based Cascaded Model Predictive Control Approach

Highly Dynamic Quadruped Locomotion via Whole-Body Impulse Control and Model Predictive Control

Robust Optimal Planning and Control of Non-Periodic Bipedal Locomotion with A Centroidal Momentum Model

Imitating and Finetuning Model Predictive Control for Robust and Symmetric Quadrupedal Locomotion

Variable Horizon MPC With Swing Foot Dynamics for Bipedal Walking Control

Fast Whole-Body Motion Control of Humanoid Robots with Inertia Constraints

Terrain-Adaptive, ALIP-Based Bipedal Locomotion Controller via Model Predictive Control and Virtual Constraints

Real-Time Control of Full Actuated Biped Robot Based on Nonlinear Model Predictive Control

Dynamic Loco-manipulation on HECTOR: Humanoid for Enhanced ConTrol and Open-source Research

Tailoring Solution Accuracy for Fast Whole-body Model Predictive Control of Legged Robots

Continuous Dynamic Bipedal Jumping via Real-time Variable-model Optimization

Three-dimensional Variable Center of Mass Height Biped Walking Using a New Model and Nonlinear Model Predictive Control

Dynamic Locomotion in the MIT Cheetah 3 Through Convex Model-Predictive Control

Discrete time model predictive control for humanoid walking with step adjustment

Contact-implicit Model Predictive Control: Controlling diverse quadruped motions without pre-planned contact modes or trajectories

CDM-MPC: An Integrated Dynamic Planning and Control Framework for Bipedal Robots Jumping

Contingency Model Predictive Control for Bipedal Locomotion on Moving Surfaces with a Linear Inverted Pendulum Model

A Robust Closed-Loop Biped Locomotion Planner Based on Time Varying Model Predictive Control

RL-augmented MPC Framework for Agile and Robust Bipedal Footstep Locomotion Planning and Control