Descriptive Model-based Learning and Control for Bipedal Locomotion

TL;DR

The paper tackles the challenge of robust bipedal balance in high-DoF robots by criticizing prescriptive trajectory-tracking approaches that force the full robot to follow low-dimensional references. It introduces a descriptive model-based learning and control framework that maintains balance via a minimal projection while keeping the remaining degrees of freedom free to pursue secondary tasks, enabling efficient and robust locomotion. Major contributions include a descriptive inverted pendulum model with stepping and push-off control, a reinforcement learning–based foot-placement strategy, velocity regulation through orbital parameters, a 2D humanoid control architecture combining balance and posture control with a supervisory layer, and simulation validation on Ranger Max demonstrating stable walking across speeds and slopes with disturbance rejection. The framework promises improved robustness and natural gait in real-world scenarios and paves the way for extensions to rough terrains, stairs, and 3D hardware implementations.

Abstract

Bipedal balance is challenging due to its multi-phase, hybrid nature and high-dimensional state space. Traditional balance control approaches for bipedal robots rely on low-dimensional models for locomotion planning and reactive control, constraining the full robot to behave like these simplified models. This involves tracking preset reference paths for the Center of Mass and upper body obtained through low-dimensional models, often resulting in inefficient walking patterns with bent knees. However, we observe that bipedal balance is inherently low-dimensional and can be effectively described with simple state and action descriptors in a low-dimensional state space. This allows the robot's motion to evolve freely in its high-dimensional state space, only constraining its projection in the low-dimensional state space. In this work, we propose a novel control approach that avoids prescribing a low-dimensional model to the full model. Instead, our control framework uses a descriptive model with the minimum degrees of freedom necessary to maintain balance, allowing the remaining degrees of freedom to evolve freely in the high-dimensional space. This results in an efficient human-like walking gait and improved robustness.

Figures (15)

• Figure 1: Ranger-Max Humanoid
• Figure 2: Full order Model vs Reduced order Model
• Figure 3: Prescriptive Model based Control Block Diagram
• Figure 4: LIPM Walking Gait
• Figure 5: IP Model