Biomechanical Comparisons Reveal Divergence of Human and Humanoid Gaits

Luying Feng; Yaochu Jin; Hanze Hu; Wei Chen

cs.ROFeb 25, 2026

Biomechanical Comparisons Reveal Divergence of Human and Humanoid Gaits

Authors: Luying Feng, Yaochu Jin, Hanze Hu, Wei Chen

Organizations: Zhejiang University, Hangzhou, China. · School of Engineering, Westlake University, Hangzhou, China. · Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo, China.

Abstract

It remains challenging to achieve human-like locomotion in legged robots due to fundamental discrepancies between biological and mechanical structures. Although imitation learning has emerged as a promising approach for generating natural robotic movements, simply replicating joint angle trajectories fails to capture the underlying principles of human motion. This study proposes a Gait Divergence Analysis Framework (GDAF), a unified biomechanical evaluation framework that systematically quantifies kinematic and kinetic discrepancies between humans and bipedal robots. We apply GDAF to systematically compare human and humanoid locomotion across 28 walking speeds. To enable reproducible analysis, we collect and release a speed-continuous humanoid locomotion dataset from a state-of-the-art humanoid controller. We further provide an open-source implementation of GDAF, including analysis, visualization, and MuJoCo-based tools, enabling quantitative, interpretable, and reproducible biomechanical analysis of humanoid locomotion. Results demonstrate that despite visually human-like motion generated by modern humanoid controllers, significant biomechanical divergence persists across speeds. Robots exhibit systematic deviations in gait symmetry, energy distribution, and joint coordination, indicating that substantial room remains for improving the biomechanical fidelity and energetic efficiency of humanoid locomotion. This work provides a quantitative benchmark for evaluating humanoid locomotion and offers data and versatile tools to support the development of more human-like and energetically efficient locomotion controllers. The data and code will be made publicly available upon acceptance of the paper.

Figures & tables

Fig. 1 : Joint mapping between humanoid robot and human.

Fig. 2 : Comparison of lower-limb joint angles, moments, and power between human subjects and humanoid robot at three walking speeds.

Fig. 3 : Comparison of torque-angle loops between human subjects and humanoid robot at three walking speeds.

Panel of Figure 4 — Fig. 4 : Multidimensional similarity heatmaps of joint angles, moments, and power across walking speeds

Fig. 5 : Bilateral symmetry index (SI) across joint pairs for human and robot.

Fig. 6 : Positive/negative work from power profiles.

Category	Metric	0.5	0.75	1.0	1.25	1.5
Symmetry	$\mathrm{SI}^{\mathrm{R}}_{v}$ ( $\mathrm{SI}^{\mathrm{H}}_{v}$ )	0.8292 (0.8067)	0.8488 (0.7259)	0.8254 (0.6252)	0.7511 (0.5545)	0.6772 (0.4936)
	$A^{\mathrm{W,R}}_{v}$ ( $A^{\mathrm{W,H}}_{v}$ )	3.3708 (0.3127)	0.9428 (0.2172)	0.9911 (0.1416)	0.6542 (0.1870)	0.2507 (0.1355)
	$\mathcal{S}^{\mathrm{R}}_{v}$	0.5831	0.4715	0.4623	0.4083	0.3511
Human-likeness	$R^{\mathrm{wav}}_{v}$	0.4316	0.4182	0.4458	0.5668	0.5856
	$d^{\mathrm{work}}_{v}$	3.0759	1.1461	0.9081	0.9929	0.5966
	$\mathcal{H}_{v}$	0.4380	0.3482	0.3225	0.2662	0.2371

TABLE I : Comprehensive GDAF indices per speed (m/s). Symmetry rows: robot value (human reference in parentheses).

Panel of Figure 8 — Fig. A.1 : Joint angles, moments, and power between human subject and humanoid robot under 28 speeds.

Fig. A.2 : MuJoCo-based gait visualization tool. A 29-DOF Unitree G1 model is driven by recorded human or robotic joint trajectories. The right panel shows joint kinematics and kinetics over the gait cycle, synchronized with the simulated motion.

Humanoid Robot Locomotion

Humanoid robot locomotion studies how bipedal robots walk, run, recover from disturbances, and traverse varied terrain.

See all 145 papers

HuMBLE: Human Motion-Driven Behavior Learning for Embodied Locomotion
Oct 7, 2026Mike Zhang, Dongho Kang, Kevin Bergamin +19Human-to-Robot Motion Imitation Humanoid Robot Locomotion
RAI Institute, USA · Boston Dynamics, USA · Carnegie Mellon University, USA
Imitation Learning from Human Motion Alone Does Not Guarantee Biomechanically Plausible Gait Kinetics
Mar 12, 2026Xinyi Liu, Jangwhan Ahn, Edgar Lobaton +2Gait Analysis Imitation Learning
Joint Department of Biomedical Engineering, University of North Carolina - Chapel Hill, Chapel Hill, NC, USA · Department of Electrical and Computer Engineering, North Carolina State University, Raleigh, NC, USA · School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ, USA
PHUMA: Physically Reliable Humanoid Locomotion Dataset
Oct 30, 2025Kyungmin Lee, Sibeen Kim, Youngdo Lee +6Humanoid Robot Locomotion Robot Imitation Learning
KAIST

Oct 7, 2026J/K move · Enter open · S save

HuMBLE: Human Motion-Driven Behavior Learning for Embodied Locomotion

Mike Zhang, Dongho Kang, Kevin Bergamin, Nicola Burger+18

RAI Institute, USA · Boston Dynamics, USA · Carnegie Mellon University, USA

Human-to-Robot Motion Imitation Humanoid Robot Locomotion RL for Legged Locomotion Robot Imitation Learning

Open paper AskPDF arXiv

Despite recent advances in humanoid locomotion, controllers optimized for command tracking and robustness tend to produce mechanical gaits, whereas controllers tied to human motion data often fail to generalize to commands outside the data distribution. This work introduces a learning framework that balances these competing objectives to synthesize real-time steerable, robust, and biomimetic locomotion policies from human data. Using an in-house curated locomotion dataset covering diverse speeds and directions, we first learn a natural locomotion prior policy through a teacher-student distillation process. Specifically, we train a full-body reference-conditioned policy with Reinforcement Learning (RL), then distill it into a lightweight prior policy conditioned solely on proprioception and a planar torso-velocity steering command. Next, we fine-tune the prior policy with multi-task RL to expand command coverage and robustness beyond the data distribution, pairing a goal-conditioned task that tracks arbitrary commands with a reference-guided task that tracks the human data as an explicit style regularizer. We validate our framework on three humanoid robots: the Boston Dynamics Atlas R1, Atlas D1, and Unitree G1. Experimental results demonstrate robust performance across real-world scenarios, including direct user-controlled locomotion in indoor and outdoor environments, and integration as the locomotion layer within hierarchical control stacks. Benchmarks against Tabula Rasa RL policies trained without human data and ablation studies confirm that our framework yields a lightweight, deployable policy that reconstructs coordinated whole-body behavior from a steering command, retaining the human gait characteristics while remaining robust and fully steerable.

Abstract

Humanoid Robot Locomotion

Explore similar work

HuMBLE: Human Motion-Driven Behavior Learning for Embodied Locomotion

Imitation Learning from Human Motion Alone Does Not Guarantee Biomechanically Plausible Gait Kinetics

PHUMA: Physically Reliable Humanoid Locomotion Dataset