Pith. sign in

REVIEW 5 cited by

Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.11470 v1 pith:D34DVYO2 submitted 2025-06-13 cs.RO

Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion

classification cs.RO
keywords diffusionlocomotionmodellearningleggedpolicyresidualacross
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Generalizing locomotion policies across diverse legged robots with varying morphologies is a key challenge due to differences in observation/action dimensions and system dynamics. In this work, we propose Multi-Loco, a novel unified framework combining a morphology-agnostic generative diffusion model with a lightweight residual policy optimized via reinforcement learning (RL). The diffusion model captures morphology-invariant locomotion patterns from diverse cross-embodiment datasets, improving generalization and robustness. The residual policy is shared across all embodiments and refines the actions generated by the diffusion model, enhancing task-aware performance and robustness for real-world deployment. We evaluated our method with a rich library of four legged robots in both simulation and real-world experiments. Compared to a standard RL framework with PPO, our approach -- replacing the Gaussian policy with a diffusion model and residual term -- achieves a 10.35% average return improvement, with gains up to 13.57% in wheeled-biped locomotion tasks. These results highlight the benefits of cross-embodiment data and composite generative architectures in learning robust, generalized locomotion skills.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Rapid co-design of Buoyancy-assisted robots for Challenging Locomotion using Gaussian Evolutionary Specialists

    cs.RO 2026-06 unverdicted novelty 6.0

    GES framework uses Gaussian-partitioned specialist policies to co-optimize morphology and control for buoyancy-assisted legged robots, reporting 5-25% performance gains, 3x hardware obstacle improvement, and 37% faste...

  2. Any2Any: Efficient Cross-Embodiment Transfer for Humanoid Whole-Body Tracking

    cs.RO 2026-05 unverdicted novelty 6.0

    Any2Any transfers pretrained humanoid whole-body tracking policies to new embodiments with 1% of original training cost via kinematic alignment and parameter-efficient fine-tuning.

  3. DynaWM: Dynamics-Aware Distillation with World Model and Momentum Targets for Smooth Locomotion over Continuous Stairs

    cs.RO 2026-06 unverdicted novelty 5.0

    DynaWM adds a world model as dynamics regularizer and momentum targets to teacher-student distillation, yielding better terrain encoding and smoother stair traversal for bipedal-wheeled robots.

  4. Any2Any: Efficient Cross-Embodiment Transfer for Humanoid Whole-Body Tracking

    cs.RO 2026-05 unverdicted novelty 5.0

    Any2Any transfers humanoid whole-body tracking models across embodiments via kinematic alignment followed by targeted PEFT, matching full-training performance with 1% of the data and compute on tested platforms.

  5. Towards a Multi-Embodied Grasping Agent

    cs.RO 2025-10 unverdicted novelty 5.0

    A JAX-implemented flow-based equivariant model for multi-embodiment grasping that deduces kinematics from geometry to support variable-DoF grippers with a new dataset of 25k scenes and 20M grasps.