# WOLF-VLA: Whole-Body Humanoid Optimal Locomotion Framework for Vision-Language-Action Learning

Source: [WOLF-VLA: Whole-Body Humanoid Optimal Locomotion Framework for Vision-Language-Action Learning](<https://www.sota2.com/research/paper/6a3cb617206542251d7ca578>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-07-01T05:00:12.177Z

Authors: Melya Boukheddimi, Omar Adjali, Daniel Sontag, Frank Kirchner

Published: 2026-06-24T08:59:59.000Z

## Abstract

Vision-Language-Action (VLA) models have recently demonstrated strong generalization in robotic manipulation, yet their applicability to whole-body, contact-rich humanoid locomotion remains severely underexplored due to data scarcity, the absence of dynamically consistent demonstrations, and the difficulty of encoding optimality and safety in learning-based pipelines. This work introduces a unified framework WOLF-VLA that integrates whole-body optimal-control (OC) motion synthesis with large-scale multi-modal dataset to train VLAs capable of generating humanoid locomotion policies directly from natural-language instructions. We construct a comprehensive dataset of dynamically feasible humanoid trajectories across six locomotion-related task families, each parameterized by environmental variations, object colors, placements, and visual distractors. We train a VLA model using the collected joint trajectories, ego-centric visual observations and natural language instruction, yielding a policy that exhibits strong reasoning and robustness to initial-condition variability, and competitive performance across several tasks and environment settings. A systematic ablation study demonstrates the impact of each modality on the model performance. The full dataset, model checkpoints, and benchmarking simulation suite will be openly released, establishing a reproducible dynamically consistent benchmark for whole-body humanoid locomotion rich VLA control and enabling future research in scalable transfer of instruction-driven locomotion policies.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Walk and Climb Stairs Up (W.CS.U) on WOLF-VLA](<https://www.sota2.com/research/sota/walk-and-climb-stairs-up-w-cs-u-on-wolf-vla.md>): Walk and Climb Stairs Up (W.CS.U); WOLF-VLA; Soft Success Rate: 63.3.

- [Walk and Climb Stairs Up and Down (W.CS.U/D) on WOLF-VLA](<https://www.sota2.com/research/sota/walk-and-climb-stairs-up-and-down-w-cs-u-d-on-wolf-vla.md>): Walk and Climb Stairs Up and Down (W.CS.U/D); WOLF-VLA; Soft Success Rate: 68.3.

- [Walk Around (WA) on WOLF-VLA](<https://www.sota2.com/research/sota/walk-around-wa-on-wolf-vla.md>): Walk Around (WA); WOLF-VLA; Success Rate: 100.

- [Walk Forward (WF) on WOLF-VLA](<https://www.sota2.com/research/sota/walk-forward-wf-on-wolf-vla.md>): Walk Forward (WF); WOLF-VLA; Success Rate: 100.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2606.25591>)

[PDF](<https://arxiv.org/pdf/2606.25591.pdf>)