# EmbodiSkill: Skill-Aware Reflection for Self-Evolving Embodied Agents

Source: [EmbodiSkill: Skill-Aware Reflection for Self-Evolving Embodied Agents](<https://www.sota2.com/research/paper/6a02b477dbc836b87d824bc0>)

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

Data updated: 2026-07-15T05:00:01.325Z

Authors: Ruofei Ju, Xinrui Wang, Xin Ding, Yifan Yang, Hao Wu, Shiqi Jiang, Qianxi Zhang, Hao Wen, Xiangyu Li, Weijun Wang, Kun Li, Yunxin Liu, Haipeng Dai, Wei Wang, Ting Cao

Published: 2026-05-11T10:33:29.000Z

## Abstract

Embodied agents can benefit from skills that guide object search, action execution, and state changes across diverse environments. Since embodied environments vary across layouts, object states, and other execution factors, these skills must self-evolve from trajectories generated during task execution. However, existing skill self-evolution methods are mainly developed in digital environments and often convert trajectories into coarse skill updates. Directly applying this paradigm to embodied settings is problematic, because a failed task execution may reflect not only incorrect skill content, but also an execution lapse in which the agent fails to follow valid guidance. We propose EmbodiSkill, a training-free framework for embodied skill self-evolution through skill-aware reflection and targeted revision. EmbodiSkill interprets each trajectory with respect to the current skill, uses skill-changing evidence to update the skill body, and uses execution-lapse evidence to preserve and emphasize valid guidance. Experiments on ALFWorld and EmbodiedBench show that EmbodiSkill consistently improves embodied task success. On ALFWorld, EmbodiSkill enables a frozen Qwen3.5-27B executor to reach 93.28% task success, outperforming GPT-5.2 used as a direct agent without skills by 31.58%. These results show that skill-aware self-evolution helps embodied agents accumulate reusable procedural knowledge from their own trajectories.

## 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.

- [Embodied Task Execution on ALFWorld](<https://www.sota2.com/research/sota/embodied-task-execution-on-alfworld.md>): Embodied Task Execution; AlfWorld; Success Rate: 93.28.

- [Embodied Task Success on EmbodiedBench EB-Habitat 1.0 (test)](<https://www.sota2.com/research/sota/embodied-task-success-on-embodiedbench-eb-habitat-1-0-test.md>): Embodied Task Success; EmbodiedBench EB-Habitat 1.0 (test); Average Task Success: 52.33.

- [Embodied Task Success on EmbodiedBench EB-Navigation 1.0 (test)](<https://www.sota2.com/research/sota/embodied-task-success-on-embodiedbench-eb-navigation-1-0-test.md>): Embodied Task Success; EmbodiedBench EB-Navigation 1.0 (test); Average Success Rate: 61.33.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.10332>)

[PDF](<https://arxiv.org/pdf/2605.10332.pdf>)
