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

TL;DR

EmbodiSkill is a training-free framework that enables embodied agents to self-evolve skills through reflection and targeted revision, significantly improving task success in diverse environments.

Contribution

It introduces a novel skill-aware reflection method for self-evolving embodied agents without additional training, addressing limitations of existing trajectory-based skill updates.

Findings

EmbodiSkill improves task success rates on ALFWorld and EmbodiedBench.

It enables a frozen Qwen3.5-27B to reach 93.28% success on ALFWorld.

It outperforms GPT-5.2 used as a direct agent by 31.58%.

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.