AtlasVA: Self-Evolving Visual Skill Memory for Teacher-Free VLM Agents
Summary
AtlasVA is a teacher-free visual skill memory framework for vision-language model agents that uses spatial heatmaps, visual exemplars, and symbolic text skills to improve spatial decision-making in long-horizon tasks, outperforming baselines on several benchmarks.
View Cached Full Text
Cached at: 05/19/26, 06:30 AM
Paper page - AtlasVA: Self-Evolving Visual Skill Memory for Teacher-Free VLM Agents
Source: https://huggingface.co/papers/2605.17933
Abstract
AtlasVA is a teacher-free visual skill memory framework for vision-language model agents that uses spatial heatmaps, visual exemplars, and symbolic text skills to improve spatial decision-making in long-horizon tasks.
Vision-language model(VLM) agents increasingly rely onmemory-augmented reinforcement learningto reuse experience across long-horizon tasks, yet most existing frameworks store memory as text and depend on proprietary teacher models to summarize or refine it. This design is poorly matched to spatial decision making: geometric priors are compressed into lossy language, and sparse interaction is often supervised through delayed textual feedback rather than dense visually grounded signals. We argue that reusable experience for VLM agents should remain visually grounded. Based on this insight, we propose AtlasVA, a teacher-freevisual skill memoryframework that organizes memory into three complementary layers:spatial heatmaps,visual exemplars, andsymbolic text skills. AtlasVA further evolves danger andaffinity atlases directly fromtrajectory statisticsand lightweightgrid heuristics, and reuses these self-evolving atlases aspotential-based shaping rewardsfor reinforcement learning. This unifies perception, memory, and optimization without external LLM supervision. Experiments on Sokoban, FrozenLake, 3D embodied navigation, and 3D robotic manipulation benchmarks show that AtlasVA consistently outperforms text-centric memory baselines and competitive VLM agents, with especially strong gains on spatially intensive tasks. Homepage: https://wangpan-ustc.github.io/AtlasvaWeb
View arXiv pageView PDFProject pageAdd to collection
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2605.17933 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2605.17933 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2605.17933 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies
EventVLA introduces a sparse visual evidence memory framework for long-horizon robotic manipulation, achieving an average success rate improvement of +40% over state-of-the-art memory-augmented VLAs.
AtlasVLA: Persistent World-Ego State Modeling for Vision-Language-Action Models
AtlasVLA introduces a dual-memory architecture with persistent world-ego state modeling to overcome perception and task-progress forgetting in vision-language-action models, achieving state-of-the-art long-horizon manipulation from a single wrist camera.
SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models
SimpleMemVLA introduces a simple memory mechanism for Vision-Language-Action models by feeding intact timestamped video history into a pretrained VLM backbone, achieving state-of-the-art results on long-horizon manipulation tasks without dedicated memory modules.
InSight: Self-Guided Skill Acquisition via Steerable VLAs
InSight presents a framework for autonomous skill acquisition in vision-language-action (VLA) models by enabling steerability at the primitive-action level and using a VLM-guided data flywheel to generate demonstrations, achieving manipulation tasks like block flipping and pouring without human demonstrations.
VisualThink-VLA: Visual Intermediate Reasoning for Effective and Low-Latency Vision-Language-Action Policies
VisualThink-VLA introduces a visual intermediate reasoning framework for vision-language-action policies that preserves spatial precision and dramatically reduces latency compared to text-based reasoning, achieving sub-second inference and state-of-the-art success rates on robot manipulation benchmarks.