用语言和视频快速适配机器人策略，无需额外训练。
Robotic Policy Adaptation via Weight-Space Meta-Learning
- 通过权重空间元学习生成任务专用的LoRA参数，不需微调。
- 在LIBERO上未见数据集性能提升约2倍，未见任务提升14倍。
- 适合需要快速部署新任务的机器人应用，尤其真实场景中表现优。
视觉-语言-动作（VLA）模型正成为机器人操作的有前景范式，能从大规模示范数据和动作标签中训练通用策略。然而，将其适配到新任务通常仍需特定任务的示范、动作标注及额外微调，导致部署成本高且难以扩展。我们提出WIZARD，一种权重空间元学习框架，通过为冻结的VLA策略生成任务专用的LoRA参数，跳过特定任务的微调。仅需语言指令和一段示范视频，WIZARD即可在单次前向传播中预测适配权重，无需目标任务动作标签或测试时优化。元训练过程中，WIZARD学习将任务证据直接映射到专家级LoRA更新，捕捉任务间的权重空间关系。在LIBERO上的实验表明，WIZARD在未见数据集集合上性能提升约2倍，在未见任务上提升约14倍。在Franka Emika Panda机器人上，WIZARD持续优于真实域适应基线，证明生成的适配器可提供超越仿真环境的任务级专业化能力。
原文摘要 · Abstract (English)
Vision-Language-Action (VLA) models are emerging as a promising paradigm for robotic manipulation, enabling general-purpose policies trained from large corpora of demonstrations and action labels. However, adapting these models to new tasks still typically requires task-specific demonstrations, action annotations, and additional fine-tuning, making deployment costly and difficult to scale. We propose WIZARD, a weight-space meta-learning framework that sidesteps task-specific fine-tuning by generating task-specific LoRA parameters for a frozen VLA policy. Given only a language instruction and a short demonstration video, WIZARD predicts the corresponding adaptation weights in a single forward pass, without target-task action labels or test-time optimization. During meta-training, WIZARD learns to map task evidence directly to expert LoRA updates, capturing relationships between tasks in weight space. Experiments on LIBERO show that WIZARD improves performance by up to ~2x on unseen dataset collections and up to ~14x on unseen tasks. On a Franka Emika Panda, WIZARD consistently improves over a real-domain adapted baseline, showing that generated adapters provide task-level specialization beyond simulation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。