Robotic Policy Adaptation via Weight-Space Meta-Learning
For robotic manipulation, WIZARD enables zero-shot task adaptation without action labels or test-time optimization, reducing deployment cost.
WIZARD generates task-specific LoRA parameters for frozen VLA policies from a language instruction and demonstration video, eliminating fine-tuning. It achieves up to ~2x improvement on unseen dataset collections and up to ~14x on unseen tasks in LIBERO, and improves real-robot performance.
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.