SkillWeaver: Agentic Exploration over Neural Interaction Skills for Scalable Robot Data Generation

2026-09-28

AI Summary

This study addresses the high cost and limited scalability of robot data collection across diverse environments and long-horizon tasks by proposing a vision-language model (VLM)-based agent framework for autonomous experience generation. The approach integrates reusable parameterized policies with agent-driven exploration, eliminating reliance on predefined execution pipelines. Closed-loop manipulation is achieved through VLM-guided reasoning and verifier-directed tree search, further supporting simulation-to-real transfer. Experimental results demonstrate that the framework autonomously generates 39,100 demonstration trajectories, substantially enhancing model generalization capabilities. It achieves zero-shot and few-shot cross-domain transfer, establishing a novel paradigm for scalable robot learning.

Abstract

Large-scale demonstrations have driven unprecedented progress in robot learning, yet collecting robot data through teleoperation is expensive and difficult to scale to diverse environments and long-horizon tasks. Simulation offers a scalable alternative, but existing data-generation pipelines often rely on open-loop controllers, scripted skill sequences, or task-specific programs. We introduce SkillWeaver, an agentic framework that autonomously generates robot experience by exploring over Neural Interaction Skills (NIS): reusable, parameterized, closed-loop policies that expose learned physical interaction capabilities to a reasoning agent. Given a task and a simulated environment, a VLM agent reasons about what to do next, invokes and parameterizes NIS to interact with the environment, observes their outcomes, and generates verification, reflection, and memory to guide subsequent exploration. We instantiate NIS as reinforcement-learned policies for closed-loop, contact-rich manipulation and organize exploration as verifier-guided tree search, enabling the agent to discover successful long-horizon behaviors without relying on predetermined execution pipelines. SkillWeaver scales autonomously to 39.1K demonstrations across 14.1K scenes, which we distill into visuomotor policies. Across simulation benchmarks and real-world manipulation, training on SkillWeaver-generated experience substantially improves generalization to novel objects, spatial configurations, tasks, and environments, and enables zero- and few-shot sim-to-sim and sim-to-real transfer. Our results suggest agentic exploration over neural interaction skills as a scalable alternative for robot data generation.

Problem

• robot data generation
• scalability
• long-horizon tasks
• teleoperation
• simulation

Innovation

• Neural Interaction Skills
• Agentic Exploration
• Vision-Language Model
• Tree Search
• Sim-to-Real Transfer