y0news
← Feed
Back to feed
🧠 AI🟢 BullishImportance 6/10

REFINE-DP: Diffusion Policy Fine-tuning for Humanoid Loco-manipulation via Reinforcement Learning

arXiv – CS AI|Zhaoyuan Gu, Yipu Chen, Zimeng Chai, Alfred Cueva, Thong Nguyen, Yifan Wu, Huishu Xue, Minji Kim, Isaac Legene, Fukang Liu, Matthew Kim, Ayan Barula, Yongxin Chen, Ye Zhao|
🤖AI Summary

Researchers developed REFINE-DP, a hierarchical framework that combines diffusion policies with reinforcement learning to enable humanoid robots to perform complex loco-manipulation tasks. The system achieves over 90% success rate in simulation and demonstrates smooth autonomous execution in real-world environments for tasks like door traversal and object transport.

Key Takeaways
  • REFINE-DP addresses the challenge of deploying diffusion policies on humanoid robots by jointly optimizing high-level planning and low-level control.
  • The framework uses PPO-based diffusion policy gradient to fine-tune motion planners while simultaneously updating controllers to reduce distributional mismatch.
  • The system achieves over 90% success rate in simulation, even for out-of-distribution cases not seen in training data.
  • Real-world validation shows smooth autonomous task execution in dynamic environments for complex loco-manipulation tasks.
  • The approach substantially outperforms pre-trained diffusion policy baselines, demonstrating the importance of RL fine-tuning for humanoid robotics.
Read Original →via arXiv – CS AI
Act on this with AI
Stay ahead of the market.
Connect your wallet to an AI agent. It reads balances, proposes swaps and bridges across 15 chains — you keep full control of your keys.
Connect Wallet to AI →How it works
Related Articles