SAM-RL: Sensing-Aware Model-based Reinforcement Learning via Differentiable Physics-based Simulation and Rendering
Jun Lv, Yunhai Feng, Cheng Zhang, Shuang Zhao, Lin Shao* and Cewu Lu*
Abstract:
Model-based reinforcement learning (MBRL) is recognized with the potential to be significantly more sample efficient than model-free RL. How an accurate model can be developed automatically and efficiently from raw sensory inputs (such as images), especially for complex environments and tasks, is a challenging problem that hinders the broad application of MBRL in the real world. In this work, we propose a sensing-aware model-based reinforcement learning called SAM-RL. Leveraging differentiable physics-based simulation and rendering, SAM-RL automatically updates the model by comparing rendered images with real raw images and produces the policy efficiently. Our sensing-aware learning pipeline allows a robot to select an informative viewpoint to monitor the task process. We apply our framework to real-world experiments for accomplishing three manipulation tasks: robotic assembly, tool manipulation, and deformable object manipulation. We demonstrate the effectiveness of SAM-RL via extensive experiments.
Paper:
Video:
Video:
Diff Rendering Video
![](https://www.google.com/images/icons/product/drive-32.png)
![](https://www.google.com/images/icons/product/drive-32.png)
Bibtex
@article{lv2022sam,
title={SAM-RL: Sensing-Aware Model-Based Reinforcement Learning via Differentiable Physics-Based Simulation and Rendering},
author={Lv, Jun and Feng, Yunhai and Zhang, Cheng and Zhao, Shuang and Shao, Lin and Lu, Cewu},
booktitle={Proceedings of Robotics: Science and Systems (RSS)},
year={2023},
}