代表(政治)
可微函数
水准点(测量)
物理引擎
计算机科学
强化学习
管道(软件)
对象(语法)
人工神经网络
国家(计算机科学)
状态空间
人工智能
动力学仿真
运动(物理)
机械臂
机器人
循环神经网络
算法
数学
数学分析
统计
物理
大地测量学
量子力学
政治
政治学
法学
程序设计语言
地理
作者
Sirui Chen,Yunhao Liu,Shang Wen Yao,Jialong Li,Tingxiang Fan,Jia Pan
出处
期刊:IEEE robotics and automation letters
日期:2022-10-01
卷期号:7 (4): 9533-9540
被引量:2
标识
DOI:10.1109/lra.2022.3192209
摘要
Dynamic state representation learning is essential for robot learning. Good latent space that can accurately describe dynamic transition and constraints can significantly accelerate reinforcement learning training as well as reduce motion planning complexity. However, deformable object have very complicated dynamics and is hard to be represented directly by a neural network without any prior physics information. We propose DiffSRL, an end-to-end dynamic state representation learning pipeline that uses differentiable physics engine to teach neural network how to represent high dimensional pointcloud data collected from deformable objects. Our specially designed loss function can guide neural network aware physics constraints and feasibility. We benchmark the performance of our methods as well as other state representation algorithms with multiple downstream tasks on PlasticineLab. Our model demonstrates superior performance most of the time on all tasks. We also demonstrate our model's performance in real hardware setting with two manipulation tasks on a UR-5 robot arm.
科研通智能强力驱动
Strongly Powered by AbleSci AI