On-Robot Learning With Equivariant Models
On-Robot Learning With Equivariant Models
复制标题
DOI：
--
复制
发表时间：
2022-03
期刊：
影响因子：
--
通讯作者：
Dian Wang;Ming Jia;Xu Zhu;R. Walters;Robert W. Platt
中科院分区：
文献类型：
--
作者：
Dian Wang;Ming Jia;Xu Zhu;R. Walters;Robert W. Platt
近年来，在计算机视觉和强化学习中，等变神经网络模型被证明能够提高任务的样本效率。本文在机器人上策略学习的背景下探讨了这一思想，其中策略必须完全在物理机器人系统上学习，而不参考模型、模拟器或离线数据集。我们专注于等变SAC在机器人操作中的应用，并探索了该算法的一些变体。最终，我们展示了在不到一到两个小时的挂钟时间内，通过在机器人上的体验完全学习几项非琐碎操作任务的能力。
Recently, equivariant neural network models have been shown to improve sample efficiency for tasks in computer vision and reinforcement learning. This paper explores this idea in the context of on-robot policy learning in which a policy must be learned entirely on a physical robotic system without reference to a model, a simulator, or an offline dataset. We focus on applications of Equivariant SAC to robotic manipulation and explore a number of variations of the algorithm. Ultimately, we demonstrate the ability to learn several non-trivial manipulation tasks completely through on-robot experiences in less than an hour or two of wall clock time.