arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2011.06698cs.ROcs.CVcs.LG
收藏

基于中层视觉表征的鲁棒策略:操纵和导航的实验研究

Robust Policies via Mid-Level Visual Representations: An Experimental Study in Manipulation and Navigation

University of California, Berkeley(加州大学伯克利分校);Stanford University(斯坦福大学);New York University(纽约大学);Swiss Federal Institute of Technology Lausanne (EPFL)(洛桑联邦理工学院)

Bryan Chen, Alexander Sax, Gene Lewis, Iro Armeni, Silvio Savarese, Amir Zamir, Jitendra Malik, Lerrel Pinto

摘要:

基于视觉的机器人通常将控制回路分为一个感知模块和一个单独的控制模块。可以对整个系统进行端到端的训练(例如,使用深度RL),但“从头开始”进行训练会带来很高的样本复杂性成本,并且最终结果通常很脆弱,如果测试环境与训练环境不同,则会意外失败。我们研究了使用中级视觉表示(传统计算机视觉目标异步学习的特征)作为端到端RL框架中通用且易于解码的感知状态的影响。中层表示编码的不变性的世界,我们表明,他们的援助泛化,提高样本的复杂性,并导致更高的最终性能。与其他包含不变性的方法(如域随机化)相比,异步训练的中级表示扩展得更好:无论是更难的问题还是更大的域转移。在实践中,这意味着中级表示可以用于成功地训练策略,以用于域随机化和从头开始学习失败的任务。我们报告结果的操纵和导航任务,并为导航包括zero-shot模拟到真实的机器人实验。

英文摘要:

Vision-based robotics often separates the control loop into one module for perception and a separate module for control. It is possible to train the whole system end-to-end (e.g. with deep RL), but doing it "from scratch" comes with a high sample complexity cost and the final result is often brittle, failing unexpectedly if the test environment differs from that of training. We study the effects of using mid-level visual representations (features learned asynchronously for traditional computer vision objectives), as a generic and easy-to-decode perceptual state in an end-to-end RL framework. Mid-level representations encode invariances about the world, and we show that they aid generalization, improve sample complexity, and lead to a higher final performance. Compared to other approaches for incorporating invariances, such as domain randomization, asynchronously trained mid-level representations scale better: both to harder problems and to larger domain shifts. In practice, this means that mid-level representations could be used to successfully train policies for tasks where domain randomization and learning-from-scratch failed. We report results on both manipulation and navigation tasks, and for navigation include zero-shot sim-to-real experiments on real robots.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑