arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2004.04954cs.CVcs.AIcs.LGcs.RO
收藏

学习在没有任何监督的真实感环境中进行视觉导航

Learning to Visually Navigate in Photorealistic Environments Without any Supervision

Univ. Grenoble Alpes(格勒诺布尔阿尔卑斯大学);Inria(法国国家信息与自动化研究所);CNRS(法国国家科学研究中心);Grenoble INP(格勒诺布尔国立理工学院);Facebook AI Research(脸书人工智能研究院)

Lina Mezghani, Sainbayar Sukhbaatar, Arthur Szlam, Armand Joulin, Piotr Bojanowski

摘要:

在现实环境中学习导航是一项具有挑战性的任务,其中代理必须完全依赖视觉输入,部分原因是缺乏位置信息使得难以在训练期间提供监督。在本文中,我们介绍了一种新的方法来学习导航图像输入没有外部监督或奖励。我们的方法包括三个阶段:学习第一人称视角的良好表现,然后学习使用记忆进行探索,最后学习通过设定自己的目标进行导航。该模型仅使用内在奖励进行训练,因此可以应用于任何具有图像观察的环境。我们通过训练一个代理来导航具有挑战性的照片般逼真的环境,从吉布森数据集与RGB输入只显示我们的方法的好处。

英文摘要:

Learning to navigate in a realistic setting where an agent must rely solely on visual inputs is a challenging task, in part because the lack of position information makes it difficult to provide supervision during training. In this paper, we introduce a novel approach for learning to navigate from image inputs without external supervision or reward. Our approach consists of three stages: learning a good representation of first-person views, then learning to explore using memory, and finally learning to navigate by setting its own goals. The model is trained with intrinsic rewards only so that it can be applied to any environment with image observations. We show the benefits of our approach by training an agent to navigate challenging photo-realistic environments from the Gibson dataset with RGB inputs only.

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。

↑