One of the key challenges in visual imitation learning is collecting large amounts of expert demonstrations for a given task. While methods for collecting human demonstrations are becoming easier with teleoperation methods and the use of low-cost assistive tools, we often still require 100-1000 demonstrations for every task to learn a visual representation and policy. To address this, we turn to an alternate form of data that does not require task-specific demonstrations -- play. Playing is a fundamental method children use to learn a set of skills and behaviors and visual representations in early learning. Importantly, play data is diverse, task-agnostic, and relatively cheap to obtain. In this work, we propose to use playful interactions in a self-supervised manner to learn visual representations for downstream tasks. We collect 2 hours of playful data in 19 diverse environments and use self-predictive learning to extract visual representations. Given these representations, we train policies using imitation learning for two downstream tasks: Pushing and Stacking. We demonstrate that our visual representations generalize better than standard behavior cloning and can achieve similar performance with only half the number of required demonstrations. Our representations, which are trained from scratch, compare favorably against ImageNet pretrained representations. Finally, we provide an experimental analysis on the effects of different pretraining modes on downstream task learning.
The Turkish Ice Cream Robot: Examining Playful Deception in Social Human-Robot Interactions
Department of Artificial Intelligence, Korea University(人工智能系,韩国大学);Ingenuity Labs Research Institute and the Department of Electrical and Computer Engineering, Queen’s University(创新实验室研究机构和电气与计算机工程系,皇后大学);Department of Artificial Intelligence, Yonsei University(人工智能系,延世大学);Robotics Institute, Carnegie Mellon University(机器人研究所,卡内基梅隆大学)
ImageDDI: Image-enhanced Molecular Motif Sequence Representation for Drug-Drug Interaction Prediction
College of Computer science and Electronic Engineering, Hunan University(计算机科学与电子工程学院,湖南大学);Department of Integrative Biotechnology, Yonsei University(整合生物技术系,延世大学)
Neighbour-level Message Interaction Encoding for Improved Representation Learning on Graphs
School of Electrical and Data Engineering, Faculty of Engineering and Information Technology, University of Technology Sydney(悉尼科技大学工程与信息技术学院工程与数据工程学院)
MSRFormer: Road Network Representation Learning using Multi-scale Feature Fusion of Heterogeneous Spatial Interactions
Information Engineering University(信息工程大学);Fujian Normal University(福建师范大学);Quanzhou Institute of Equipment Manufacturing, Haixi Institute, Chinese Academy of Sciences(中国科学院海西研究院泉州装备制造研究所);Norwegian University of Science and Technology(挪威科技大学);China Centre for Resources Satellite Data and Application(中国资源卫星应用中心);Henan Twenty First Century Aerospace Technology Co, Ltd.(河南二十一世纪航天技术有限公司);Shandong University of Science and Technology(山东科技大学)
DensePhysNet: Learning Dense Physical Object Representations via Multi-step Dynamic Interactions
Shanghai Jiao Tong University(上海交通大学);Massachusetts Institute of Technology(麻省理工学院);Princeton University(普林斯顿大学);Google(谷歌);MIT Center for Brains, Minds and Machines(麻省理工学院大脑、心智与机器中心);Columbia University(哥伦比亚大学)