The success of deep learning depends heavily on the availability of large datasets, but in robotic manipulation there are many learning problems for which such datasets do not exist. Collecting these datasets is time-consuming and expensive, and therefore learning from small datasets is an important open problem. Within computer vision, a common approach to a lack of data is data augmentation. Data augmentation is the process of creating additional training examples by modifying existing ones. However, because the types of tasks and data differ, the methods used in computer vision cannot be easily adapted to manipulation. Therefore, we propose a data augmentation method for robotic manipulation. We argue that augmentations should be valid, relevant, and diverse. We use these principles to formalize augmentation as an optimization problem, with the objective function derived from physics and knowledge of the manipulation domain. This method applies rigid body transformations to trajectories of geometric state and action data. We test our method in two scenarios: 1) learning the dynamics of planar pushing of rigid cylinders, and 2) learning a constraint checker for rope manipulation. These two scenarios have different data and label types, yet in both scenarios, training on our augmented data significantly improves performance on downstream tasks. We also show how our augmentation method can be used on real-robot data to enable more data-efficient online learning.
Object-Centric Instruction Augmentation for Robotic Manipulation
School of Computer Science, East China Normal University(华东师范大学计算机科学学院);Midea Group(美的集团);Department of Mathematics, School of Science, Shanghai University(上海大学理学院数学系)
Shape-Space Deformer: Unified Visuo-Tactile Representations for Robotic Manipulation of Deformable Objects
CSIRO Robotics, Data61, CSIRO(CSIRO机器人部、Data61、CSIRO);School of Electrical Engineering and Computer Science (EECS), The University of Queensland(电气工程与计算机科学学院(EECS)、昆士兰大学)
Benchmarking and Enhancing Skill-Level Memory for Partially Observable Robotic Manipulation
University of Science and Technology of China(中国科学技术大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);Zhejiang University(浙江大学);Shanghai Jiaotong University(上海交通大学);Nanjing University(南京大学)
FoAM: Foresight-Augmented Multi-Task Imitation Policy for Robotic Manipulation
University of Southern California(南加州大学);Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所);School of Artificial Intelligence, Shanghai Jiao Tong University(上海交通大学人工智能学院)
MV-WAM: Manifold-Aware World Action Model with Value Augmentation
State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University(北京大学计算机学院多媒体信息处理国家重点实验室);Beijing Innovation Center of Humanoid Robotics(北京人形机器人创新中心)
DeformGen: Dynamics-Based Topology Augmentation for Deformable Manipulation Policy Learning
Shanghai Jiao Tong University(上海交通大学);Eastern Institute of Technology, Ningbo(宁波东方理工大学);Tsinghua University(清华大学);The Hong Kong Polytechnic University(香港理工大学);University of Science and Technology of China(中国科学技术大学);Zhongguancun Academy(中关村学院);Microsoft Research(微软研究院)
Stabilizing Deep Q-Learning with ConvNets and Vision Transformers under Data Augmentation
University of California, San Diego(加利福尼亚大学圣迭戈分校)
虽然通过强化学习(RL)训练的智能体可以直接从视觉观察中解决越来越具有挑战性的任务,但将学习到的技能推广到新的环境仍然非常具有挑战性。广泛使用数据增强是一种很有前途的提高RL泛化能力的技术,但它通常会降低样本效率,甚至会导致发散。在本文中,我们调查的不稳定性的原因时,使用数据增强在常见的非政策RL算法。我们确定了两个问题,都植根于高方差Q目标。基于我们的研究结果,我们提出了一个简单而有效的技术,稳定这类算法下增强。我们使用ConvNets和视觉Transformer(ViT)对基于DeepMind Control Suite的一系列基准以及机器人操作任务进行了广泛的基于图像的RL的实证评估。我们的方法大大提高了增强条件下ConvNets的稳定性和样本效率,并在具有不可见视觉效果的环境中实现了与基于图像的RL的最先进方法竞争的泛化结果。我们进一步表明,我们的方法可扩展到基于ViT的架构的RL,并且数据增强在这种情况下可能特别重要。
AXIS: A Growable Community-Driven Data Engine for Scalable Robot Manipulation
Axis Robotics(轴机器人公司);University of California, Berkeley(加州大学伯克利分校);Georgia Institute of Technology(佐治亚理工学院);Texas A&M University(德州农工大学);Johns Hopkins University(约翰·霍普金斯大学);University of Pennsylvania(宾夕法尼亚大学);University of Michigan(密歇根大学);National University of Singapore(新加坡国立大学);Nanyang Technological University(南洋理工大学)
$\mathbf{M^3A}$ Policy: Mutable Material Manipulation Augmentation Policy through Photometric Re-rendering
Tsinghua University(清华大学);MARS Lab, Nanyang Technological University(南洋理工大学MARS实验室);University of California, Berkeley(加州大学伯克利分校);The University of Tokyo(东京大学)
RESTA: Time-Scaling Augmentation for Data-Efficient Robot Imitation Learning
Global College, Shanghai Jiao Tong University(上海交通大学Global College);Shanghai Institute of Physical Intelligence and Robotics (PAIR)(上海物理智能与机器人研究所(PAIR));Duke Kunshan University(昆山杜克大学)
MinInter: Minimizing Trajectory Interpolation During Data Augmentation for Imitation Learning
School of Automation and Intelligent Manufacturing, Southern University of Science and Technology(南方科技大学自动化与智能制造学院);Pengcheng Laboratory(鹏城实验室);Carnegie Mellon University(卡内基梅隆大学)
IA-VLA: Input Augmentation for Vision-Language-Action models in settings with semantically complex tasks
Intelligent Robotics Group at the Department of Electrical Engineering and Automation, School of Electrical Engineering, Aalto University(Aalto大学电气工程学院电气工程与自动化系智能机器人组);Biomimetics and Intelligent Systems Group at the Faculty of Information Technology and Electrical Engineering, University of Oulu(奥卢大学信息科技与电气工程学院仿生学与智能系统组);Section of Mechanical Technology at the Department of Engineering Technology and Didactics, Technical University of Denmark(丹麦技术大学工程技术与教学系机械技术部门)
DeGuV: Depth-Guided Visual Reinforcement Learning for Generalization and Interpretability in Manipulation
Cognitive Robotics Lab, School of Computer Science, University of Manchester(认知机器人实验室,计算机科学学院,曼彻斯特大学);Learning and Adaptive Robotics Lab, Department of Computer Science and Engineering, University of Texas at Arlington(学习与自适应机器人实验室,计算机科学与工程系,德克萨斯大学阿灵顿分校)