Transporter Networks: Rearranging the Visual World for Robotic Manipulation

Abstract

Robotic manipulation can be formulated as inducing a sequence of spatial displacements: where the space being moved can encompass an object, part of an object, or end effector. In this work, we propose the Transporter Network, a simple model architecture that rearranges deep features to infer spatial displacements from visual input -which can parameterize robot actions. It makes no assumptions of objectness (e.g. canonical poses, models, or keypoints), it exploits spatial symmetries, and is orders of magnitude more sample efficient than our benchmarked alternatives in learning vision-based manipulation tasks: from stacking a pyramid of blocks, to assembling kits with unseen objects; from manipulating deformable ropes, to pushing piles of small objects with closed-loop feedback. Our method can represent complex multi-modal policy distributions and generalizes to multi-step sequential tasks, as well as 6DoF pick-and-place. Experiments on 10 simulated tasks show that it learns faster and generalizes better than a variety of end-to-end baselines, including policies that use ground-truth object poses. We validate our methods with hardware in the real world. Experiment videos and code will be made available at https://transporternets.github.io

Assistive Tele-op: Leveraging Transformers to Collect Robotic Task Demonstrations

“…We evaluated the transformer and Assistive Tele-op system across a variety of tasks representing industrial [18,31], household [4,8], and caregiving [7,10] task scenarios. For this, we used both existing data from the Roboturk [18] simulation dataset covering pick-and-place and nut assembly tasks, as well as new data for other tasks that we collected in VR.…”

Equivariant $Q$ Learning in Spatial Action Spaces

“…Variations on this approach have been explored in [22,23]. The FCN approach has been adapted to a variety of different manipulation tasks with different action primitives [24,25,26,27,28,1,29,30,31]. In this paper, we extend the work above by proposing new equivariant architectures for the spatial action space setting.…”

“…Recent work demonstrates the use of data augmentation improves the data efficiency and the policy's performance in reinforcement learning [7,8,9]. In the context of robotics, data augmentation is often used to generate additional samples [1,10,11]. In contrast to learning the equivariance property using data augmentation, our work utilizes the equivariant network to hard code the symmetries in the structure of the network to achieve better sample efficiency.…”

“…The agent's value function or policy typically considers different perspectives on the same scene to be different world states. A popular way to combat this problem is through visual data augmentation, i.e., to create additional samples or experiences by randomly translating and rotating observed images [1] but keeping the same labels. This can be used in conjunction with a contrastive term in the loss function which helps the system learn an invariant latent representation [2,3].…”

RLDS: an Ecosystem to Generate, Share and Use Datasets in Reinforcement Learning

“…Recently, with the growing interest of the research community for IL, LfD or ORL, a lot of datasets have been released. They mainly focus on robotics [15,44,48,40], some with a particular focus on human data [31,32,39]. Some works include datasets for discrete action-environments like games [19,28].…”