skip to main content

Explore Research Products in the NSF-PAR

Title: Learning Rope Manipulation Policies Using Dense Object Descriptors Trained on Synthetic Depth Data
Robotic manipulation of deformable 1D objects such as ropes, cables, and hoses is challenging due to the lack of high-fidelity analytic models and large configuration spaces. Furthermore, learning end-to-end manipulation policies directly from images and physical interaction requires significant time on a robot and can fail to generalize across tasks. We address these challenges using interpretable deep visual representations for rope, extending recent work on dense object descriptors for robot manipulation. This facilitates the design of interpretable and transferable geometric policies built on top of the learned representations, decoupling visual reasoning and control. We present an approach that learns point-pair correspondences between initial and goal rope configurations, which implicitly encodes geometric structure, entirely in simulation from synthetic depth images. We demonstrate that the learned representation - dense depth object descriptors (DDODs) - can be used to manipulate a real rope into a variety of different arrangements either by learning from demonstrations or using interpretable geometric policies. In 50 trials of a knot-tying task with the ABB YuMi Robot, the system achieves a 66% knot-tying success rate from previously unseen configurations. See https://tinyurl.com/rope-learning for supplementary material and videos.  more » « less
Award ID(s):
1730628
NSF-PAR ID:
10221258
Author(s) / Creator(s):
; ; ; ; ; ; ;
Date Published:
Journal Name:
2020 IEEE International Conference on Robotics and Automation (ICRA)
Page Range / eLocation ID:
9411 to 9418
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ( , International Conference on Machine Learning)
    Learning sensorimotor control policies from high-dimensional images crucially relies on the quality of the underlying visual representations. Prior works show that structured latent space such as visual keypoints often outperforms unstructured representations for robotic control. However, most of these representations, whether structured or unstructured are learned in a 2D space even though the control tasks are usually performed in a 3D environment. In this work, we propose a framework to learn such a 3D geometric structure directly from images in an end-to-end unsupervised manner. The input images are embedded into latent 3D keypoints via a differentiable encoder which is trained to optimize both a multi-view consistency loss and downstream task objective. These discovered 3D keypoints tend to meaningfully capture robot joints as well as object movements in a consistent manner across both time and 3D space. The proposed approach outperforms prior state-of-art methods across a variety of reinforcement learning benchmarks. Code and videos at https://buoyancy99.github.io/unsup-3d-keypoints/ 
    more » « less
  2. ; ; ( , International Conference on Machine Learning)
    null (Ed.)
    Learning sensorimotor control policies from highdimensional images crucially relies on the quality of the underlying visual representations. Prior works show that structured latent space such as visual keypoints often outperforms unstructured representations for robotic control. However, most of these representations, whether structured or unstructured are learned in a 2D space even though the control tasks are usually performed in a 3D environment. In this work, we propose a framework to learn such a 3D geometric structure directly from images in an end-toend unsupervised manner. The input images are embedded into latent 3D keypoints via a differentiable encoder which is trained to optimize both a multi-view consistency loss and downstream task objective. These discovered 3D keypoints tend to meaningfully capture robot joints as well as object movements in a consistent manner across both time and 3D space. The proposed approach outperforms prior state-of-art methods across a variety of reinforcement learning benchmarks. Code and videos at https://buoyancy99.github. io/unsup-3d-keypoints/. 
    more » « less
  3. ; ; ; ; ; ; ( , 5th Annual Conference on Robot Learning)
    We propose a visually-grounded library of behaviors approach for learning to manipulate diverse objects across varying initial and goal configurations and camera placements. Our key innovation is to disentangle the standard image-to-action mapping into two separate modules that use different types of perceptual input:(1) a behavior selector which conditions on intrinsic and semantically-rich object appearance features to select the behaviors that can successfully perform the desired tasks on the object in hand, and (2) a library of behaviors each of which conditions on extrinsic and abstract object properties, such as object location and pose, to predict actions to execute over time. The selector uses a semantically-rich 3D object feature representation extracted from images in a differential end-to-end manner. This representation is trained to be view-invariant and affordance-aware using self-supervision, by predicting varying views and successful object manipulations. We test our framework on pushing and grasping diverse objects in simulation as well as transporting rigid, granular, and liquid food ingredients in a real robot setup. Our model outperforms image-to-action mappings that do not factorize static and dynamic object properties. We further ablate the contribution of the selector's input and show the benefits of the proposed view-predictive, affordance-aware 3D visual object representations. 
    more » « less
  4. ( , Conference on Robot Learning)
    3D perceptual representations are well suited for robot manipulation as they easily encode occlusions and simplify spatial reasoning. Many manipulation tasks require high spatial precision in end-effector pose prediction, which typically demands high-resolution 3D feature grids that are computationally expensive to process. As a result, most manipulation policies operate directly in 2D, foregoing 3D inductive biases. In this paper, we introduce Act3D, a manipulation policy transformer that represents the robot’s workspace using a 3D feature field with adaptive resolutions dependent on the task at hand. The model lifts 2D pre-trained features to 3D using sensed depth, and attends to them to compute features for sampled 3D points. It samples 3D point grids in a coarse to fine manner, featurizes them using relative-position attention, and selects where to focus the next round of point sampling. In this way, it efficiently computes 3D action maps of high spatial resolution. Act3D sets a new state-of-the-art in RLBench, an established manipulation benchmark, where it achieves 10% absolute improvement over the previous SOTA 2D multi-view policy on 74 RLBench tasks and 22% absolute improvement with 3x less compute over the previous SOTA 3D policy. We quantify the importance of relative spatial attention, large-scale vision-language pre-trained 2D backbones, and weight tying across coarse-to-fine attentions in ablative experiments. Code and videos are available at our project site: https://act3d.github.io/. 
    more » « less
  5. ; ; ; ( , The International Journal of Robotics Research)
    null (Ed.)
    We present a framework for deformable object manipulation that interleaves planning and control, enabling complex manipulation tasks without relying on high-fidelity modeling or simulation. The key question we address is when should we use planning and when should we use control to achieve the task? Planners are designed to find paths through complex configuration spaces, but for highly underactuated systems, such as deformable objects, achieving a specific configuration is very difficult even with high-fidelity models. Conversely, controllers can be designed to achieve specific configurations, but they can be trapped in undesirable local minima owing to obstacles. Our approach consists of three components: (1) a global motion planner to generate gross motion of the deformable object; (2) a local controller for refinement of the configuration of the deformable object; and (3) a novel deadlock prediction algorithm to determine when to use planning versus control. By separating planning from control we are able to use different representations of the deformable object, reducing overall complexity and enabling efficient computation of motion. We provide a detailed proof of probabilistic completeness for our planner, which is valid despite the fact that our system is underactuated and we do not have a steering function. We then demonstrate that our framework is able to successfully perform several manipulation tasks with rope and cloth in simulation, which cannot be performed using either our controller or planner alone. These experiments suggest that our planner can generate paths efficiently, taking under a second on average to find a feasible path in three out of four scenarios. We also show that our framework is effective on a 16-degree-of-freedom physical robot, where reachability and dual-arm constraints make the planning more difficult. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email