VisuoSpatial Foresight for Physical Sequential Fabric Manipulation

Abstract

Robotic fabric manipulation has applications in home robotics, textiles, senior care and surgery. Existing fabric manipulation techniques, however, are designed for specific tasks, making it difficult to generalize across different but related tasks. We build upon the Visual Foresight framework to learn fabric dynamics that can be efficiently reused to accomplish different sequential fabric manipulation tasks with a single goal-conditioned policy. We extend our earlier work on VisuoSpatial Foresight (VSF), which learns visual dynamics on domain randomized RGB images and depth maps simultaneously and completely in simulation. In this earlier work, we evaluated VSF on multi-step fabric smoothing and folding tasks against 5 baseline methods in simulation and on the da Vinci Research Kit (dVRK) surgical robot without any demonstrations at train or test time. A key finding was that depth sensing significantly improves performance: RGBD data yields an 80% improvement in fabric folding success rate in simulation over pure RGB data. In this work, we vary 4 components of VSF, including data generation, visual dynamics model, cost function, and optimization procedure. Results suggest that training visual dynamics models using longer, corner-based actions can improve the efficiency of fabric folding by 76% and enable a physical sequential fabric folding task that VSF could not previously perform with 90% reliability. Code, data, videos, and supplementary material are available at https://sites.google.com/view/fabric-vsf/.

arXiv commentJournal extension of prior work on VSF to appear in Autonomous Robots S.I. 207. arXiv admin note: text overlap with arXiv:2003.09044

References in corpus (18)

• Continuous control with deep reinforcement learning
• Leveraging Demonstrations for Deep Reinforcement Learning on Robotics Problems with Sparse Rewards
• Stochastic Video Generation with a Learned Prior
• Safe Controller Optimization for Quadrotors with Gaussian Processes
• Visual Foresight: Model-Based Deep Reinforcement Learning for Vision-Based Robotic Control
• Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models
• Cautious NMPC with Gaussian Process Dynamics for Autonomous Miniature Race Cars
• Learning Predictive Representations for Deformable Objects Using Contrastive Estimation
• SoftGym: Benchmarking Deep Reinforcement Learning for Deformable Object Manipulation
• Few-Shot Goal Inference for Visuomotor Learning and Planning
• VideoFlow: A Conditional Flow-Based Model for Stochastic Video Generation
• Goal-Aware Prediction: Learning to Model What Matters
• On-Policy Robot Imitation Learning from a Converging Supervisor
• VisuoSpatial Foresight for Multi-Step, Multi-Task Fabric Manipulation
• Recovery RL: Safe Reinforcement Learning with Learned Recovery Zones
• Learning Arbitrary-Goal Fabric Folding with One Hour of Real Robot Experience
• Time Reversal as Self-Supervision
• MMGSD: Multi-Modal Gaussian Shape Descriptors for Correspondence Matching in 1D and 2D Deformable Objects