Robust Grasp Planning Over Uncertain Shape Completions

Abstract

We present a method for planning robust grasps over uncertain shape completed objects. For shape completion, a deep neural network is trained to take a partial view of the object as input and outputs the completed shape as a voxel grid. The key part of the network is dropout layers which are enabled not only during training but also at run-time to generate a set of shape samples representing the shape uncertainty through Monte Carlo sampling. Given the set of shape completed objects, we generate grasp candidates on the mean object shape but evaluate them based on their joint performance in terms of analytical grasp metrics on all the shape candidates. We experimentally validate and benchmark our method against another state-of-the-art method with a Barrett hand on 90000 grasps in simulation and 200 grasps on a real Franka Emika Panda. All experimental results show statistically significant improvements both in terms of grasp quality metrics and grasp success rate, demonstrating that planning shape-uncertainty-aware grasps brings significant advantages over solely planning on a single shape estimate, especially when dealing with complex or unknown objects.

arXiv comment7 pages, 5 figures. Accepted to 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2019)

References in corpus (5)

• Adam: A Method for Stochastic Optimization
• Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
• Scalable Bayesian Optimization Using Deep Neural Networks
• RGB-D Object Detection and Semantic Segmentation for Autonomous Manipulation in Clutter
• Multi-Modal Geometric Learning for Grasping and Manipulation

Cited by in corpus (16)

• Vision-based Robotic Grasping From Object Localization, Object Pose Estimation to Grasp Estimation for Parallel Grippers: A Review
• Multi-FinGAN: Generative Coarse-To-Fine Sampling of Multi-Finger Grasps
• Robotic Pick-and-Place With Uncertain Object Instance Segmentation and Shape Completion
• Beyond Top-Grasps Through Scene Completion
• Active Visuo-Haptic Object Shape Completion
• TransSC: Transformer-based Shape Completion for Grasp Evaluation
• Amodal 3D Reconstruction for Robotic Manipulation via Stability and Connectivity
• Synergies Between Affordance and Geometry: 6-DoF Grasp Detection via Implicit Representations
• RGB Matters: Learning 7-DoF Grasp Poses on Monocular RGBD Images
• Shape Completion with Prediction of Uncertain Regions
• DDGC: Generative Deep Dexterous Grasping in Clutter
• Deformation-Aware Data-Driven Grasp Synthesis
• PCF-Grasp: Converting Point Completion to Geometry Feature to Enhance 6-DoF Grasp
• Dexterous grasp data augmentation based on grasp synthesis with fingertip workspace cloud and contact-aware sampling
• Towards synthesizing grasps for 3D deformable objects with physics-based simulation
• Robust shape estimation with false-positive contact detection