SegICP: Integrated Deep Semantic Segmentation and Pose Estimation

Abstract

Recent robotic manipulation competitions have highlighted that sophisticated robots still struggle to achieve fast and reliable perception of task-relevant objects in complex, realistic scenarios. To improve these systems' perceptive speed and robustness, we present SegICP, a novel integrated solution to object recognition and pose estimation. SegICP couples convolutional neural networks and multi-hypothesis point cloud registration to achieve both robust pixel-wise semantic segmentation as well as accurate and real-time 6-DOF pose estimation for relevant objects. Our architecture achieves 1cm position error and <5^\circ$ angle error in real time without an initial seed. We evaluate and benchmark SegICP against an annotated dataset generated by motion capture.

arXiv commentIROS camera-ready

References in corpus (3)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• Semantic Understanding of Scenes through the ADE20K Dataset

Cited by in corpus (25)

• Vision-based Robotic Grasping From Object Localization, Object Pose Estimation to Grasp Estimation for Parallel Grippers: A Review
• DeepICP: An End-to-End Deep Neural Network for 3D Point Cloud Registration
• Cross-view Semantic Segmentation for Sensing Surroundings
• Fruit Detection, Segmentation and 3D Visualisation of Environments in Apple Orchards
• PoseCNN: A Convolutional Neural Network for 6D Object Pose Estimation in Cluttered Scenes
• FECANet: Boosting Few-Shot Semantic Segmentation with Feature-Enhanced Context-Aware Network
• Robotic Pick-and-Place of Novel Objects in Clutter with Multi-Affordance Grasping and Cross-Domain Image Matching
• TEASER: Fast and Certifiable Point Cloud Registration
• Real-Time Object Pose Estimation with Pose Interpreter Networks
• Point Cloud Registration using Representative Overlapping Points
• TriVoC: Efficient Voting-based Consensus Maximization for Robust Point Cloud Registration with Extreme Outlier Ratios
• BIM-assisted object recognition for the on-site autonomous robotic assembly of discrete structures
• Learning Orientation Distributions for Object Pose Estimation
• Learning a Task-specific Descriptor for Robust Matching of 3D Point Clouds
• Automatic Robot Hand-Eye Calibration Enabled by Learning-Based 3D Vision
• Self-supervised 6D Object Pose Estimation for Robot Manipulation
• Robust Partial-to-Partial Point Cloud Registration in a Full Range
• A Polynomial-time Solution for Robust Registration with Extreme Outlier Rates
• SketchParse : Towards Rich Descriptions for Poorly Drawn Sketches using Multi-Task Hierarchical Deep Networks
• Rapid Pose Label Generation through Sparse Representation of Unknown Objects
• Joint Pose and Shape Estimation of Vehicles from LiDAR Data
• A Dynamical Perspective on Point Cloud Registration
• 3D Convolution on RGB-D Point Clouds for Accurate Model-free Object Pose Estimation
• DANIEL: A Fast and Robust Consensus Maximization Method for Point Cloud Registration with High Outlier Ratios
• In pixels we trust: From Pixel Labeling to Object Localization and Scene Categorization