Abstract:We present DeepICP - a novel end-to-end learning-based 3D point cloud registration framework that achieves comparable registration accuracy to prior state-of-the-art geometric methods. Different from other keypoint based methods where a RANSAC procedure is usually needed, we implement the use of various deep neural network structures to establish an end-to-end trainable network. Our keypoint detector is trained through this end-to-end structure and enables the system to avoid the inference of dynamic objects, leverages the help of sufficiently salient features on stationary objects, and as a result, achieves high robustness. Rather than searching the corresponding points among existing points, the key contribution is that we innovatively generate them based on learned matching probabilities among a group of candidates, which can boost the registration accuracy. Our loss function incorporates both the local similarity and the global geometric constraints to ensure all above network designs can converge towards the right direction. We comprehensively validate the effectiveness of our approach using both the KITTI dataset and the Apollo-SouthBay dataset. Results demonstrate that our method achieves comparable or better performance than the state-of-the-art geometry-based methods. Detailed ablation and visualization analysis are included to further illustrate the behavior and insights of our network. The low registration error and high robustness of our method makes it attractive for substantial applications relying on the point cloud registration task.

Deep Learning Based Pose Estimation Method Using 3D Point Clouds

A Deep Learnable Framework for 3D Point Clouds Pose Transformation Regression

Object Pose Estimation with Point Cloud Data for Robot Grasping

Simulation and Deep Learning on Point Clouds for Robot Grasping

PA-Pose: Partial Point Cloud Fusion Based on Reliable Alignment for 6D Pose Tracking

Deep Learning Based 6D Pose Estimation for Bin-Picking Using 3D Point Clouds

A Registration Method for 3D Point Clouds with Convolutional Neural Network.

Head Pose Estimation Using Deep Neural Networks and 3D Point Clouds.

DeepICP: An End-to-End Deep Neural Network for 3D Point Cloud Registration

3D Convolution on RGB-D Point Clouds for Accurate Model-free Object Pose Estimation

A Hybrid Approach for Cross-modality Pose Estimation Between Image and Point Cloud

6-D Object Pose Estimation Using Multiscale Point Cloud Transformer.

A multilevel object pose estimation algorithm based on point cloud keypoints

6D Object Pose Regression Via Supervised Learning on Point Clouds.

Instance segmentation based 6D pose estimation of industrial objects using point clouds for robotic bin-picking

MVPointNet: Multi-View Network for 3D Object Based on Point Cloud

Sequential 3D Human Pose and Shape Estimation from Point Clouds

Sparse Convolution Based 6D Pose Estimation for Robotic Bin-Picking with Point Clouds

An Integrated Framework for 3-D Modeling, Object Detection, and Pose Estimation from Point-Clouds

Rgb-D Fusion for Point-Cloud-Based 3d Human Pose Estimation

PointSkelCNN: Deep Learning-Based 3D Human Skeleton Extraction from Point Clouds