DeepPCO: End-to-End Point Cloud Odometry through Deep Parallel Neural Network

Abstract

Odometry is of key importance for localization in the absence of a map. There is considerable work in the area of visual odometry (VO), and recent advances in deep learning have brought novel approaches to VO, which directly learn salient features from raw images. These learning-based approaches have led to more accurate and robust VO systems. However, they have not been well applied to point cloud data yet. In this work, we investigate how to exploit deep learning to estimate point cloud odometry (PCO), which may serve as a critical component in point cloud-based downstream tasks or learning-based systems. Specifically, we propose a novel end-to-end deep parallel neural network called DeepPCO, which can estimate the 6-DOF poses using consecutive point clouds. It consists of two parallel sub-networks to estimate 3-D translation and orientation respectively rather than a single neural network. We validate our approach on KITTI Visual Odometry/SLAM benchmark dataset with different baselines. Experiments demonstrate that the proposed approach achieves good performance in terms of pose accuracy.

LodoNet

“…There are few deep learning-based approaches for LiDAR odometry estimation that have comparable results. DeepPCO [32] only reports the results on its validation sequences. However they did not specify the unit of and .…”

“…The odometry is finally estimated by solving the SVD from the matched keypoint pairs. DeepPCO [32] generates panoramic-view of depth image projection to feed to it neural networks. L3-Net [16] proposes a learning-based LiDAR localization system by comparing the network-based feature vector between the current LiDAR frame and pre-build point cloud map followed by a recurrent neural network based smoothness module.…”

“…Odometry results on KITTI dataset Velas et al[30] DeepPCO[32] LO-Net[12] ours seq t_rel r_rel t_rel r_rel t_rel r_rel t_rel r_rel t_rel r_rel t_rel r_rel t_rel r_rel t_rel r_rel t_rel r_rel…”

Deep Learning Point Cloud Odometry

“…LO-Net managed to achieve competitive results to the classical baseline for LIDAR odometry which is LOAM. DeepPCO (Wang et al 2019) is an end-to-end LIDAR odometry framework that is composed of two sub-networks: a translation estimation network and a flow orientation network. The sub-networks form a deep parallel framework that regresses 6-DoF pose.…”

Deeplio: Deep Lidar Inertial Sensor Fusion for Odometry Estimation

“…Recently some efforts have been made to apply learning-based methods -i.e. neural networks, on LiDAR measurements, to estimate the pose of a mobile agent over time (Wang et al, 2019, Velas et al, 2018. Analogously to these works we propose a supervised learning-based end-to-end trainable fusion method, which utilizes neural networks to extract relevant features from the multi-modal LiDAR and IMU measurements and fuses these latent features to regress the motion encoded in them.…”