Dual LIDAR online calibration and mapping and perception system

Abstract

Making autonomous mobile devices capable of autonomous positioning and map building in a GPS-denied environment as well as being able to circumvent people in the operating area simultaneously, is essential for many UGV appliances. In this paper, a mapping and positioning system using dual LIDAR is proposed for positioning and building maps. The system may also detect people in the operating space of the equipment using infrared heat maps. The localization and mapping information from each LIDAR is unified in a loosely coupled approach after simultaneous extrinsic calibration. We propose a method for constructing factor graph using LIDAR point cloud geometric features to optimally solve the dual LIDAR extrinsic in real time. We tested the localization accuracy with a publicly available dataset and three real-world scenarios and compared it with three existing methods. The test results in open-source datasets and real scenarios show that our proposed method improves the position and attitude estimation by up to 50% compared to the three existing methods, and the proposed dual LIDAR extrinsic calibration method can achieve an estimation accuracy of 0.05m for the translational extrinsic and 1deg for the rotational extrinsic. In addition, our proposed method achieves well position and attitude estimation when other existing methods show severe drift in trajectory estimation in real scenarios. Thus, our proposed method is suitable for obtaining high accuracy measurements of LIDAR extrinsic parameters and for dense mapping and accurate localization in the environment of GNSS-denied and human mobility.

Hierarchical fusion based high precision SLAM for solid-state lidar

“…However, in standalone configurations utilizing a single LiDAR sensor, the scan matching-based approach of LiDAR odometry is sensitive to the scan rate, and the sensed point cloud is prone to distortion caused by the sensor motion. Consequently, this sensitivity can lead to a degradation in the performance of the odometry system, especially in high-speed or complex scenarios where the sensor motion is more pronounced [9].…”

“…Since the initialization does not have the state of the previous keyframe, it is assumed that the current frame is similar to the reference frame, and then the point cloud tracking changes is used to calculate the initial threshold. The initial threshold is defined as equation (9).…”

Automated Method for SLAM Evaluation in GNSS-Denied Areas

“…Most research on multi-LiDAR system configurations is conducted in the field of autonomous driving for road scenarios [35,36]. However, also for other scenarios, this is crucial due to the increasing use of multi-LiDAR systems in recent works [14,37,38]. By using the proposed time synchronization method, more combinations of different FOVs, scanning lines, and spatial alignments for different platforms can be tested using the benefits of time correspondence in GNSS-denied areas.…”

Automatic multi-LiDAR calibration and refinement method

“…The extrinsic parameters were calibrated manually in the early decade. There are now many sensor extrinsic calibration methods which can be divided into target-based ones [2,[5][6][7][8][9][10][11][12], motion-based ones [1,[13][14][15][16][17][18][19][20][21][22] and target-less ones [23][24][25][26]. Target-based methods need auxiliary objects that have distinguished geometric features.…”