| Summary: | In recent years, LiDAR (Light Detection and Ranging) technology has found extensive applications in SLAM (Simultaneous Localization and Mapping), delivering impressive results across a wide array of scenarios. However, LiDAR-based SLAM systems still face significant challenges in complex environments characterized by degenerate geometrical features, such as long corridors, and dynamic objects. The inherent limitations in purely geometric sensing can lead to failures in maintaining accurate localization and mapping.
To address these challenges, recent advancements have introduced 4D RADAR technology, which, despite having lower geometric accuracy compared to LiDAR, offers the unique ability to detect the velocity of objects. This velocity information can be crucial in scenarios where LiDAR-based SLAM systems struggle, particularly in environments with moving objects or ambiguous geometrical features.
In this dissertation, the whole framework architecture was designed and the the function of the GM6020 motor rotating at a given speed was implemented. The corresponding time and angle information was recorded, as well as the process of obtaining and optimising the initial point cloud data from the lidar. As this system involves the rotation and integration of 3D coordinates, there are bound to be errors (usually due to a slight error in the angle and distance between the lidar and the motor in the ideal building system). In order to minimise the effect of this error, the initial point cloud data is first processed and reduced by voxel grid sampling, then the processed point cloud data is subjected to KD tree nearest neighbour search with suitable processing method, which gives the result of calculated normal vector and a satisfied evaluation of the quality of plane fitting. Finally, non-linear least squares optimisation is carried out using Ceres solver to optimise the alignment process of point cloud data to get the result after plane fitting, which reduces the effect of errors due to system construction.
|
|---|