Faster GICP with Voxel-Based Computation: Dynamic voxel size refinement for Enhanced Accuracy

This paper introduces a method to enhance the real-time performance and robustness of the Generalized Iterative Closest Point (GICP) algorithm through Voxelized Covariance Calculation(VCC). Our approach builds on the Voxelized Generalized Iterative Closest Point (VGICP) and addresses the issues related to KDTree bottlenecks and voxel size parameters in VGICP. By improving the calculation of covariances using voxelization, the proposed method significantly reduces the computational resources required for this task, which is a primary bottleneck in GICP. Additionally, it ensures robustness by matching points based on voxel downsizing at each iteration. The algorithm was tested using the Kitti benchmark dataset, and experimental results confirm that it outperforms the existing VGICP, nano-GICP method in terms of both performance and accuracy. These enhancements in computational efficiency and accuracy make the proposed method a promising solution for real-time applications requiring precise point cloud registration.