Optimal Target Shape for LiDAR Pose Estimation

Targets are essential in problems such as object tracking cluttered or textureless environments, camera (and multi-sensor) calibration tasks, and simultaneous localization mapping (SLAM). Target shapes for these tasks typically symmetric (square, rectangular, circular) work well structured, dense sensor data pixel arrays (i.e., image). However, lead to pose ambiguity when using sparse LiDAR point clouds suffer from the quantization uncertainty of LiDAR. This paper introduces concept optimizing target shape remove clouds. A is designed induce large gradients at edge points under rotation translation relative ameliorate associated with cloud sparseness. Moreover, given a shape, we present means that leverages target's geometry estimate vertices while globally estimating pose. Both simulation experimental results (verified by motion capture system) confirm optimal global solver, achieve centimeter error few degrees even partially illuminated placed 30 meters away. All implementations datasets available https://github.com/UMich-BipedLab/optimal_shape_global_pose_estimation.