Quantitative Assessment of mmWave-Based Point Cloud Generation Pipelines for Target Detection

Abstract

This thesis tackles the challenge of employing quantifiable metrics to assess the quality of point clouds generated by various distinct pipelines using TI IWR6843AOP mmWave FMCW radar. This study focuses on developing quantifiable metrics to evaluate point cloud quality for both static targets, such as the corner reflectors used in this research, and human targets. For static point targets, this study introduces metrics that combine Euclidean distance errors with range, azimuth, and elevation angle errors, providing a more comprehensive assessment compared to using Euclidean distance errors alone. For human targets, this thesis introduces metrics from two perspectives. The first focuses on coverage, employing the Euclidean distance to the human mesh surface to quantify errors between the ground truth human mesh and the point cloud. Additionally, It calculates the Euclidean distance between each point and all joints, selecting the minimum distance to determine the closest joint and evaluating the percentage of points reflected from each body segment. The second focus is on consistency. Point cloud consistency across consecutive frames is assessed by analyzing the mean and maximum intensity values and calculating Hausdorff distances to evaluate the stability of the point cloud distribution.