Research on radar feature-level cooperative alignment technique based on point cloud processing

Aiming at the problems of more information loss in networked radar track-level alignment and high communication rate requirement for signal-level alignment, this paper proposes a radar feature-level cooperative alignment method based on point cloud processing. First, the distance-Doppler (RD) domain data collected by multiple radars are preprocessed, which are segmented into RD data of individual targets and discretized into point cloud form, retaining the data above a certain threshold. Then, the point cloud normal distribution transform (NDT) alignment algorithm is utilized to achieve same-target alignment in complex environments by taking the minimum of the alignment root mean square error (RMSE) as the same-target discrimination criterion. Finally, the practicality and robustness of this paper's method in complex scenarios with different interference environments, different radar spatial angle intervals, and different signal-to-noise ratios are verified through simulation experiments and performance analysis.