A Two-Step Multiframe Assignment Method for Multiple Extended Target Tracking With Azimuth Ambiguity Based on Pseudo Measurement Set

Abstract

Autonomous vehicle technology uses perception modules to detect and track people and objects around autonomous vehicles. These perception modules often employ high-resolution radars for tracking due to their relatively low cost and high angle resolution. A major challenge in using high-resolution radar for advanced driver assistance systems (ADAS) or autonomous vehicles is the azimuth ambiguity and related split ambiguity for extended targets, which are caused by grating lobes due to the large physical distance between radar antenna elements relative to signal wavelength. These grating lobes give false measurements in a direction different from the target's actual direction. Due to azimuth ambiguity, data association with multiframe measurements using traditional methods is limited and generates large missed detection errors. To address this limitation, a two-step multiframe assignment method is proposed to resolve the split and azimuth ambiguity separately. In the first step, each ambiguous radar measurement (cluster) is used to generate a pseudo measurement set (PMS). Then, the split ambiguity is resolved by the PMS-to-PMS association, resulting in a merged PMS (MPMS); in the second step, the azimuth ambiguity is resolved by the Track-to-MPMS association. Simulation results indicate that the proposed method outperforms the pseudo 3-D assignment (P3DA)-PMS method in reducing false detection errors when resolving azimuth ambiguity, due to the utilization of additional frame data. The effectiveness of the proposed method in addressing azimuth ambiguity is also demonstrated using real data.