Research and Implementation of 77GHz Automotive Radar Target Detection Technology

Today, with the rapid development of autonomous driving technology, ensuring the accuracy of information when entering subsequent tracking algorithms is crucial. To address potential issues of target loss and information ambiguity in radar signal processing, a fast signal processing system suitable for high-resolution radar has been researched and proposed. This system relies on the MIMO radar framework. After signal preprocessing, the MIMO radar's equivalent aperture provides ADC data from multiple channels, thereby improving angular resolution. After obtaining discrete valid points in the R-D-A domain, dimensionality compression is performed to compress the R-D-A domain into the R-A domain. The valid points in the R-A domain are then transformed into the real Cartesian coordinate system and subjected to weighted DBSCAN clustering. Finally, computing the centroids of the valid cluster points yields the exact relative position of the targets for subsequent algorithm processing. This framework aims to minimize false alarms while utilizing the subsequent weighted clustering process to remove noise points and invalid targets. Overall, the algorithm has low complexity and achieves good target detection performance.