Trajectory Prediction Method of Millimeter-Wave Radar Based on Markov Model for Roadside Installation Scenario

Abstract

In the application of roadside traffic detection, when the wide-area millimeter-wave radar detects the target, due to the limits of target occlusion, detection angle and detection range, the problem of missing target trajectory points will be inevitable. Aiming at solving this problem, a target trajectory prediction method based on Markov is proposed. This method predicts the four characteristic values of the target: radial distance, radial speed, horizontal distance and horizontal speed respectively. In order to ensure the accuracy of trajectory prediction, the trajectory pre-processing is carried out by means of threshold filtering and smooth filtering, and the Markov prediction model is established after the data preprocessing completed, then the trajectory prediction is finally achieved. In this paper, the Markov model is used as a sample to predict the target trajectory by predicting the four eigenvalues of the actual data. The results show that the Markov model is used to realize the target trajectory prediction of millimeter-wave radar detection with high accuracy, which can be applied to actual traffic detection.