Study On Highway Guardrail Detection Method Using Lidar

Environmental perception is a fundamental technology for path planning and decision-making in intelligent vehicles.The safest area and its road structure can be obtained by guardrail detection.Therefore,guardrail detection is an essential component in environmental perception,and it has become a major study and development topic in the field of intelligent vehicles.To the best of our knowledge,there is little literature dealing with guardrail detection.This thesis studies the problem of highway guardrail detection using the lidar.In order to solve this problem,the feature of highway guardrail detection is extracted.Then,a redundant feature point elimination algorithm is proposed to remove the outlier from other lane guardrails.Furthermore,a joint guardrail detection and tracking algorithm is developed to improve the accuracy,robustness,and efficiency of guardrail detection.The main contributions of this thesis are given as follows:1.The state-of-the-art algorithms and modeling methods for guardrail detection in intelligent vehicles are given,with a summary of their pros and cons.The problem of highway guardrail detection is summarized,and the motivation of this thesis is given.2.A distance-based redundant feature point elimination method for highway guardrail detection is developed and a metric for performance evaluation of highway guardrail detection is proposed.Firstly,the characteristic of highway guardrail based on lidar is analyzed in detail.The height and corner features of the highway guardrail are considered,and the corresponding feature extraction algorithm is constructed.To determine the outlier from other road’s guardrails,a robust guardrail detection method by distance-based redundant feature point elimination is proposed.Furthermore,a metric for performance evaluation of highway guardrail detection is given.Simulation experiments demonstrate the effectiveness of the proposed method.3.A joint guardrail detection and tracking method is proposed.In order to improve the accuracy,robustness,and efficiency of the guardrail detection,a joint guardrail detection and tracking method is proposed.In the proposed method,an adaptive interest region method is developed using the result of the tracking from the last frame.Considering the relationship between the guardrails on both sides of the lane,a joint highway guardrail detection and tracking is formulated.Computer simulations show that the proposed method has improved performance and a lower computation cost compared with the state-of-the-art algorithms.4.A software system of guardrail detection for a vehicle is developed.The functions and its corresponding interfaces for guardrail detection and tracking are established.Consequently,compared with other methods,the proposed methods have better performance in metrics of accuracy,robustness,and efficiency.This thesis provides future directions for academic and engineering.