Lane estimation using lateral histogram in radar based ACC system

Abstract

In Radar based ACC system, road structures makes multi-path in radar beam, so it degrades radar detection performance. To improve ACC performance, we propose a new method for recognizing road environment by radar only sensor based ACC system. The LES (Lane Estimation System) is statistical method that estimates lane characteristic by radar object (position, velocity) and in-vehicle information (vehicle speed, path-radius). The LES convert this simple data to statistical data, such as lateral histogram and lane cluster. Using this statistical data, we can estimate road environment and improve ACC performance. To prove validation of our method, we apply LES to real ACC system. In Yong-Seo highway (Korea), we assessed accuracy of our system and summarized improvement detection performance.