Lane Tracking Using Multilayer Laser Scanner To Enhance Vehicle Navigation And Safety Systems

It is of great interest in advanced driver assistance systems to prevent lane departure warning. Over half of all traffic fatalities in 2007 were caused by unintended lane departure [1]. Lane departure warning (LDW) systems often include passive sensors such as cameras to detect the lane and warn the driver of a lane departure. The most critical element of any lane keeping assistance system is the availability of reliable lane position information. Currently available LDW systems have shown that cameras can provide lane position information; however these systems suffer from technical limitations in areas where lane markings may be missing or difficult to detect due to lighting, rain, or snow [3][4]. However, LiDAR (Light Detection and Ranging) can be used to supplement other LDW sensors by providing lane position data accurately and consistently even in cases of varying outdoor conditions. It can provide additional robustness to a navigation solution by providing a lane position in the case of GPS outages.The objective of this research is to show that a multilayer LiDAR is capable of detecting and tracking lane markings that could supplement a LDW or on road navigation system. This will be realized using an Ibeo multilayer laser scanner that is capable measuring both distance and reflectivity data. The algorithm implemented is operating using the principle that the lane's surface is less reflective than the lane markings [6]. Therefore points of high reflectivity of the scan act as potential lane markings.Based on the distance to the lane markings, the user's position in the lane can be determined which will act as supplemental information to a navigation or safety system. Currently lane markings are capable of being found during post processing on both static and dynamic tests. Dynamic real-time data was taken at a NCAT test track aboard a Hyundai Sonata. LiDAR results of these tests were overlaid onto vision data, which were acquired simultaneously with the LiDAR data for a rough visual metric. Additionally the test track previously mentioned has been surveyed and will serve as a truth measurement in order to validate the methods used to acquire and track lane markings.This paper will cover in-depth the algorithm and processes used to acquire lane markings as well as techniques used to mitigate false readings.