Abstract
The Advanced Driver Assistance Systems (ADAS) has meant an improvement in the safety of road vehicle traffic. However, to achieve positive results it is necessary that these systems are able to judge in a coherent manner situations that entail risk and avoid giving false alarms that lead to incorrect and disturbing actions for the traffic that make the driver lose his confidence in the system. In this sense, the ACC systems can have operational problems, e.g. in areas of curves, which should be avoided by means of later filtering. Furthermore, in these ACC systems, besides the mere obstacle detection, it is important obtaining other variables not always considered, such as the relative speed, probability, localisation and impact angle, obstacle identification, etc. The developed system is based on the identification of obstacles using a long range lidar system Sick LRS 1000 located on the front of the vehicle. In turn, the detection of potential obstacles is filtered by means of algorithms that use the positioning of the vehicle in a detailed digital map of the road and sensors of the vehicle dynamics (mainly speed and yaw angle), considering the impact on the accuracy of the results of the positioning. Thus, it is possible to determine the region of interest in every moment, avoiding false alarms due to obstacles located out of the limits of the road or in adjacent lanes that entail no risk for the movement of the vehicle with the on-board system. Moreover, innovative algorithms have been included in the system for the identification of the type of vehicle detected by the lidar, as well as its kinematics, which are necessary data for assessing potential risk situations. This identification is based on the adjustment of the longitudinal and transverse axes of the detected vehicle, from the interpretation of the points detected by the lidar. This adjustment is done globally with all the points belonging to an obstacle, and unlike other presented methods it does not require the definition of tolerances, which improves the automatic process of obstacle description, making it more robust. Operational tests of the system have been made, comparing the inferred kinematics from the lidar measures and the actual kinematics registered by an inertial system (non-contact speed sensor L-CE Correvit and gyroscopic platform RMS FES 33) boarded in the detected vehicle, obtaining positive results.
KEYWORDS – Laserscanner, obstacle recognition, kinematic description, collision warning, ADAS