Heterogeneous Fusion of Video, LIDAR and ESP Data for Automotive ACC Vehicle Tracking

This contribution introduces an approach to cross-calibrate automotive vision and ranging sensors. The resulting sensor alignment allows the incorporation of multiple sensor data into a common detection framework. Exemplarily, we show how a realtime vehicle detection system, intended for emergency breaking or ACC applications, benefits from the low level fusion of multibeam lidar and vision sensor measurements in discrimination performance and computational complexity. The following tracking stage estimates the position and velocities of targets to permit distance control, time to collision measurements and further situation analysis

I. Introduction

Several driver assistance systems observing the vehicle environment have already been introduced to the market. The first generation of such systems, introduced for Active Cruise Control (ACC) applications, were realized with long range radar or lidar technology. As a classification feature to distinguish between real targets of interest and background measurements, these systems exclusively used the measured and tracked target velocities over ground. Needless to say, that such a scheme recognizes only moving targets and advanced applications as emergency breaking and ACC stop-and-go rely on a robust non-moving object detection, especially in traffic jam situations for instance. In addition, the same is the case for safety systems related to vulnerable road users (VRU), who are often just standing in urban areas. This is why the currently available second generation of ACC systems, solves the problem by taking additional sensor data features into account for a non moving object detection. But due to legal reasons and a remaining uncertainty the fully automatic breaking action is released by the driver after a warning signal.