The CARSENSE program, grouping together 12 major industrial and research partners (Autocruise, BMW, CRF, Renault, IBEO, Jena-Optronik, Thales Airborne Systems, TRW Automotive, INRIA, INRETS-LEOST, ENSMP, LCPC-LIVIC) and sponsored by the EC, is willing to develop a sensor system, the purpose of which is to provide sufficient information on the car environment to assist in low speed driving in complex (urban) environments. This article describes the main results so far of the program regarding improvement of individual sensors and merger of the information from these sensors in a fusion unit. Some innovative aspects have been already integrated in the system. After the introduction of Adaptive Cruise Control (ACC) into the market in 1999 all surveys and experimental assessments have shown a high interest and product acceptance for such systems. However, these first "advanced driver assistance systems" (ADAS) are very much limited to use on motorways or urban expressways without crossings. The traffic situations evaluated by these systems consist only of other vehicles, such as cars and trucks moving in simple patterns. Processing is restricted and can be focused on few, well defined detected objects. Nevertheless, even for these relatively simple situations, these first systems cannot cope reliably with fixed obstacles. They also occasionally behave in an unexpected manner, causing surprise to the driver in 'cut-in' situations. Here, the width of the sensor beam may not fully cover the area in front of the vehicle, resulting in a late response from the ADAS.
Abstract