Low Speed Automation Using Multiple Sensors -- CARSENSE Underway in Europe
IVsource.net
31 July 2000

The project recognizes the early success and public acceptance of Adaptive Cruise Control (ACC) -- introduced in 1999 in Europe -- and sees this as the first in a string of increasingly more advanced functions.  As ADAS supports more complex driving tasks and gains wider use, CARSENSE will focus on extending the capability  to more complex situations in dense traffic environments in and around urban areas.  Traffic there is characterized by lower speeds, traffic jams, tight curves, traffic signs, crossings, and ancillary traffic participants such as motorbikes, bicycles or pedestrians. 

ADASE Conclusions Highlighted Needs

CARSENSE is an outgrowth of the Advanced Driver Assistance Systems for Europe (ADASE) project conducted during the EC's 4th Framework Program, which concluded that a crucial need for ADAS advancement is a significant boost in the performance of the systems that monitor the driving environment.  According to the CARSENSE project website, this calls for higher range and precision and higher reliability of the sensor information.  The way to accomplish this is to improve existing sensors like radar, laser, and image processing and -- importantly -- to fuse the information/output of these different sensor systems with appropriate scene models to achieve better accuracy, redundancy, robustness, and an increase of the information content.  Thus, CARSENSE aims to develop it all: a sensing system, sensor fusion techniques, and an appropriate flexible architecture for driver assistance systems to advance the state of the art in ADAS for complex traffic and driving situations ... at low speeds. 

The ADASE project concluded that driver assistance at low speeds is the most viable next step after introduction of ACC.  To do this requires two fundamental ADAS building blocks: reliable information on stationary objects; and a wider short-range field of view (as opposed to the long-range sensors used for high speed collision warning).  While automated low speed driving itself will not actually be implemented in CARSENSE, it will nevertheless drive the requirements for both the sensor system and the architecture. 

Key Innovations Built Into Project

Major areas of innovation are expected from CARSENSE in the areas of:

Introduction of a complex scenario for driver assistance systems, opening the way towards assisted/autonomous driving in dense urban traffic 

Innovative sensor system architecture, which enables easy exchange and upgrade of hardware and software modules (e.g., radar, laser, video) 

Fusion of individual sensors into a system with improved performance

Improvement of individual sensors using new object detection and image processing methods 

Creative teaming of complementary European companies 

• A real-time algorithm allows calculation of the orientation and the lateral pose of a vehicle with respect to the observed road. This approach provides robust measures when lane-markings are dash, partially missing, perturbed by shadows, highlights, other vehicles or noise.
• Contrary to usual approaches, the second technique is based on an efficient curve detector, which can automatically handle occlusion, by vehicles, signs, light spots or shadows, or low image contrast. Shapes in 2D images are described by their boundaries represented by linearly parameterized curves. No particular markings or road lighting conditions are assumed, such that the lane discrimination is based only on geometrical considerations. 

• Given the low speed application, the aim is to detect obstacles located at less than 50 meters (160 feet) in front of the test vehicle. For CARSENSE, an obstacle is any vehicle, motor bike, bicycle or pedestrian within the test vehicle trajectory.  The binocular vision and a multisensor approach allow successful detection and locaxtion of such objects.  This figure illustrates how data fusion exploits differences in sensor coverage, utilizes complimentary sensor information, and combines different sensor modes.

Test Vehicle Alfa 156 Sportwagon 2.0 Selespeed

• The partnership blends private sector product developers with research institutions. BMW, Fiat, and Renault are participating auto OEMs, with TRW Automotive responsible for image processing, fusion, and system aspects.  Thomson-CSF Detexis is handling the radar sensor, with Jena-Optronik providng video camera technology, and IBEO providing the laser sensor.

• The Institut National de Recherche en Informatique et Automatique (INRIA) of France, with its significant experience in image processing and sensor fusion, is developing the fusion algorithms, along with Ecole Nationale des Mines de Paris (ENSMP).

 
• The Laboratoire Central des Ponts et Chausses (LCPC), specializing in road construction technologies, is assessing the infrastructure aspects of the project, and the overall architecture is being defined by road safety experts at Institut National de Recherche sur les Transports et le Securitie (INRETS).  The LIVIC laboratory, a joint venture of LCPC and INRETS, is expected to play a key role as well.

IBEO Laserscanner LD Automotive

• Project coordination is being provided by Autocruise, a recently created joint venture of Thomson-CSF and TRW.  Autocruise is involved in the development, production and commercialization of radar sensors for advanced driver assistance systems.

... For more information contact Dr. Jochen Langheim,LucasVarity Thomson-CSF Autocruise Ltd., at jochen.langheim@detexis.thomson-csf.com ... or check out the project's detailed web site at www.lara.prd.fr/carsense/carsense.htm.

[Top]