Along with energy efficiency, road safety is one of the key strategic innovation issues in vehicle systems development. As the absolute number of accidents decreases, the relative number of injuries and fatal incidents with people outside continuously increases. For years, the spotlight has been on passive passenger safety systems. Conventional systems are available for detecting a car’s surroundings using imaging; high hopes are also being held for “cooperative sensory systems”, which offer a variety of inherent benefits. Based at the Baden-Württemberg Cooperative State University Lörrach the Steinbeis Innovation Center for Embedded Design and Networking (sizedn) has joined forces with automotive manufacturers to develop the safety sensor solutions of the future.
For years, modern top-of-the-range cars have been equipped with sensor systems that provide drivers passengers with images to enhance safety and driving comfort. These include driver assistance systems such as night vision, lane keeping assistants (LKA), lane change assist (LCA), traffic sign recognition and much more. The systems are expensive, both because of the inbuilt cameras and the image processing – which becomes more and more complex, the more functions the system offers. Just one example: spotting the risk posed by pedestrians involves a whole host of issues affecting illumination and the car’s immediate surroundings.
To make these systems viable in as many cars as possible, costs need to be kept low. As things currently stand, this will only be possible if image-processing hardware can be found. The PROPEDES project (Predictive Pedestrian Protection at Night) is a plan to develop a predictive night-time visibility system for cars that will protect pedes trians by networking images provided by sensors around the car (video/radar). The project is sponsored by the German Federal Ministry of Education and spearheaded by e|enova, a car electronics innovation consortium. The Lörrach-based Steinbeis Innovation Center is a member of the PROPEDES project consortium working alongside Daimler, Robert Bosch and ProDesign. The aim of the project is to identify the basic algorithms needed to process, evaluate and size images and subsequently transfer this information to hardware and software. The team also plans to develop and evaluate flexible FPGA video processors. Other tasks include programming FPGA-based development platforms using soft cores or embedded microcontroller cores and integrating these into the development process.
Sensors that also provide images bring a number of advantages, especially as they enable image data to be used in various ways, and can recognize all kinds of objects. This contrasts, however, to a string of intrinsic disadvantages, such as difficulties when objects are partly hidden, weak signals caused by passive reflections, and high levels of interference under poor visibility. The biggest challenge, however, is how to categorize objects, i.e. how to put a recognized object into a meaningful group, or type of behavior (such as a pedestrian or cyclist).
One way to compensate for these disadvantages is to use cooperative sensor systems. Recognized objects receive a tag or active element. This makes it possible to exchange information via wireless systems so that at least the object’s category can be transmitted. By drawing on secondary radar techniques used in aerospace technology, it is also possible to evaluate the properties of electromagnetic waves to make another estimate of an object’s relative position. This enables the sensors to pinpoint the position and category of an object. A similar approach is already being used by WATCHOVER, an EU project in which the Steinbeis Innovation Center in Lörrach is also closely involved. This project already resulted in the first jointly-developed sensor, which subsequently became part of the sensor fusion (interconnection of all sensor data) in an on-board unit. This project and the Bavarian Project Amulett – which has already made significant progress with wireless positioning – have given rise to a consortium that is part of Ko-FAS (“Cooperative Sensor Systems and Cooperative Perception for Preventative Safety in Road Traffic”). Sponsored by the Federal Ministry of Economics and Technology (BMWi), the consortium has set its particular sights on cooperative sensor systems.
In a sub-project called Ko-TAG, the Steinbeis Innovation Center is working with BMW Research and Technology, Continental Safety Engineering International, Daimler, the Fraunhofer Institute for Integrated Circuits, and Technische Universität München. This project involves researching transponder-based cooperative sensor technology for two primary applications: protecting vulnerable road users and providing vehicle-to-vehicle safety. The joint Ko-TAG project uses transmitter/receiver modules in vehicles which send queries to transponders. These are carried by other road users and answer with response codes containing detailed information. The question- response method provides vehicles with information such as the relative position of other road users near the car – and this makes it possible to calculate the probability of a collision.