As a measure to guarantee failsafe performance, it is common practice for aircraft to install all safety-relevant systems twice. The project partners in the KIT SmartLoad project however want to find ways to detect failures to cars at an early stage and thus control imminent dangers. The project therefore is focusing on the development of new methods for fault prevention and prediction. "In the development of automated vehicles, so far existing series models have been equipped with additional technology so that the cars ultimately have more components. Our approach is to develop vehicles that do without the ballast of additional components," explains Michael Frey of the Institute for Vehicle System Technology (FAST) at KIT.
For safe and reliable operation, it is necessary to consider the entire vehicle with all its mechanical and electronic components and their interaction with the driver, explains Frey. The scientists in the SmartLoad project are testing this approach using the example of steering assistance. They use a drive that individually controls individual wheels: while a normal power steering system consists of a motor that helps the driver to turn the steering wheel, the wheels are now controlled differently on the left and right, which makes steering directly easier. This makes it possible to compensate for a power steering failure without having to install additional components,.
In addition, electric and self-driving vehicles, in which all four wheels are individually driven and steered, can perform completely new driving maneuvers. Existing standard tests based on driving cycles are not suitable for testing such cars, according to the scientists' analysis. The solution is to provide test benches that test individual components but pretend that they are installed in a vehicle that is currently undergoing a test drive. The partners in the project have test benches at their disposal that are connected in the XiL-BW-e laboratory network for electromobility and map all aspects relevant to vehicle development in real time.