The license to test

As one of the first companies in Austria, the Graz Research Center VIRTUAL VEHICLE received the official approval to carry out test drives with autonomous vehicles on public roads. The approval was granted by the Federal Ministry of Transport, Innovation and Technology.

Just at the beginning of May, Europe’s most diverse test lab for autonomous driving was launched. Under the title ‘ALP.Lab’, the VIRTUAL VEHICLE, together with the Styrian research institutes TU Graz and Joanneum Research, as well as the industrial companies AVL List and Magna Steyr, bundles its competences, in order to test automated driving systems on a large scale. ALP.Lab combines virtual and real tests, analyses, simulations and a wide range of private and public test tracks.

To fully utilise ALP.Lab’s test environment, however, a certificate from the Ministry of Transport is necessary, which officially allows to carry out test drives on public roads. The VIRTUAL VEHICLE has now received this certificate as one of the first companies in Austria.

Deep learning with the vehicle demonstrator

In fact, the certificate only applies to VIRTUAL VEHICLE’s demonstrator car in form of a Ford Mondeo Hybrid. This car is equipped with the latest ‘steer-by-wire’ and ‘brake-by-wire’ systems as well as simple driver assistance functions. Based on this vehicle, VIRTUAL VEHICLE is working on a unique research platform, which is developed in three stages: in the first step, which has already been largely completed, the researchers ensured that electronics are given full access to the vehicle, with the goal of operating the accelerator, brakes and steering by computer alone, creating an “artificial intelligence on wheels”.

Currently, the Graz researchers are in the second development stage: latest sensors (for example, Lidar, radar, cameras, GPS or Car-2-X-systems) are installed to enable a 360° field survey. At this stage, high-performance multicore computing platforms (such as NVIDIA, Infineon Aurix, and dSPACE) will also be integrated for data analysis and data fusing to prepare the third phase as best as possible.

In the final third stage, autonomous driving functions will be finally implemented. In this context, ‘deep learning’ methods are tested, which means that the car is ‘trained’ until it can follow a road independently.