With their new high-resolution cameras, the fleet vehicles constantly record situations in traffic. To evaluate this data, systems with artificial intelligence run on the new high-performance computers. They automatically recognize and filter situations in the vehicle that are particularly valuable for the developers. These rarely occurring corner cases are crucial for the development of highly automated driving functions.
Once detected in the vehicle, the valuable camera recordings are uploaded to the cloud and are available to the development teams almost in real time. The teams can train their software with this new data and deploy a new release to a separate, secure development partition of the vehicle computer via over-the-air update as part of the so-called Big Loop. There, the improved systems can be tested under real conditions without compromising the safety of the vehicle.
With the development fleet, functions of the new software platform can be tested at an early stage and successively prepared for series production. This process underlines CARIAD's software-driven and agile product development approach.