ADAS Model Vehicle – Developing and Testing Autonomous Driving Cost-Effectively

Testing driver assistance systems, fully automated driving functions and the associated software architecture on the original is often time-consuming and cost-intensive. To facilitate the work of developers and students and to make small changes immediately visible, a programmable model car was developed for the “Audi Autonomous Driving Cup” competition.

The result is a vehicle that corresponds to an Audi Q2 on a scale of 1:8. This car was equipped with the highest quality components from model sport, but also with computer components such as an Intel i7-8700t, 32 GB RAM, 512 GB SSD and a GeForce GTX 1050 Ti graphics card.

This driving computer runs the ADAS software framework ADTF, which was developed by digitalwerk. This makes it an ideal platform for developing and testing autonomous driving.

In order to develop driving functions, data on road traffic scenarios must first be collected. In the case of large automotive companies, this is done via the company’s vehicle fleets; students who are making their first attempts at programming driving functions at their university cannot, of course, fall back on such resources. But developers may also want to test new ideas on a small scale first, before continuing to work on a 1:1 scale. A model car that is remote-controlled, runs through various scenarios and is equipped with a Lidar and a Basler-Fisheye camera, is therefore a simple as well as inexpensive way to generate driving data.

Once the first data have been collected and corresponding driving programmes have been written, these can also be uploaded to the car and tested directly with it.

This makes it a perfect little helper to learn how to work with ADAS frameworks and also to implement smaller projects.