Fortune 500 tech company: Developing L4 self-driving vehicle technology in a GDPR-compliant manner

Challenge

Object recognition using videos has been increasingly gaining momentum for many ADAS programs carried out by OEMs and Tier 1. Its development requires, among others, the collection of a vast volume of video data, taken from all manner of driving scenarios, in order to train the AI systems to accurately recognize pedestrians, traffic signs, other vehicles, and so on.

Our customer developed its own L4 self-driving vehicle program. To achieve this, they equipped vehicles with a total of 6 cameras (2 front-camera, 2 in each side mirror) and planned to proactively collect data for building proprietary AI systems.

While the collection of videos is essential to the development of their autonomous driving system, it might also raise questions to individuals' privacy leading to possible GDPR violations, as well as a PR backlash.

Therefore an anonymization solution was looked for addressing these concerns. The solution had to come with an offline mode in order to meet the IT security requirements of the customer.

Solution

Celantur’s Container was thoroughly tested by the customer during a pilot phase and with support by the Celantur team. Production deployment is located on local server infrastructure in a closed network environment. Additional interfaces (stream mode) have been added to the container, enabling input and output of images as low-level file structures.

Celantur's contribution