Blog

Cameras, radars, lidars and ultrasonic sensors continuously provide information that allows the vehicle to create a three-dimensional image of its surroundings. Systems need to recognize various objects in real time, including: vehicles, people, lanes and road signs. Sensor data must be digitally processed correctly so that the vehicle always receives the “absolute truth” – known in the industry as ground truth – to calculate and implement the driving function based on it.

The German company ZF has proposed a proprietary solution in this area – ZF Annotate. The cloud-based AI service is designed to accelerate the development and verification of advanced driving automation systems. The service works – regardless of the tested set of sensors – as a stand-alone configuration. The reference data is processed and annotated in the cloud, then a comparative measurement is performed.

Artificial intelligence for greater speed and greater security
Additional data recordings from ZF sensors – the reference measurement – provide ground truth. ZF Annotate operates as a redundant configuration, independent of the configuration of the perception system to be checked, and receives the same information while driving on the road. The recorded data is then sent to the cloud and analyzed. Thanks to artificial intelligence, all relevant objects are accurately marked and classified, assigned unique identification numbers, and moving objects are tracked. Object information is part of the full description of the environment model. After this “annotation”, the software provides a precise comparative measurement. This makes ZF Annotate a modern, AI-powered validation solution for testing and training modern ADAS/AD systems from Level 2+ to Level 5.

“ZF Annotate combines the advantages of a robust and independent set of reference sensors with a scalable cloud service that uses intelligent 2D and 3D tracking algorithms,” says Dr. Holger Klein, CEO of ZF Group. Previous comparable systems relied primarily on 2D annotations and thus mapped the environment in terms of distance and horizontal angle. ZF Annotate’s 3D-enabled feature adds depth information to the data, making the solution optimize a previously labor-intensive and expensive service.

Up to 10 times faster and 80% cheaper
ZF claims that the system is capable of processing data at a cost that is 80% lower compared to manual methods. Traditionally, validation of ADAS and AD systems required extensive manual annotation, which was time-consuming and expensive. Thanks to AI, the validation process is accelerated from 12 to 2 months. “With the ability to work 24 hours a day, seven days a week, our cloud-based service performs reference data verification in an extremely short time compared to the market, without any loss of quality,” says Klaus Hofmockel, Head of R&D at the Assistance Systems Division drivers and electronics.

Develop once, deploy everywhere: flexibility
Depending on the customer’s needs, the reference sensors can also be used in the test vehicle itself or in “Chase” mode where a separate reference vehicle is used (e.g. when evaluating public parking scenarios). ZF says this flexibility allows Annotate to become independent from specific sensor manufacturers and integrate it into ongoing development projects without major modifications to test vehicles.

The company ensures that the recorded reference data is not limited to the front view only. Depending on customer requirements, a set of reference sensors can provide a comprehensive 360-degree view, providing a detailed and precise representation of the vehicle’s surroundings.

Additionally, ZF Annotate can be used in all vehicle classes, both passenger cars and commercial vehicles.

The service was presented on June 27, 2024, during the ZF World Technology Day in Jeversen, Hannover.

Source: ZF – biuro prasowe World premiere: ZF Annotate uses AI for the development of ADAS and AD systems – ZF