MENU

ZF relies massively on AI for new ADAS platform

Technology News |
By Christoph Hammerschmidt

Equipped with artificial intelligence and a fairly comprehensive sensor set, cars equipped with this system can independently perform various driving maneuvers, especially on motorways. In addition, the ZF coPILOT can be operated with voice commands and recognizes the traffic situation, senses driving behavior, and also monitors the driver in order to prevent dangerous situations through active control interventions. The core of the system is the ZF ProAI central computer and the Nvidia DRIVE platform recently introduced by ZF. The system presented in Shanghai is close to series maturity; it should be available from 2021.

The System is based on a sensor set that monitors the car n all directions. Equipped with Nvidia’s DRIVE software, the system’s driving and safety functions are expected to go well beyond the performance of regular Level 2 systems for semi-automated driving, explained Torsten Gollewski, Head of Autonomous Mobility Solutions at ZF.

The company highlighted the range of functions with a demo car on the AutoShanghai. For example, vehicles equipped with such a system can automatically drive on or off motorways or similarl multi-lane highways. For more relaxed highway driving, the ZF coPILOT combines extended cruise control with active steering support and lane keeping function. In addition, the system can proactively change lanes, overtake other vehicles, and return to the original lane. In addition, the system permanently analyzes the vehicle environment and identifies pedestrians, oncoming traffic, and intersections.


In addition, the ZF coPILOT is equipped with sensors and functions that monitor the driver and issue warnings should safety risks arise. This is the case, for example, when the driver is distracted, his line of vision is too far away from the traffic, or he exhibits signs of fatigue.

The ZF sensor set in the demonstration vehicle consists of a front radar, four corner radars, and a total of eight cameras. Two of them are directed to the front and one to the rear, two cameras each are integrated in the side mirrors on the left and right, and one monitors the driver.

In order to be able to evaluate the flood of images and data from these sensors in real time, enormous computing power is required. This is the task of the central computer ZF ProAI Gen2, the heart of the ZF coPILOT: Equipped with a powerful and Nvidia DRIVE Xavier chip, ProAI can integrate and holistically control the pre-trained algorithms for all driver assistance functions. According to the manufacturer, the system is scalable and seamlessly updatable – if desired, automobile manufacturers can also add their own features and functions or those developed by ZF.

Since in many countries highly automated driving according to SAE Level 3 is not yet permitted, the ZF coPILOT is based on partially automated driving according to Level 2, which obliges the driver to permanently monitor the traffic environment, but is intended to relieve him of many routine tasks and increase safety.

 

Related articles:

ZF demos automotive supercomputer, autonomous minibus

Artificial Intelligence arrives at ECU level

Infineon, TTTech Auto show computing platform for autonomous driving up to Level5


Share:

Linked Articles
eeNews Automotive
10s