The ProAI ‘supercomputer’ unveiled by ZF and Nvidia a year ago acts as the central control unit for a new test vehicle that ZF says demonstrates Level 4 automation.
The test vehicle can perform various driving functions, and has been developed with a special focus on urban environments. “In contrast to a trip on a freeway or rural road, it is significantly more complex in urban scenarios to create a reliable understanding of the current traffic situation, which provides the basis for appropriate actions of a computer-controlled vehicle,” said Torsten Gollewski, head of advanced engineering at ZF Friedrichshafen.
ProAI’s hardware, connected sensor sets, evaluation software and functional modules can be adapted, giving scalability. “The configuration’s modules can be adapted to the specific application according to ZF’s ‘see-think-act’ approach helping vehicles to have the necessary visual and thinking skills for urban traffic,” said Gollewski.
“The flexible architecture also allows for other automation levels in a wide variety of vehicles. At the same time, it provides information about which minimum hardware configuration is essential for which level.”
The test vehicle gains a 360° understanding of its surroundings, updated every 40 milliseconds, using cameras, lidar and radar sensors. The ProAI’s computing unit analyses this data in real time. “AI and deep-learning algorithms are used primarily to accelerate the analysis and to make the recognition more precise,” said Gollewski. “It’s about recognising recurring patterns in traffic situations from the flood of data, such as a pedestrian trying to cross the road.”
The vehicle’s possible reactions are used to calculate the longitudinal acceleration or deceleration and direction of travel and this data is saved.
ProAI is production-ready. ZF has announced that a Chinese car manufacturer will be the first to install ProAI in a vehicle with autonomous features the result of collaboration with Nvidia and Baidu.
ZF and Baidu recently presented a system based on ProAI, enabling autonomous valet parking. The function is being debuted in a test fleet belonging to Pand Auto, a Chinese car-sharing provider.
“ZF’s supercomputer is ideal for our autonomous driving applications, as it can process a huge amount of data such as HD digital maps in real time,” said Zhenyu Li, vice president and general manager of Baidu’s Intelligent Driving Group. “In addition, it combines with AI, deep-learning capabilities and meets all the current automotive standards.”