NVIDIA has been tabbed to provide the processors and AI platform for the self-driving taxi and ride-sharing tests.
By Murray Slovick, Contributing Editor
In an effort to speed up development of fully automated, driverless operation (SAE Level 4/5) in metropolitan areas, Bosch and Daimler chose California as the location of their first autonomous test fleet. In the second half of 2019, the partners will offer customers a shuttle service with automated vehicles on selected routes in an unnamed Californian metropolis (though they did later say that an automated shuttle service would be available on certain routes in a city located in Silicon Valley).
The partners are equally financing the development work. Daimler Mobility Services is envisaged as the operator of the test fleet and the app-based mobility service. The pilot project aims to demonstrate how services such as car-sharing and ride-hailing can be intelligently connected to shape the future of mobility. In addition, the partners have decided on NVIDIA as the supplier of the artificial-intelligence (AI) platform as part of their control unit network.
A decisive factor for fully automated and driverless driving in an urban environment is the reliable recognition of the vehicle’s surroundings with the aid of various sensors. Analyzing and interpreting the variety of incoming data and translating them into driving commands within a very short time requires enormous computing power.
At the same time, fully automated, self-driving vehicles in the city require a versatile, redundant system architecture and the highest level of functional safety. To do so, the necessary computing operations are performed in parallel in different circuits, with the system capable of getting instant access to these parallel computing results when necessary.
The control unit network will be used in the fleet vehicles that Daimler and Bosch will put on the roads of California in the second half of 2019. The test operation will provide information about how fully automated and driverless vehicles can be integrated into a multi-modal transport network.
Currently, Bosch and Daimler employees work together in teams in two regions—the greater Stuttgart area in Germany and around Sunnyvale in Silicon Valley. Employees from both companies share the same office space, facilitating rapid communication across working disciplines and short decision-making paths. At the same time, they have access to the entire know-how of colleagues working in the parent companies.
The personnel in this collaborative effort are developing the concepts and algorithms for the fully automated system. Daimler’s task is to bring the drive system into the car. To that end, the company is providing the necessary development vehicles, test facilities, and, later, the vehicles for the test fleet. Bosch is responsible for the components (sensors, actuators, and control units) specified during the development work. For test purposes, the partners will use their laboratories and test rigs, plus their respective test sites in Immendingen and Boxberg, Germany.
What’s more, in 2014, Mercedes-Benz obtained approval to test automated vehicles in the Sunnyvale/California region. Since 2016, the company also has comparable approval within the Sindelfingen/Böblingen region in Germany. Bosch says it was the world’s first automotive supplier to test automated driving on public roads in Germany and in the U.S. in early 2013.
For their driving system, Bosch and Daimler will rely on a control unit network made up of several individual control units. NVIDIA will supply the platform required to run the AI algorithms generated by Bosch and Daimler for the vehicle’s movement. The network of control units collates the data from all sensors with radar, video, LiDAR, and ultrasound technology (sensor data fusion), evaluates them within milliseconds, and plans the movements of the vehicle. All in all, the control unit network has a computing capacity said to be in the hundreds of trillion operations per second.
Both partners are focusing on safety as a chief concern. “The decisive factor is to introduce a safe, dependable, and mature system,” says Dr Michael Hafner, Head of Automated Driving at Daimler AG. “Safety has the highest priority, and is the constant theme of all aspects and development stages on our way to the start of series production. If in doubt, thoroughness comes before speed.”
Stephan Hönle, Senior Vice President Business Unit Automated Driving at Robert Bosch GmbH echoes Dr. Hafner’s sentiments, “Developing automated driving to a level ready for series production is like a decathlon. It’s not enough to be good in one or two areas. Like us, you have to master all disciplines. Only then will we succeed in bringing automated driving to the roads and the city safely.”