BMW Builds its Own IT Platform to Process Data From its Autonomous Vehicles
【Summary】Self-driving vehicles generate terabytes of data, which needs to be analyzed during the development process, so that deep-learning algorithms can make safer driving decisions and improve over time. The challenge for automakers is how to sort and analyze all of this data so it can be used more efficiently for self-driving vehicles. To address this problem, BMW is built its own IT platform it calls D3.
A self-driving vehicle needs more than additional hardware, these vehicles require access to vast amounts of quality data, along with millions of lines of code to safely navigate. Self-driving vehicles also generate terabytes of their own data, which needs to be analyzed during the development process, so that deep-learning algorithms can make safer driving decisions and improve over time.
The challenge for automakers is how to sort and analyze all of this data so it can be used more efficiently and shared with other self-driving vehicles.
To address this problem, BMW is built its own IT platform it calls D3. The "D3" name stands for "Data-Driven Development." D3 forms the basis for the development and validation of data for BMW's autonomous vehicles. The launch of the new BMW's D3 platform represents a key milestone on BMW's path to highly and fully automated driving.
BMW says its Data-Driven Development is an indispensable tool in securing the safety and reliability of the Level 3 autonomous systems that will be offered in the future, beginning with the upcoming BMW iNEXT in late 2021.
The BMW Group has been applying the Data-Driven Development (D3) approach for several years now. The basic principle assumes that the only way of successfully navigating complex traffic situations in the real world with a autonomous vehicle is to gather massive quantities of data for analysis. This means the algorithms and overall operation of autonomous driving have to be validated using a broad set of driving data.
How the D3 Platform Works
The quality of the data collected from BMW's development vehicles is continuously improved by utilizing a robust data qualification and filtering process. The first step in the process for BMW is to collect around 5 million kilometres (3.1 million miles) of real-life driving data from its fleet of test vehicles.
From all of this data, 2 million kilometers (1.25 million miles) of the most relevant driving data and environmental factors are then identified and extracted. This driving data subsequently undergoes regular reprocessing as autonomous driving development progresses.
The two million kilometers of data is constantly expanded by adding 240 million kilometres (150 million miles) of simulation-generated data, to ensure that real-life driving scenarios are taken into account during development.
Reprocessing 1.25 million miles of real world data combined with 1.25 million more miles of simulation requires a high performance data platform with over 230 petabytes storage capacity and the computing power of more than 100,000 cores and more than 200 GPUs.
To transfer all of this data for analysis, BMW uses a 96 x 100 Gbps (gigabyte per second) connection between the BMW Group High Performance D3 platform and the Hardware-in-the-Loop (HiL) simulation stations located at the BMW Group Autonomous Driving Campus.
The data collection fleet is comprised of 80 BMW 7-Series sedans, which are in operation on the west coast of the U.S., Germany, Israel and China. The number of vehicles is set to increase to 140 by the end of 2019.
In 2021 the production version of the BMW Vision iNEXT, which was first unveiled to the public in summer 2018, will become the first model from the BMW Group to offer a Level-3 system as an option. To support the safe autonomous driving functions in the INEXT, BMW will use data from its D3 platform.
BMW's fleet of test vehicles will begin testing higher Level-4 autonomy in 2021, which requires no driver intervention, in large-scale trials conducted in predefined urban environments.
The BMW iNEXT will rely on the automaker's D3 data center for its autonomous driving systems.
Computing Capabilities of the BMW D3 Platform
The D3 platform supports the daily collection of more than 1,500 terabytes (TB) of raw data each day generated by BMW's autonomous test fleet. Storage capacity is more than 230 Petrabytes. Computing power is provided by 100,000 cores and more than 200 GPUs.
Bandwidth to transfer this large amount of data is roughly enough to broadcast one million HD television programs simultaneously.
The data center is just a few kilometers away from the BMW Group Autonomous Driving Campus in Unterschleißheim, Germany near Munich. The close proximity to the campus is needed in order to transfer the enormous quantities of data from the campus to the platform via physical cables.
The enormous challenges involved in building a secure platform for fully automated driving were met by BMW teaming up with the leading technology partners, including DXC Technology, an IT services company. DXC was formed in 2017 by the merger of Computer Sciences Corp ( CSC) and the Enterprise Services business of Hewlett Packard.
DXC's work is to setup and run the data center and to develop applications to support the autonomous driving development process. The aim is to reduce costs and the time needed until the system is ready to market.
DXC's applications allow BMW's development teams to collect, store and manage the data from the vehicle sensors. It also makes it available for training machine learning algorithms for autonomous driving in a matter of seconds.
BMW says that using a single platform for data storage, processing and AI training lowers the hardware and software requirements, thereby reducing costs and complexity. The automaker says that data can be gathered globally but monitored centrally. This can help maximize efficiency and cut costs.
The DXC solution was developed in an open source environment, allowing agile collaboration between engineers, regardless of their location. The BMW Group is the first company in the automotive industry to apply agile working models entirely focused on autonomous driving.
The BMW Group said the high performance D3 platform was completed in the space of a few months and within budget.
The D3 platform will support work at the automaker's Autonomous Driving Campus, which opened in 2017. The campus has space for 1,800 employees. BMW opened the campus to pool together its development expertise in advanced driver assist systems (ADAS) and autonomous driving at a single location.
"We want to play a leading role in the development of safe autonomous driving," said Klaus Fröhlich, Member of the Board of Management of BMW AG, responsible for Development, at the time. "We are pursuing this goal with great diligence and systematically establishing the necessary framework along the way. One of the milestones is our Autonomous Driving Campus."
BMWs agile development process, combined access to high quality data from the D3 center, speeds up the automaker's entire software development process for autonomous driving, allowing BMW's future autonomous vehicles to handle extremely complex driving scenarios.
Originally hailing from New Jersey, Eric is a automotive & technology reporter covering the high-tech industry here in Silicon Valley. He has over 15 years of automotive experience and a bachelors degree in computer science. These skills, combined with technical writing and news reporting, allows him to fully understand and identify new and innovative technologies in the auto industry and beyond. He has worked at Uber on self-driving cars and as a technical writer, helping people to understand and work with technology.
Electric Automaker XPeng and Bosch Join the ‘ELISA Project’ to Promote the Use of Linux in Safety-Critical Automotive Applications
China's Tech Giant Baidu Plans to Rollout the World’s Largest Fully Autonomous Ride-Hailing Service by Next Year
Vietnam-based Electric Vehicle Startup VinFast Ships its First Vehicles to the U.S.
Tesla’s Model 3 is Reportedly Getting a Redesign to Make it More Appealing as Competition in the EV Segment Grows
LG Chem is Investing $3.2 Billion to Build a Cathode Manufacturing Facility for Electric Vehicle Batteries in Tennessee
Mercedes-Benz is Offering a Performance Upgrade as a Yearly Subscription on its EQ Electric Vehicles
LG Develops ‘Invisible’ Speaker Sound Technology That Could Revolutionize In-Vehicle Audio
Waymo Reveals its New Robotaxi Built by New Premium Electric Vehicle Brand Zeekr
- Elon Musk Wants to Cut Around 10% of Tesla’s Global Workforce, Blames Economic Uncertainty
- Shares of Volvo’s EV Brand Polestar to Begin Trading on the NASDAQ After Closing on its SPAC Deal With Gores Guggenheim
- Automotive Supplier MAHLE Developed a Superior Continuous Torque (SCT) Electric Vehicle Motor That Can Run ‘Indefinitely’
- Volkswagen Announces Pricing and Styling Updates for the 2023 ID.4 Electric SUV, Which is Being Assembled in Tennessee
- J.D. Power Study Finds That EVs Are More Problematic Than ICE Cars
- Qualcomm Ventures Invests in ThunderX, a Company Developing a Domain Controller-based Intelligent Vehicle Compute Platform
- Michigan-based May Mobility Closes on $111 Million Funding Round, Begins Development on Toyota’s Next-Gen Commercial Autonomous Vehicle Platform
- General Motors Announces a Key North American Nickel Supply Deal for Electric Vehicle Batteries
- BMW Expands i4 Lineup With the Affordable eDrive35
- Tesla May Build its Next Factory in South Korea, According to the Country’s Presidential Office