In 2021, the BMW Vision iNeXT will be the first BMW Group vehicle to be able to autonomously roar over the motorway. In technical jargon this means: the vision iNeXT dominates Autonomous Driving Level 3. Level 3 means that the driver can pass on the highway up to a speed of 130 km / h, the driving over a longer period of time to the vehicle. Can, does not have to – as the management at BMW emphasizes.
80 cars collect data worldwide
But for this vision to become reality in two years, there is still much to do. BMW is currently sending 80 cars a day from its Autonomous Driving Campus in Munich / Unterschleissheim, as well as at test sites on the west coast of the USA, in Israel and China, to record car traffic data. By the end of 2019 it should be 140 vehicles.
The test cars should collect around five million kilometers of real driving data. From this data, which is recorded with the help of cameras at the ends of the front and rear bumper, about two million data sets are filtered out which show particularly relevant traffic situations.
All data is stored as raw data as well as analyzed in the respective driving situation in order to derive corresponding instructions for autonomous driving. At the end of 2021, the next step will follow: A fleet of test vehicles will test the Level 4 function in city traffic – that means driving without any driver intervention. It takes again records and many kilometers driven – this time from the urban environment.
240 million simulated road kilometers
Why so much effort with driving around? “Because only the pictures of these test drives show what is really going on outside in the field,” says Alexandro Vukotich, Head of Development Driver Assistance and Autonomous Driving at BMW – and until recently still in development at Audi. But that’s not enough as a basis for developing autonomous driving. To ensure a high level of security, the two million miles actually filtered out of the system will be covered by another 240 million kilometers of simulation-generated kilometers.
AAlso interesting: Start-up teaches autonomous cars to see
The idea behind it: The complexity and variety of traffic situations on all continents can be mapped with enormous amounts of data, making autonomous driving ultimately manageable. In order to save and process all this data, BMW has created the D3 platform. D3 stands for data-driven development and has powerful power: 230 petabytes of storage capacity as well as an extremely powerful compute platform with more than 100,000 processor cores and over 200 GPUs (graphics processing units). About the power consumption, however, BMW is silent. But it should not be environmentally friendly and sustainable.
“Stay in the lane, turn left, turn right – that’s not a problem with autonomous driving, but then there is this one roundabout in France, which is somehow different from all the roundabouts scanned so far really is, “says Klaus Straub, CIO at BMW.
The engineers and developers are progressing piecemeal – from the simple left-right-straight-ahead driving situation to potential accident scenarios. And every time a new software version is created. Once the developers have programmed a new camera software version, neural networks and re-processing are used to test whether the new version is better than the old one. The system is thus constantly evolving.
New business models
So there is a lot of research and development in the new technology. And these investments will have to pay off someday. In terms of price, a BMW, which has integrated autonomous driving Level 3, in the price range of a normal BMW with optional equipment are: four-digit, comparable to what would cost a luxurious leather interior. The situation is different with autonomous driving levels 4 and 5.
“Since we do not yet know how many sensors we will need in the first generation of these vehicles, we still can not estimate how much euros an autonomous car will cost, but Level 5 is unlikely to be readily affordable to the consumer We have to think about new business models and think about whether the classic purchase model is still up to date, or more likely to be leasing or sharing models. ”
New work structures
And BMW is breaking new ground not only with regard to price models. In order to advance the topic of autonomous driving, BMW has also missed completely new work structures. “Because with the previous methods, we can not continue,” says Alexandro Vukotich. “We do not know the size of the mountain of tasks that come with autonomous driving, so we slice the mountain, and we’ve set up 80 teams, each working on autonomous driving solutions for two weeks will be tested and the next steps will be taken. “The software developers can test their newly written code on the campus immediately in the vehicle.
So you work your way step by step, relying on agile working methods. Hierarchy levels have largely served their purpose. Every single person has a lot of responsibility in the projects. But partners can also be well integrated via the agile structures and the open ecosystem. BMW is working with Daimler to create interfaces for the integration of additional services.
Whether this will be enough to depend on the competition from China, which is already testing Autonomous vehicles through the area today, must show the future. The sluggish development of the 5G expansion in Germany at least brakes the autonomous driving properly.
Also interesting: BMW boss: “We are in a technology war”
Newsletter & Messenger
Always up to date on all topics of digital life with the LEAD Newsletter and the LEAD Tech Newsletter. Whether professional or private. In your inbox or via messenger.
Subscribe to our newsletter now
Subscribe now via messenger