Computer data for autonomous driving
29.03.2019
K ameras, sensors and lidar are the sensory organs of self-driving cars. The software can process this information and derive actions from it, for example recognize a red traffic light and brake in front of it.
The purely logical processing of such data, however, does not allow an autonomous vehicle to move reliably in traffic. Even at level 3, for example longer automated driving on motorways, unpredictable sources of danger must be recognized and responded accordingly. Is the car moving in a serpentine manner in the other lane? Does the truck now switch to the left without blinking?
Storage space such as 230,000 notebooks
The swarm brain of networked and self-driving cars has to learn possible scenarios. This generates huge amounts of data. BMW feeds the information from around five million kilometers of real driving data into the new High Performance D3 platform with a test car fleet. Their storage capacity is 230 petabytes. One petabyte is equal to 1,000 terabytes. A terabyte is a hard drive size currently used in notebooks.
BMW calculates it clearly: The data volume of 230 petabytes corresponds to the company after a volume of 45 apartments of 80 square meters and a generous three meters high, all completely are full of CDs. A block of flats full of data.
Every day 1,500 terabytes of new raw data are collected for processing, which could be saved on 23,000 iPhone X.
Level 3 from 2021, level 4 later
In 2021, BMW will launch a new electric SUV based on the iNext study, the first series model that will have Level 3 autonomous driving systems on board for a surcharge. Up to a speed of 130 km /h, the computer should then be able to take control on the autobahn for a longer period of time.
At the end of 2021, BMW wants to put a fleet of test vehicles on the road, with which the next one Automation level, level 4, is tested. Here the car drives without any action from the driver, who can and will only have to intervene in certain emergency situations.
