"Every Bit is Gold" - This is the motto of many OEMs and Tier1s developing sensor-based systems in the field of driver assistance and autonomous driving. They collect as much data as possible from real test drives for validation. However, developers quickly face numerous problems with this approach.

Storing the data in the vehicle, as well as transporting it over the network, storing it in the data center and finally processing it all involves very high costs. At the same time, no information is obtained about the relevance of the collected data. It is therefore not unlikely that some recordings or scenes are duplicated or simply not usable because of to poor data quality. This comes into play when long test drives are affected due to limited storage capacity, incorrect or corrupt data in the vehicle. 

It quickly becomes clear that this approach to data harvesting is proving to be extremely inefficient and costly. Especially when you regard advancing technology and accompanying ever-increasing amounts of data. We, at b-plus, have developed a solution to this problem through our Smart Recording platform:

Behind the term Smart Recording is the combination of the AVETO Toolbox, which enables the pre-selection of interesting and relevant driving scenes at an earlier stage of the process. The recognition and tagging of raw data is already integrated into the recording process. The data stream can be filtered according to different scenarios: for example, specific objects on the road, the number of pedestrians, or even specific scenarios such as overtaking events. With Smart Recording, not all driving data is recorded and fed in, but the recorder is only triggered for relevant scenes.

Our Smart Recording solution enables monitoring of all recordings with regard to content and quality. The data is already prioritized during recording. This way, the amount of data can be reduced to 5%. This corresponds to a saving of 60 terabytes during an 8-hour test drive.