Autonomous Vehicles (AV) are capable of impressive performance, but to be suitable for everyday use on the road they must master everyday traffic. These vehicles are configured with complex sensors, cameras, lidar systems and data recording devices. The sheer amount of data generated presents a problem of how to get it off the vehicle’s limited internal storage so these expensive vehicles can go back out to experience more driving situations for AI training. The data help engineers to create simulations that advance progress during the development phase because they are continually testing new situations around the clock, seven days a week. They generate hundreds of terabytes of data every day.
A major German autonomous vehicle development company presented us with their issue – moving 244TB of data off of five cars at a time within a 24-hour period. Then, five more cars would arrive the next day with the same huge amount of data, and on and on for several months. The testing would be done in a distant city. Not only did the data have to get off the vehicles, but the units also had to be shipped back to headquarters where they would upload the full dataset into the main computer center.
The solution was to have a set of ServerPack 35 devices in the field collecting the data, removing, and shipping them while they replaced the first units with a new set. The first group arrives at the main campus and uploads via 40GbE for the fastest transfer possible. Meanwhile, the second set is done recording and sent off. Then, a third set is put in place to provide continuous round-robin recording. The only way to get enough performance was to use Acromove´s VeloΣ, our data migration accelerator software.
With this model, petabytes can be transferred in the shortest timeframe and made available immediately for analysis. It is also very safe since not all the data is in one place until it is at the secure data center plus it is all protected by AES-256 encryption. Acromove takes security to another level with our AcroTrace™ tracking solution that shows where each device is located at all times, whether the ServerPack 35 devices have experienced hard shocks or extreme temperatures, and whether they have been diverted by a bad actor seeking to steal the data in which case the data can be destroyed remotely. The system can also be operated remotely so there is no need for an expensive computer technician on location to make things work.