Unmanned transport development has made a huge leap forward over the past five years. However, although readings from onboard equipment have remained practically unchanged, the time for making decisions based on these measurements has become critical. More and more companies are joining the race to develop unmanned vehicles, and the success of each project largely depends on the quality of telematics systems and the speed of processing and analyzing the information coming from the multiple sensors. In making their forays into this promising market, dozens of companies have invested more than $10 bln in their projects. Artificial intelligence (AI) and 5G networks have radically changed the balance in the industry. To speed up the widespread introduction of autonomous vehicles, it might be a good idea to consider how data is processed in motorsport, an area where technical innovations are introduced without delay.
Data as fuel
In order for AI to learn, it needs to process a huge amount of data. In this sense, data can be viewed as fuel required for the operation of an AI engine. The data it needs is mainly supplied by telematics systems. Telematics enables controlling a vehicle as a single ‘body’ – in many ways similar to how the brain thinks. Due to their sensors and built-in navigation systems, unmanned vehicles can navigate the terrain, plan a route, react to obstacles and traffic signals, recognize people and other vehicles. With hundreds of different sensors, the system does this quickly, and makes decisions even faster.
For example, in Formula 1 racing, there may be just one thousandth of a second separating qualification from failure, which is only 6 cm between two competing race cars. Therefore, a racer also has only about a split second to make a decision.
A race car may be equipped with more than 200 sensors that will collect data equivalent to 300 GB over the course of a race. The entire scope of data generated by the company over a week will be equal to 45 TB. Artificial Intelligence and Machine Learning (ML) are so advanced that a colossal amount of data can be processed in an instant. Using AI and ML increases the accuracy of telematics processing. At the same time, modern data storage systems allow processing the data retrieved by sensors without compromising the testing and other procedures required for launching new vehicles in the market.
How it works
We cooperate with Mercedes-AMG Petronas Formula One Motorsport that uses our data storage systems in their work. The team collects data from over 250 sensors installed on their race car and analyzes the data in real time to score an advantage in the race. We process the data from the telematics systems and the race teams, pre-race and post-race results, simulation outcome, engine test data and all aerodynamic indicators. It is extremely important for the team to conduct test runs to simulate different conditions and use the obtained data to make decisions on how to act during actual races. They also use our solution to launch simulation process based on big data received during the last race in order to create a wining race car.
Changing gears with 5G
5G networks will make data processing even faster and, most importantly, will reduce the response time of telematics systems. One of the key changes that will be available with 5G is the possibility to continuously collect and process telematics data regardless of the location of the car. Teams will no longer have to return the car to the base to access the collected data and launch processing. In the city, the continuous data processing allows for improving the safety of self-driving cars. The technology of batch processing that could process data only after it was available for saving is a thing of the past. An increased capacity means that cars will spend more time on the roads, which will provide the prompt exchange of larger volumes of data. When processing big data, the unmanned car software will continue to learn and improve, both for the comfort and safety of city roads.
Driving the car of the future today
AI is imposing new requirements on companies; with ever-increasing volumes of data, IT infrastructure has to keep up with the times, easily adapt to changing conditions, and develop. At the moment, businesses that operate in the automobile industry are receiving an opportunity to make decisions based on the technologies that have been tested on race tracks.
- Vehicle insurance. Insurance companies can re-calculate insurance certificates depending on a specific motorist’s driving style. The insurer receives data from telematics systems, analyze them, and can update the tariff by raising it in case of an increased risk or reduce it for a highly careful driver. Also, insurance companies can provide motorists with an opportunity to control their driving behavior with the use of sensors, which allow drivers to see their mistakes and reduce accident risks.
- Safety. Telematics provides drivers with an opportunity to obtain some sort of extra sensory organs, with sensors and video cameras to promptly warn them of blind spots, pedestrians jaywalking, or a sudden obstacle.
To work efficiently, AI must receive data online and base their operations on predictive analytics developments based on machine learning. Data transfer rate is crucial, and it is important to exclude the possibility of losing data from numerous sensors and ensure efficient data processing. This is critical for the automobile sector, including for researchers to use data to focus on developing technologies that will allow us to see unmanned vehicles on the roads and have no concerns about core data storage infrastructure.
By Alexei Averin, Sr. Technology Consultant at Pure Storage