Generally, data collection in the automotive industry is not a new phenomenon, as there has always been a need for vehicles to have improved safety features and accurate mapping and localization capabilities. However, the leap toward connected and autonomous vehicles demonstrated that the amount of data that is being collected is increasing every year.
What type of data is collected?
As we described in our previous blog, much like global privacy measures, the EU’s GDPR law highlights rigorous requirements that companies must abide by to be compliant and ensure privacy for the public. For the automotive industry, this means paying special attention to the private information that is collected by smart applications and technologies in modern vehicles.
A study concluded that an average American vehicle could produce up to 380TB to 5100TB in just one year – if placed on a mass scale the amount of data produced is staggeringly high – so what type of data is the automotive industry collecting?
ADAS and ADS
To make the Advanced Drivers Assistant Systems (ADAS) and Autonomous Driving Systems (ADS) functional, a large amount of data is collected for algorithm development and training for perception, localization, path planning, and testing and validation of these systems. The data is usually classified into three categories:
- Owner and passenger information: like identification information and personal setting.
- Location data such as vehicle location from GPS, IMU, and sensors.
- Sensor data: Front, side, rear cameras, fisheye cameras, dash camera, inner-cabin camera, lidar, radar, ultrasonic sensors, etc.
To serve ADAS and ADS applications, the camera plays a vital role to perceive the 360-degree surroundings of the vehicles by installing over 6-10 cameras from the front, the side, and the rear. They range from different resolutions, the field of view, detection distance, and mounting position. The images and videos captured by cameras provide both static information like the position and semantic meaning of lane and traffic signs etc., and dynamic information such as the type, location, speed, and movement prediction of vehicles, pedestrians, and other VRUs. The data is then fused with other sensors’ output to interpret environmental conditions in a robust, high quality, accurate, and safe manner.
High-definition map (HD map) plays an important role in modern connected and autonomous driving on different aspects, for example, to precisely localize themselves and to accurately navigate on a lane-level basis. It contains information about the exact positions of pedestrian crossings, traffic lights/signs, barriers and more. To produce the inch-perfect accurate HD maps, large amounts of street-level data need to be collected and processed via the fleets with various sensors installed. This is often done using collection cars which use various sensors, such as LiDAR, GPS, IMU, antenna, and cameras to collect point cloud and image data. Through the various sensors that are used, collection cars pick up street-level information that includes the road, its surroundings, people, and of course other vehicles.
Generally, dashcam footage is mainly used by consumers, for reasons that range from personal entertainment to providing evidence of accidents and events on the road. For the automotive industry, dashcam footage can be used against the ADAS and AD sensors, as the footage is low resolution, the device is light, easy to install, and cost-effective compared to advanced methods of data collection, such as preinstalled sensors for mass production. This makes it a prime tool used by the automotive industry, particularly for the companies that deal with AI and data-driven technology and for vehicle validation and verification.
The flexibility of the dashcam makes it a preferred tool for AI-based autonomous driving startups, where the dashcam collects long-hour driving scenarios from various countries, road conditions, and driving scenes, which then facilitates the machine learning process for the AI technology embedded in autonomous driving systems. This is also used by vehicle testing facilities and institutions to collect data where the vehicle has had a malfunction, or if the vehicle system responded incorrectly. This recorded data is vital to present as evidence and helps the analysis of the problem.
Should you be concerned about the data you’re collecting?
From these three applications, we can easily conclude that a massive amount of data is being collected across the automotive industry for different purposes. Among the data, the image and video footage have a high chance to include personal information such as faces, bodies, and license plates. According to the GDPR, companies are not permitted to store or use data that contains an individual’s personal information without any consent. As visual data, including image and video footage, is often collected on a mass scale, it becomes impossible for companies to obtain the public’s consent. Therefore, to overcome this challenge, companies should anonymize all data that contains personal information from their database – this ensures that no private information is stored, processed, or leaked, and still allows for the purposes of data collection to be realized.
AI-powered GDPR Compliance
Fortunately, NavInfo Europe’s AI-powered GDPR compliance service provides a quick and fool-proof solution. Personal information markers such as faces and license plates can be detected and blurred by high-accuracy and resource-efficient AI models. Additionally, the deep learning method enables high-performance processing of all kinds of visual data from different types of cameras, across different regions, as well as during different times of the day. In parallel with the anonymization service, we offer the infrastructure for GDPR compliant data handling, which includes data management and pre-checkup, setting up of the anonymization pipeline, data anonymization, and data validation and finalization.
Our approach enables organizations to make use of a resource-efficient and precise alternative to eliminating personal data from collected data. We aim to ensure that our customers and OEM players can accelerate the adoption of connected and autonomous vehicles, while still being able to comply with privacy regulations.