Fully autonomous driving is on everyone’s lips. Some expect more safety on the road. The others are afraid of the autonomy of artificial intelligence and the ethical consequences. Fully autonomous driving (level 5 of autonomous driving 1 ) on our freeways is still a thing of the future.
On the other hand, the increasingly sophisticated driver assistance systems in modern vehicles, intended to make life easier for the driver and improve safety, have long been a reality. These assistance systems are currently hurdling to level 3, highly automated driving (HAD). And that not only takes place in the vehicle but also in the cloud. You can read here why this is the case and its meaning for software development.
Vehicle Sensor Data Is Not Sufficient For Highly Autonomous Driving
Today’s driver assistance functions are mostly based on evaluating data from different sensors and systems of a vehicle such as a camera, ultrasound, lidar, or radar. These recognize partial sections of the environment and are combined to form a uniform environment model. This environment model provides a wide range of information on moving objects, static obstacles, the course of roads, and more. The vehicle’s “field of vision” is restricted to this area.
However, this is not enough to consistently continue on the path to highly automated driving. Rather, the previous environment model for calculating and checking the route must be significantly expanded. This is done with information coming from outside the sensor range. The driving functions can be dynamically improved with them, and even completely new driving functions can be developed. This means that IT systems outside the vehicle are becoming part of vehicle dynamics functions for the first time.
IT Systems Outside The Vehicle Are Becoming Part Of The Driver Assistance Systems
With the new vehicle variants of many manufacturers, additional vehicle functions are coming onto the road, supported and expanded by IT systems in the cloud. For example, the backend in the cloud keeps the map material up to date and provides dynamic information. This information is based on both external information sources and aggregated sensor data from many vehicles. IT systems in the background collect this data from various sources, process it, and deliver it back to the vehicles.
A HAF IT system first determines whether the information in question is relevant for a vehicle. The vehicle’s location serves as a basis, and all events occurring in the vicinity of this location are potentially relevant to a vehicle. The radius is determined depending on the type of event. Since the location of a vehicle constantly changes while driving, only the changed information must be transmitted to the vehicle.
Vehicle And HAF Systems Are In Constant Exchange
The IT system provides various environmental conditions such as speed limits, construction sites, or traffic light phases. These can be transmitted to the vehicles as so-called events. The vehicles can register for one or more events to receive notifications about the individual events.
There can be very different types of events in which the vehicles or their drivers can be interested, for example, information on variable electronic speed limits. Due to traffic conditions, construction sites, or other circumstances, these may show different values. Another type of event is warnings of dangerous driving conditions such as black ice. Such early warnings should help to avoid accidents.
The vehicle can also be informed via route approval events which sections are suitable for autonomous driving. RTTI information on the current traffic situation is also possible, as are events that transmit data on the duration of traffic light phases. With the latter, it is in principle possible to determine the speed accordingly to arrive at the next traffic light with green.
An IT system for HAF forms the basis for a comprehensive platform that enables further digital services in hazard warning, navigation, traffic information, and parking. Here, too, it is true that the automobile manufacturers have only implemented some of these events, and much is still a future dream.
An Agile Cloud-Based Approach Is A Success Factor For HAF Systems
For the highly automated driving to function reliably with the support of an IT system in the background and to be expanded in the future, the developed backend software requirements are high. Reliability and performance must be ensured and scalability for future expansions.
Double slash develops HAF backend systems with customers and partners that collect and evaluate information and distribute the dynamic information calculated to the vehicle fleet. During development, our experts identified the following success factors:
Cloud-Based Approach And Microservice Architecture
The technological foundation of a HAF application is the customer cloud. The cloud-based application is used to guarantee the high availability of the system in all regions and to be able to serve a constantly growing vehicle fleet. The IT system is based on a flexible microservice architecture with separate, independent services. Current cloud technologies are used for monitoring and logging.
SAFE – Agile Collaboration On A Large Scale
An agile process model makes flexible reactions to changes possible. Since the project is to be seen in a larger overall context with various IT systems in automated driving, the SAFE (“Scaled Agile Framework”) model is ideal here.
The Agile Manifesto is the basis for SAFE, from which – as for Scrum – the SAFE principles are derived. However, SAFE goes one step further, as the framework is intended to coordinate multiple agile teams and projects. These Scrum Teams work towards a common goal; their projects are part of a larger product. By merging the individual teams, dependencies can be identified, and redundancies avoided. Your sprints have a common rhythm, and the results of the individual teams are coordinated through overall planning at the program increment level 2.
Conclusion
For highly automated driving (HAD), a large amount of information is required outside the sensor range of a vehicle so that the driver assistance system can learn. For this purpose, IT systems outside of the vehicle provide processed dynamic information based on a wide variety of information sources. Such systems have particularly high performance, reliability, and scalability requirements. We rely on the latest cloud technologies and an agile but scalable process model to ensure this. With the latter, you can react quickly and flexibly to changes without neglecting dependencies on other projects in the overall context.
Also Read: The Benefits Of Cloud Computing