Software plays a key role for the automotive industry. What strategy is Continental pursuing and what is planned for the future? In July 2021, Software expert Dr. Karsten Michels from Continental Automotive provides insights in an interview.
More than 90 percent of innovations in automotive come from software and electronics development. Software is the catalyst of modern mobility, and without it much of what we expect today would not be possible: a high level of comfort, seamless connectivity both in communications and in connecting different means of transport, efficient use of resources and much more. Autonomous mobility from driverless trains to people movers through to individual travel – all of that would be inconceivable without software.
The classic application of software in cars is embedded, developed for the respective use case and standardized only to a small extent. That is not particularly efficient. Even just the required scope of functions makes this approach impossible for the future. Software will increasingly no longer be specifically developed for each individual vehicle. The proportion of software that is bought or is open source is growing at a fast pace. This increases the necessity to standardize interfaces and other general conditions. The objective is to have software that, primarily thanks to the tools used, is developed in a fully automated way, is free of bugs and is secure. It must, of course, also meet all the requirements of customers and end customers and make the best possible use of the existing hardware resources. Simply perfect software that resolves all problems and doesn’t create any new ones.
The development of the user interface has to keep pace with the increasing complexity of the software. The vehicle must guide the passenger without patronizing so as to offer the required functions from the vast range of possibilities in an understandable form. The vehicle thus becomes an assistant and concierge. The vehicle must have a certain intuition and operate with foresight. Artificial intelligence plays a particularly prominent and decisive role in autonomous driving.
The number of types and in particular the volume of data are constantly increasing. That is also one of the reasons why the architectures in the vehicle will change. Sensors will have standardized interfaces and send their data to zone controllers. These process the data independently to some extent. For the most part, however, they forward the data to high-performance computers that bundle many of the functions and become central data centers. When organized in this way, the number of interfaces can be reduced. This allows the data flow in the vehicle to be much better orchestrated. The data flow into the infrastructure – into the cloud, for edge computing or any infrastructure elements – such as for traffic management (V2X) – then takes place via broadband and with low latency by means of 5G or subsequent standards through a central communication control unit or a high-performance computer.
The rapidly growing requirements in terms of functionality of the vehicles, such as user experience or advanced driver assistance systems, demand increasingly complex software and data flows. These can no longer be implemented by adding further traditional control units. The requirements of the on-board power supply alone would exceed the scope for installation in the car. The solution is new architectures that are used to consolidate many functions and place them in high-performance computers. Using these architectures reduces the complexity. At the same time, the vehicle remains easy to build and is able to continuously process updates. This is imperative in order to ensure the life-long security of the vehicle.
Requirements in terms of data security and also privacy are increasing. Besides firewalls, additional measures are being taken at the communication interfaces. Communication between the control units within the vehicle is therefore secured.
Even after a possible breach of the firewall, a check takes place in the individual control units to verify whether or not the relevant data was sent by an authorized sender. It cannot go any further without authentication. While software prevents intrusion, a check is run in parallel to see if illegal intrusion took place. Intrusion detection software modules are used for this. In the unlikely event that an intrusion took place, the vehicle reports this independently. It gives a warning and a check is then carried out as to what the countermeasure may be. In order to be able to provide the best possible technology for this and to make vehicles as secure as possible, Continental acquired Argus, which is among the leading security experts in the sector.
The quality of the software depends on development processes. Numerous tools and processes are required to securely manage the complexity of today’s software developments. To give an example: The ICAS1 high-performance computer developed for VW’s ID.3 contains around 20 million lines of code to map the many thousands of requirements. The software is developed by 19 different parties in diverse locations. This only works if the software architectures are stored in tools, are linked with requirements and the software is then automatically and continuously created and tested in the development process. The teams that implement such a project must be managed in accordance with this complexity. Continental uses agile methods for this.
Agile software development plays to its strengths particularly when projects are of a certain complexity. It allows complex overall tasks to be subdivided into packets, which are worked on and tested by small teams on their own responsibility. Ownership by the individual teams gives rise to enormous motivation. The small packets are planned in great detail in what are called sprints. A review takes place after each sprint. What has been completed, what is outstanding – and then can a new plan be made based on this? The status is always visible. That leads to very controlled development, which is, however, very flexible and fast at the same time.