The automotive industry is currently experiencing a major change due to software. Manufacturers and suppliers need solutions to guarantee the security and smooth functioning of the vehicle software. AI-based vehicle software intelligence helps them understand software dependencies, discover redundant code, and document updates.
Many industries have undergone upheaval due to new technologies. The automotive industry is currently experiencing a major change due to software. According to the current Automotive Software Survey, 56% of those surveyed assume that car manufacturers will generate more than five percent of their sales by 2027 from the sale of software that is transferred to the vehicle via OTA updates. 21% of the experts even assume that software sales will already account for more than ten percent of sales in 2027. Therefore, some automobile manufacturers will change into software groups because there is increasing sales and profit potential in the sale of software functions.
Demands on workers, life cycles of vehicles, and business models for manufacturers are turned upside down when car manufacturers transform into software companies. If OEMs want to lead the competition, they have to address the topic of Vehicle Software Intelligence (VSI). VSI is a category of solutions based on AI that provides detailed insights into vehicle software. Vehicle Software Intelligence solutions map the dependencies between different software systems in the vehicle and their changes and functional behavior. They help to increase software quality and security. We present three use cases for car manufacturers to rely on vehicle software intelligence solutions.
Understand Software Dependencies
A current study by the Technical University of Munich in cooperation with the BMW Group examines the dependencies of a modern vehicle software system. This shows that there are 1,451 dependencies between the 94 software systems. Vehicle software intelligence uncovers these interdependencies and analyzes the behavior of the software systems. This allows vehicle manufacturers to track how changes in one system affect every line of code independent of other systems. This transparency is crucial for proactive maintenance, vehicle safety, and the ability to implement new regulations.
Discover Superfluous Code
Much of the vehicle software is developed by the open-source community. Some vehicles still run software code that was developed many years ago. In addition, the manufacturer’s software interacts with the systems of numerous suppliers. As a result, it is often difficult for car manufacturers to obtain ASIL-D (Automotive Safety Integrity Level) certification, which stipulates that no extra code may run on vehicles.
Vehicle Software Intelligence solutions enable manufacturers to track down extra code – increasing safety and compliance with ISO 26262 ASIL functional safety certification.
Document Software Updates
According to the current Automotive Software Survey, every vehicle will receive between two and six over-the-air (OTA) updates from 2025 onwards. The World Forum for the Harmonization of Vehicle Regulations (UNECE WP.29) guidelines states that updates must be secure and documented. Vehicles only retain their type approval without additional tests if the manufacturer can prove that an update only fixes an error or that it is a security patch.
With the help of AI-based vehicle software intelligence, vehicle manufacturers can prove which lines of code have been added and which functions are affected by a software update. This speeds up the type approval process and reduces the associated costs.
Vehicle Software Intelligence In Three Steps
In the following, we describe an example of what an AI and machine learning-based vehicle software intelligence solution looks like in three steps.
1. Validate: The software’s structure, relationships, and dependencies are checked before being integrated into the vehicle’s devices. An algorithm identifies and maps at a functional level which software is running in the proposed update and which existing software is affected. This technology also shows which vehicle functions require renewed type approval. The knowledge gained from this greatly simplifies the official approval process.
2. Recognize: Machine learning algorithms analyze the behavior and relationships between the vehicle’s millions of software code lines. This allows automakers to detect anomalies in software behavior that could indicate a software configuration change, software bug, or hack. The Vehicle Software Management solution can predict which monsters will become problems and lead to system failures. This makes it possible to identify potential errors before they occur.
3. Update: The last step is the over-the-air update (OTA update). Unlike solutions that check the software binaries for changes, line-of-code updates consist of much smaller update files since only the changed lines of code are updated. This reduces the amount of data to be transmitted and thus also the transmission costs. Line-of-code updates also allow controllers to be updated without taking them offline. This means that the daily use of the vehicle is not affected.
Business-Critical Benefits Through Vehicle Software Intelligence
A modern car has more than 100 separate ECUs using different chipsets with different memory sizes, clock speeds, and operating systems. For example, the control units for driver assistance systems are connected via several in-vehicle networks with other protocols and even Ethernet. They are offered to the OEMs by various providers, integrators, and suppliers.
AI-based Vehicle Software Intelligence supports car manufacturers in understanding software dependencies, discovering redundant code, and documenting updates. As a result, VSI solutions help ensure the safety of connected and autonomous vehicle systems while reducing costs. This offers business-critical advantages for automobile manufacturers and significantly increases consumer acceptance of driver assistance systems.