The Role of AI and AI Tools in Automotive Software Development and Testing

Artificial Intelligence (AI) is a hot topic in many industries and applications. But it’s not a new concept for the automotive industry. AI, specifically machine learning – the process of teaching machines to learn from data and improve over time – has been used for quite some time, initially for enhancing physics-based approaches in development.

However, the arrival of connected vehicles and the creation of data from various sources, such as customer usage and fleet data, have significantly transformed the landscape. In the past, data had to be generated, today the industry leverages vast amounts of data from these sources to train AI models in real-time, enhancing the development process.

AI in Automotive

AI's impact on the automotive industry is complex. It is crucial to distinguish between different AI applications, such as machine learning models that enhance physical models, large language models that improve software development efficiency, and AI tools that aid in understanding and processing documentation. These applications are critical in various aspects of automotive development, from improving user interfaces to enabling autonomous driving.

One of the primary challenges in integrating AI into automotive development is managing the vast amounts of data generated. Traditional methods of data validation and processing are no longer sufficient. Instead, advanced infrastructure and platforms are required to effectively handle and filter this data. Despite these challenges, the benefits of AI are substantial and can significantly reduce the time and cost associated with manual processes. For example:

• Data from a single source can be used in different ways by various teams (e.g., calibration and software development)
• Testing processes can be streamlined
• Development activities can be optimized and continuously operate

AI in Software Development and Testing

In software development and testing, AI tools have been instrumental in improving efficiency and reliability for years. For example, AI can generate test cases from requirements, reducing the time and potential errors associated with manual processes. Afterall, AI can work 24/7, but as with anything data driven, results and analysis are only as good as the data it’s using.

Additionally, AI can analyze video and create artificial data from real-world scenarios that can’t be measured physically and would be too expensive to create a tool for, enhancing test coverage and confidence in the results. It also can be valuable in using requirements to generate test cases. Previously this was a manual process where an engineer had to read and interpret requirements and then create test cases. Using AI, it can all be automated, significantly decreasing the time involved and opportunity for errors and/or misinterpretations.

However, you can’t take shortcuts in software development and testing. The integration of AI in safety-critical areas, such as autonomous driving, requires rigorous validation and testing to ensure reliability and safety. So, for now, AI is a tool to improve development time and create confidence in testing results.

The Future of AI in Automotive

Looking ahead, the potential applications of AI in the automotive industry are vast. As computing power becomes more affordable and accessible, AI's role in development and testing is expected to expand. For example, there’s limited computing power in a vehicle, so companies are looking at extending the computing power to the cloud.

However, the industry must address regulatory and liability issues, particularly concerning autonomous vehicles. If a mistake is made and an accident occurs, who is liable? The car made the decision, so is it the OEM, the software supplier, or others? This is a complex facet that industry must work through as autonomous technologies advance.

That raises the question of how fast we move forward. No one wants to be last, but moving too fast can result in losing some control as to what is going on, and when it comes to safety, that’s not acceptable. Because our work at ETAS deals with vehicle safety, we focus on how to make a vehicle safe and secure, so we need to be in control of and rely on the output of AI tools. We see a future where there’s a balance between leveraging AI for efficiency and ensuring safety and reliability through stringent validation processes.

ETAS's Approach to AI

ETAS is at the forefront of integrating AI into automotive development and our commitment to innovation ensures we will remain a key player in the evolving landscape of automotive AI. Our approach includes using AI to process and optimize data, generate test cases, and enhance model reliability. For example, we use AI tools on both sides of the development chain:

• Left Side/Coding: Generative AI and a chatbot tool helps developers using the ETAS tool chain with common issues in AUTOSAR, c-code generation, ARXML files and configuring base software. In doing so, the AI tool learns the common issues, providing an answer to something it already knows, or making it easier for a human to take over the process for an undocumented issue.
• Right Side/Validation: In the AI calibration suite, large data is used to optimize calibration based on real-world information. Traditionally, this was a time-consuming process done manually by a person. By using AI, there’s significant efficiency improvements and the knowledge is retained, unlike if a person with the knowledge were to leave the company.

Another example is the ETAS Embedded AI Coder, which enables safe, quick and efficient deployment of AI functions on electronic control units (ECUs). By using this tool’s state-of-the-art performance, development time is significantly reduced, which frees up traditional resources needed to generate C code. And, when incorporating ETAS’s ASCMO, it can generate AI models for multiple domains, providing a one-stop solution for model development and C code generation. By working together, these tools expand the embedded coder ecosystem, allowing non-AI experts to conduct model generation.

AI and AI tools are revolutionizing the automotive industry, offering unprecedented opportunities for efficiency and innovation. As the industry continues to evolve, the integration of AI will play a crucial role in shaping the future of automotive development and testing. ETAS's proactive approach to leveraging AI underscores the transformative potential of these technologies in creating safer, more efficient vehicles.