Modern Advanced Driver Assistance System (ADAS) control units have greatly increased the complexity of tasks faced by mechanics and auto electricians. These professionals must now manage an extensive range of instructions, manuals, and tools.
An initial outcome of the AI-Cares project, led by SNJ Media with support from BI-REX and contributions from BitBang, focuses on streamlining the recalibration process for the aforementioned control units. The project introduces a GenAI-powered system designed to quickly identify the required tools and manual references, facilitating more efficient interventions.
The AI-CARES project is financed by the European Union – Next Generation EU.
AI for Navigating Complex Knowledge
The growing prominence of ADAS control units in modern vehicles has dramatically increased the complexity of tasks and skills demanded of mechanics and auto electricians. These professionals must contend with a wide range of instructions, manuals, and tools that vary significantly across brands and models, requiring them to handle diverse materials such as text, images, schematics, and videos.
Generative AI provides innovative solutions to manage this complexity by simplifying the search for answers and connections within unstructured databases while offering operators a user-friendly natural language interface. Achieving these results, however, demands identifying the most suitable architecture and technologies, along with meticulous training to ensure fast, reliable responses and long-term sustainability.
Crafting a Tailored Solution
BitBang conducted a detailed analysis of the needs, context, existing manuals, activity history, requirements, and the constraints and opportunities related to infrastructure and technology to develop a customized GenAI system.
The Proof of Concept was developed using Amazon Web Services architecture. A suitable model was selected and trained with Retrieval-Augmented Generation (RAG). The RAG approach was chosen for its ability to minimize hallucination risks by equipping the Large Language Model with a library of secure, domain-specific sources.
To ensure high-quality responses, the team refined prompting and tokenization mechanisms while establishing an objective framework to evaluate results. Metrics such as Precision, Recall, and F-Measure were used to guide response optimization through fine-tuning and successive iterations.
This robust and meticulously calibrated system offers a secure foundation for expanding the project in various directions, including the types of content analyzed, interaction methods, response selection, and overall operational flow. The ultimate goal? Delivering a reliable solution that accelerates activities while enhancing both employee and customer satisfaction.
Project Collaboration
SNJ Media is an innovative SME that develops customized IT solutions and offers related software and IT services. The company launched the AI-Cares project to bring the benefits of Artificial Intelligence to the automotive sector, with a particular focus on enhancing post-sale services. The ADAS application serves as a prime example of the potential created by this synergy.
BI-REX, the Competence Center dedicated to Industry 4.0 and Big Data, supported the project’s launch in 2023.
BitBang is a consultancy firm specializing in Data and AI and a BI-REX consortium member. It brings its specialized expertise to the AI-Cares project, focusing on the development and refinement of the AI model’s technological components.
