Advanced Driver Assistance Systems (ADAS) are advancing rapidly, narrowing the gap between today’s semi-automated features and the fully autonomous vehicles of the future. Developing and validating these systems requires test vehicles that not only capture real-world conditions but also simulate rare edge cases and guarantee safety across millions of scenarios. Although hardware platforms, sensors, and vehicle integration often attract the most attention, the real backbone of these vehicles lies in the software stack. Before describing my work in this field at Mercedes-Benz, it is therefore worth outlining some essential aspects of ADAS software.

At the core of every ADAS test vehicle is a layered software architecture. The process begins with the Sensor Data Acquisition Layer, which gathers input from LiDAR, radar, cameras, GPS/IMU, and other sensors. This layer must guarantee precise timestamping, accurate synchronization, and reliable high-bandwidth logging, often reaching several gigabits per second. The data then flows into the Middleware and Communication Layer, where it is distributed between sensors, vehicle buses such as CAN, FlexRay, or Ethernet, and higher-level applications. Because timing is critical in such systems, deterministic frameworks like AUTOSAR Adaptive, ROS 2, or tailored middleware solutions are often employed to ensure real-time performance. Building on this, the Data Processing and Perception Stack converts raw sensor input into a structured view of the environment, identifying objects, estimating free space, and predicting trajectories. For testing purposes, production-grade algorithms frequently run alongside experimental approaches, allowing direct comparison under identical driving conditions. To complete the architecture, Scenario and Event Triggering Software continuously monitors the driving environment, automatically flagging relevant test situations such as cut-ins, pedestrian crossings, or distinctive road features, thereby streamlining post-drive analysis and enabling large-scale validation.

Equally important as the in-vehicle software is the development and deployment process behind it. Modern ADAS projects adopt best practices from broader software engineering domains, with Over-the-Air updates enabling rapid and seamless distribution of new software to entire test fleets. Continuous Integration and Continuous Delivery pipelines automate the full cycle of building, testing, and validating code—ranging from unit tests and integration tests to end-to-end system checks—before deployment. Containerization technologies such as Docker or LXC, together with orchestration frameworks, provide the scalability and version control necessary to coordinate dozens or even hundreds of test vehicles simultaneously. Finally, mature deployment and monitoring strategies ensure that every software version in the field can be traced and managed reliably, a prerequisite for both technical progress and operational efficiency.

Now, how does my expertise in Software Engineering fit into this endeavor of ADAS Software Development?

First and foremost, I must note that all of my work is subject to a confidentiality agreement with Mercedes-Benz AG, which prevents me from sharing specific technical details or implementations.

That said, my journey began in September 2024, when I moved to Stuttgart for six months to pursue the opportunity of writing my Bachelor’s thesis in cooperation with Mercedes-Benz. The thesis, titled Development and Evaluation of Micro-Front-End Architectures for Modular Web Applications in Vehicle Prototypes Based on Objective Requirements Elicitation, revolved around the development of a web application in the context of ADAS test vehicles. Within this project, I contributed to both the system architecture and the implementation of the back end (server side) and front end (client side), with my thesis focusing primarily on front-end system architecture and its implementation.

After successfully completing my thesis, I continued with a full-time internship over the summer, further supporting the development team and contributing to the steady improvement of our software. My work during this period was centered on software architecture, complemented by enhancements to back-end functionality and improvements to UI/UX elements.

With the summer of 2025 now coming to an end, this winter semester I am starting into my Master’s degree in Information Systems at the Technical University of Munich. Alongside my studies, I will continue to support the Mercedes-Benz Test Vehicle Architecture development team as a working student. I look forward to this next chapter with gratitude and excitement, as it allows me to deepen both my personal and professional growth.

Projects