Thesis in the field of Map Learning from September 2024

In Group Research & Mercedes-Benz Cars Development (RD), we are shaping the automobile generations of the future. We are already working today on vehicles that will secure Daimler's technological leadership in the future. Autonomous driving is one of the strategic areas of development at Mercedes-Benz.

A central building block for automated driving is a complete understanding of the vehicle environment. The detection of the environment is primarily based on sensors such as lidar, camera and radar, but a digital road map also plays an important role. The road map provides contextual information for evaluating the current situation, which cannot be recognized by the sensors or only with difficulty, and thus completes the environment detection. The road map also provides important planning information about the route ahead.

A road map for automated driving must be significantly more accurate, more complete, more up-to-date and safer than the map previously used in navigation systems. To achieve this, sensor data from the vehicles is used, among other things, to update the road map and guarantee a correct HD map at all times via a closed-loop process. In the "Map Learning" team, we are working on developing such a closed-loop process for creating highly accurate maps.

Your task will focus on processing the sensor data from our vehicle fleet to create highly accurate and detailed digital road maps. You will create and develop scalable concepts and implement them together with our team. The focus is on deriving a consistent lane model from the processed sensor data using deep learning algorithms.

You will face these challenges:

• Familiarization with existing software framework and state of the art algorithms
• Concept extension and development, as well as implementation of a learning-based approach for generating rich representations, in particular from the field of deep learning, graph deep learning, geometric deep learning and foundation models
• Development and implementation of an approach for estimating the required scope and ensuring the diversity of the training data used
• Concept development and development of an evaluation scheme to evaluate the results against an existing ground truth map and alternative approaches

The final topic will be determined in consultation with the university, you and us.

• Master's degree in the field of computer science, artificial intelligence, robotics or comparable
• Confident written and spoken German and English skills
• Strong programming skills in Python
• Experience in working with ML frameworks (e.g. PyTorch, TensorFlow, scikit-learn)
• High degree of initiative and ability to work in a team
• Strong communication skills

Additional information:

Of course, we are not entirely without formalities. Please apply online only and attach a CV, current certificate of enrollment stating the semester of study, current transcript of records, relevant certificates (max. total size of attachments 5 MB) and mark your application documents as "relevant for this application" in the online form.

Further information on the recruitment criteria can be found"here".

Nationals from countries outside the European Economic Area should send their residence/work permit with their application.

We particularly welcome online applications from severely disabled persons and persons with equivalent disabilities. If you have any questions, you can also contact the site's representative for severely disabled employees at SBV-Sindelfingen@mercedes-benz.com, who will be happy to support you in the further application process after your application.

Please understand that we no longer accept paper applications and that there is no entitlement to return postage.

If you have any questions about the application process, please contact HR Services by e-mail at myhrservice@mercedes-benz.com or by phone: 0711/17-99000 (Monday to Friday between 10 a.m. - 12 p.m. and 1 p.m. - 3 p.m.).