Design and Adaption of Neural Networks for Autonomous Driving

Hilfskraftstelle

Themen-Schwerpunkt: Elektromobilität, Maschinelles Lernen, Mobilität, Real Time Data Management, Software-Entwicklung
Studiengänge: Elektrotechnik, Informatik, Verwandte Studiengänge

Umfeld

When developing highly automated vehicles in a real-world environment, large amounts of data are generated that, according to current knowledge, can no longer be processed using conventional processing and analysis methods. Numerous sensors such as cameras, radar or lidar generate this data. As part of the KIsSME research project, solutions are being developed that address the on-board selection of data in real time. During driving operation, it has to be decided whether the data is relevant, sub-critical or even critical, so that a recording can be triggered, based on the situation.

Aufgaben

We are looking for a student (m/f/d) with good knowledge in the ML domain. As part of the project, we strive for an object detector built upon a state-of-the-art convolutional neural network architecture (f.i. YOLO) specifically built for autonomous driving. Your responsibilities include but are not limited to:

  • Literature research
  • Evaluation of state-of-the-art object detection NNs
  • Domain specific adaption of an existing architecture

Wir bieten

  • An inter-disciplinary work environment with partners from science, industry and users
  • A pleasant work atmosphere
  • Constructive cooperation
  • Excellent technical equipment
  • Fully automated coffee machine with premium beans

Wir erwarten

  • Required skills:
  • Programming in C
  • Machine Learning Theory: DNN, CNN, …
  • Development on Linux systems
  • Desirable skills:
  • ML-Frameworks and data curation: ONNX, SciKit Learn, …
  • Very good knowledge of German or English
  • Motivation and commitment

Bewerbung

We are looking forward to your application in PDF form including the following content:

  • Curriculum vitae
  • Current transcript of records

Weitere Informationen