Master Thesis in AI-based AD/ADAS Virtual Validation

IAV GmbH; Gifhorn; Vor Ort; EUR 8.928 - 13.392; Vollzeit

Zusammenfassung

IAV GmbH is seeking a Master Thesis candidate to explore AI-based virtual validation for AD/ADAS. Work spans vision-based perception, scene understanding, and physical sensor modeling in virtual environments.

You will collaborate with an international team across locations and contribute to cross-site research activities. The scope covers offline AI methods, high-fidelity sensor models, and integration with simulation frameworks such as dSPACE ASM, CarMaker, and MATLAB/Simulink.

Qualifikationen

• Currently enrolled in a degree program in Computer Science, Robotics, Mechatronics, Electrical Engineering, or a related field.
• Advanced knowledge in Computer Vision / Perception, AI / Deep Learning / LLMs / VLMs, or Simulation / Physical Modeling.
• Strong programming skills in Python (C++ or MATLAB is a plus).
• Experience with modern AI frameworks (e.g., PyTorch, OpenCV, transformer-based models).
• Ideally experience with cloud environments or distributed systems.
• Interest in data-driven, scientific, and model-based analysis approaches.
• Independent, analytical, and structured working style.

Aufgaben

• Topic Area 1: Vision-Based Perception & Scene Understanding (Virtual Validation).
• Develop innovative visual perception approaches for offline analysis and virtual validation.
• Research methods for scene understanding, reasoning, and contextual interpretation.
• Develop and analyze hybrid AI architectures combining classical CV with LLMs/VLMs.
• Investigate robust scenario recognition in complex simulations.
• Validate and evaluate model behavior from simulation or recorded data.
• Topic Area 2: Physical Sensor Modeling & Simulation.
• Advance physically based sensor models for camera, radar, and lidar in virtual envs.
• Apply and extend frameworks like CarMaker, Ansys, MATLAB/Simulink.
• Improve realism of synthetic sensor data for ADAS/AD validation.
• Integrate and compare sensor models for offline analysis.
• Explore efficient, scalable validation pipelines.

Kenntnisse

Python

C++

MATLAB

OpenCV

PyTorch

Transformer models

LLMs

Cloud environments

Ausbildung

Master student in Computer Science/Robotics

Tools

dSPACE ASM

CarMaker

Ansys

MATLAB/Simulink

Jobbeschreibung

Master Thesis in AI-based AD/ADAS Virtual Validation

As part of our development and research activities in the field of automated driving (AD/ADAS), we are looking for motivated students who want to contribute to innovative and future‑oriented topics in the context of virtual validation. The focus is explicitly on offline‑based methods for data analysis as well as the development and improvement of simulation and perception models for virtual validation. There is no real‑time ECU or in‑vehicle implementation work involved. We focus on cutting‑edge AI methods for vehicle perception as well as high‑fidelity physical sensor models for virtual development and testing environments. You will work at the intersection of artificial intelligence, simulation, and data‑driven modeling. Depending on your interests, we will jointly select a suitable topic in the interview and flexibly define the scope for a master thesis, mini thesis, internship, or working student position. You will also be part of an international team, collaborating closely with colleagues across multiple global locations and contributing to cross‑site development and research activities.

Your Responsibilities

Topic Area 1: Vision‑Based Perception & Scene Understanding (Virtual Validation)

• Further development of innovative approaches in visual perception for automated driving in the context of offline analysis and virtual validation.
• Research and implementation of methods for scene understanding, reasoning, and contextual interpretation of traffic situations.
• Development and analysis of hybrid AI architectures (classical computer vision combined with vision and large language models).
• Investigation of robust scenario recognition methods in complex on‑ and off‑highway simulation environments.
• Validation and evaluation of model behavior based on simulation or recorded measurement data.

Topic Area 2: Physical Sensor Modeling & Simulation

• Further development of physically based sensor models for camera, radar, and lidar systems within virtual development environments.
• Application and extension of simulation frameworks such as dSPACE ASM, CarMaker, Ansys, and MATLAB/Simulink.
• Improvement of realism and accuracy of synthetic sensor data for ADAS/AD validation processes.
• Integration and comparison of different sensor models in the context of data‑driven offline analysis.
• Investigation of innovative approaches for improving efficiency and scalability of simulation and validation pipelines.

Necessary Skills

• Currently enrolled in a degree program in Computer Science, Robotics, Mechatronics, Electrical Engineering, or a related field.
• Advanced knowledge in at least one of the following areas: Computer Vision / Perception, Artificial Intelligence / Deep Learning / LLMs / VLMs, or Simulation / Physical Modeling.
• Strong programming skills in Python (C++ or MATLAB is a plus).
• Experience with modern AI frameworks (e.g., PyTorch, OpenCV, transformer‑based models).
• Ideally experience with cloud environments or distributed systems.
• Interest in data‑driven, scientific, and model‑based analysis approaches.
• Independent, analytical, and structured working style.

Desired Skills

• Ability to work independently and collaborate within a large, interdisciplinary team.
• Strong communication skills and ability to present technical findings clearly.
• Experience with virtual validation of automated driving systems is a plus.

Compensation: Attractively compensated in accordance with our company wage agreements.

Diversity and equal opportunity are important to us. What matters to us is the individual, with his or her character and strengths.