Curriculum vitae / web edition
Stefano Carlo Lambertenghi
Autonomous-systems researcher working across software testing, simulation, and physical vehicle experiments in Munich.
Current work
Scientific Researcher
Responsible for the development, deployment, operation, and maintenance of Fortuna, a Volkswagen Passat adapted and certified for autonomous driving on public roads. Restored the platform to full operational status and lead its use for research and testing.
Designed and ran nearly 100 hours of autonomous-driving experiments on public roads; deployed and adapted Autoware and Apollo, including trajectory-planning changes for narrow two-way urban roads, and end-to-end models from NVIDIA DAVE-2 to recent CARLA Leaderboard agents.
Doctoral Candidate
Topic: measuring and mitigating the sim-to-real gap in testing of cyber-physical systems.
Earlier experience
Intern
Developed ultra-low-power deep-learning-based monocular relative localization for nano-quadrotors, enabling real-time onboard inference on resource-constrained embedded hardware.
Intern
Developed FPGA-based vehicle classification algorithms using support vector machines for intelligent guidance systems.
Research platforms
Fortuna · Full-size autonomous-driving research vehicle
Work across drive-by-wire control, ROS, camera and LiDAR systems, modular and end-to-end driving stacks, simulation integration, and repeatable real-world experiments.
Donkey Car · Small-scale research vehicle
An open-source small-scale platform with its own Unity digital twin, used to compare simulation, vehicle-in-the-loop, and mixed-reality testing against real driving before validating results on Fortuna at full scale.
Education
Doctoral Candidate
MSc in Artificial Intelligence
Thesis: drone-to-drone on-board relative pose estimation and control of nano-class autonomous UAVs. GPA 9.21, summa cum laude.
Bachelor’s in IT Science and Technology
Thesis: interactive holographic displays for advertisement. GPA 97/110.
Selected publications
Real-World Perturbation Testing of Autonomous Driving Systems
Misbehavior Forecasting for Focused Autonomous Driving Systems Testing
Cam2Sim: Neural Scenario Reconstruction for Closed-Loop Autonomous Driving Simulation
PerturbationDrive: A Framework for Perturbation-Based Testing of ADAS
A Multi-Modality Evaluation of the Reality Gap in Autonomous Driving Systems
Ultra-Low Power Deep Learning-based Monocular Relative Localization Onboard Nano-Quadrotors
Skills
- Programming
- Python, C/C++/C#, Java, Swift, MATLAB
- Frameworks & tools
- PyTorch, TensorFlow, OpenCV, ROS, Docker, Unity, Blender, Autoware, Apollo
- Languages
- Italian (native), English (C1)
Profiles
Research and code
Google Scholar · DBLP · GitHub · fortiss/TUM research group · LinkedIn