Autonomous Simulation
My master's thesis: a simulation tool for testing self-driving off-road vehicles in realistic virtual terrain before they reach a real field.
- client
- Robotics Research Lab, RPTU Kaiserslautern
- my role
- Master's thesis
- year
- 2023
- status
- completed
01 the brief
Testing autonomous vehicles off-road is slow, expensive and risky. The lab needed a way to run the same situations again and again, including critical ones, without a real vehicle in the field.
02 what i built
I built a high-fidelity simulation environment in Unreal Engine and connected it to Finroc, the lab's C++/Java robotics framework. On top of it I designed learning scenarios, routine tests and critical test cases, and wrote a Python tool that turns the simulation data into clear performance and safety results.
03 key features
- 01Realistic off-road environment in Unreal Engine (C++)
- 02Connected to the Finroc robotics framework
- 03Learning scenarios, routine tests and critical test cases
- 04Python analysis tool for performance and safety results
04 outcome
Vehicles can be tested repeatably, including situations that would be too risky to stage for real.
05 gallery
// images_upload_pending
Visual assets for this deployment will be uploaded soon.