Virtual collision detection of simulated railway vehicle movements
- The VAL-Project, ‘Vollautomatische Abdrück Lokomotive’, is a research and development project of Nuremberg Institute of Technology Georg Simon Ohm (THN) in collaboration with DB Cargo, that seeks to automate shunting operations. As part of this project, the THN is developing a test methodology to validate the decision-making of the automated railway driving system. Therefore, a scenario-based testing approach is chosen. [1] Based on the large number of possible test scenarios, a hybrid test strategy combining virtual simulation (LAB) and real-world (FIELD) tests is discussed. The aim is to run the majority of the required test runs in the LAB, i.e. a photorealistic virtual simulation environment developed at THN [2]. To handle the virtual testing scope, formalised scenario descriptions and a dedicated test case set-up methodology are used [1]. For test run rating, evaluation criteria derived from project specifications, driving service regulations, and other relevant sources are considered and subsequently specific rules defined. ToThe VAL-Project, ‘Vollautomatische Abdrück Lokomotive’, is a research and development project of Nuremberg Institute of Technology Georg Simon Ohm (THN) in collaboration with DB Cargo, that seeks to automate shunting operations. As part of this project, the THN is developing a test methodology to validate the decision-making of the automated railway driving system. Therefore, a scenario-based testing approach is chosen. [1] Based on the large number of possible test scenarios, a hybrid test strategy combining virtual simulation (LAB) and real-world (FIELD) tests is discussed. The aim is to run the majority of the required test runs in the LAB, i.e. a photorealistic virtual simulation environment developed at THN [2]. To handle the virtual testing scope, formalised scenario descriptions and a dedicated test case set-up methodology are used [1]. For test run rating, evaluation criteria derived from project specifications, driving service regulations, and other relevant sources are considered and subsequently specific rules defined. To automatically determine the outcome of a virtual test run, a methodology for monitoring compliance with these criteria is required. A crucial element of this monitoring process involves detecting collisions within the Unreal® Engine virtual environment, which is the focus of this paper.
For this purpose, the native collision detection capabilities of the Unreal® Engine are outlined, while the Separating Axis Theorem (SAT), as well as the ray casting algorithm are introduced for mathematical collision detection. In order to apply these mathematical methods to this explicit engineering purpose, a methodical multistep filtering procedure is developed, to consider objects based on their distance to the locomotive and their proximity to the track, i.e. their relevance to the safe vehicle movement.…


| Author: | Jan Feldhoff, Steffen Schäfer, Christina Theresia Singer |
|---|---|
| ISBN: | 978-3988853363 |
| Parent Title (English): | Conference on Applied Research in Engineering Sciences: ARC 2024 |
| Publisher: | Rediroma-Verlag |
| Document Type: | conference proceeding (article) |
| Language: | English |
| Date of first Publication: | 2024/06/17 |
| Release Date: | 2024/12/16 |
| Volume: | 2024 |
| Pagenumber: | 6 |
| Konferenzangabe: | Applied Research Conference, 13th annual, 2024, Nürnberg |
| institutes: | Institut für Fahrzeugtechnik |
| Research Themes: | Mobilität, Verkehr, Logistik |
