Speakers
Description
We cooperate with IXPERTA s.r.o. company in the development of a software simulator capable of generating the training data for creation of an obstacle detection system used in a railway vehicle. The project is supported by Technology Agency of the Czech Republic (TACR), in the program of industrial research and experimental development TREND and ends in 2022.
The simulator will allow to create a virtual 3D environment of a railway track as a virtual replica of a real track. Together with this, acquisition of various sensors’ output from the simulated environment will be available. This includes RGB camera, LIDAR, thermal camera etc., which will try to mimic their real counterparts mounted on a railway vehicle.
In the virtual environment, it will be possible to simulate multiple critical scenarios that might arise on a real track. Besides, it will be possible to mimic different weather conditions such as rain, snow, fog and influence the generated output data. The simulator will therefore provide virtualization of a railway track conditions and it will carry out test runs in a laboratory environment.
As a part of the simulator development, it is necessary to specify a set of scenarios that can occur and select parameters of the virtual environment that can be influenced. In this way simulator’s complexity will be limited while also ensuring required variability and quality of the output data to fulfil the training of the obstacle detection system. To provide high quality of the generated data we utilize ray tracing as a rendering method. Due to the computational complexity, we perform rendering on the IT4Innovations’ infrastructure.
In our contribution we will describe our goals in more detail, and we will present our actual progress in the project.
