Competence Center AI EngineeringProblem definition
A method of "Visible Light Communication" has been developed for communication between vehicles. The method, which uses high-frequency headlight flickering for message exchange with the help of an optical sensor, is to be tested. For this purpose, it must be examined to what extent the received image data can be processed in order to recognize vehicles by means of neural networks and at the same time extract the message data. Only a single prototype of the system exists, which also lacks road approval, so that real public traffic situations, and potential interference between multiple transmitters, cannot currently be tested.
Goal
The communication system is to be evaluated in typical traffic situations of connected driving in order to evaluate the robustness of object recognition and message exchange. In particular, interference effects between several transmitters and bright ambient light are to be quantitatively tested.
Solution expertise
The simulation environment OCTANE already supports the simulation of traffic situations and physically-based sensors. Optical effects of the receiving system (optical low-pass filter, rolling shutter effect of the CMOS sensor) and radiometric profiles of vehicle head- and taillights complement the image simulation. Neural networks are used to identify other vehicles in the unfiltered image data.
Test bed
OCTANE simulation platform with photometric shader and object detection in sensor data.
