Autonomous driving has made significant progress in recent years, particularly in good visibility and lighting conditions. However, night-time environments continue to pose a major challenge, as limited lighting reduces the performance of many sensor systems. Unlike active sensors such as LiDAR or radar, passive sensors do not emit their own signals but rely exclusively on existing radiation from the environment, which offers advantages such as lower energy consumption and a reduced electromagnetic signature. This is particularly relevant in a military context, as active sensors are potentially detectable and could therefore reveal a system’s position.
The aim of the project is therefore to investigate approaches to improving autonomous navigation at night, utilising predominantly passive sensors. The focus is particularly on camera systems operating in the visible and infrared spectral ranges. Through the use of modern image processing and machine learning techniques, the aim is to develop methods that enable robust object detection and scene analysis even under difficult lighting conditions. The results of the study are intended to help make automated systems more reliable for night-time operational scenarios. This is relevant both for civilian applications, such as in autonomous logistics, and for military applications where vehicles must operate discreetly and safely in environments with restricted visibility.
