Active NIR illumination for improved camera view in automated driving application

The advance of automated vehicles imposes increasing requirements on the sensor system of vehicles. Besides the ongoing development of perception algorithms, different hardware approaches exist in order to improve the detection of infrastructure and road users. In the far-field in front of the vehicle, the detection of infrastructure and road users relies on camera and LiDAR systems. However, the reliability of both systems is influenced by weather conditions, especially due to reflections from snow, rain, or fog and the camera by the ambient lighting as well. Optimized algorithms are implemented to improve the vision of both systems but limitations remain. RaDAR is proven to work more reliably in adverse weather conditions but struggles in providing sufficient data for detailed object classification. In combination with data fusion, the sensor systems can provide a partly redundant perception of the road and its users. This paper aims to provide a proof of concept for the improvement of the vision of camera systems in low light by using active NIR illumination. For this purpose, the spectral emission of visible and near-infrared sources is compared with the sensitivity of a camera. Considering regulatory emission limits, an optimal wavelength for additional NIR lighting is determined.