Currently, almost all visual sensors in autonomous driving systems are RGB sensors in the visible wavelengths, emulating human visions. Such practices naturally exclude a lot of useful information in other spectral ranges, severely degrading the detectivity of machine vision and leading to fatal mistakes.

Infrared spectral imaging can improve the reliability of autonomous driving in a large scale. First, it can provide more accurate obstacle detection and recognition. By detecting infrared signals, an autonomous driving vehicle can classify human bodies, animals, and other vehicles with improved accuracy. Second, spectral imaging can be used in traffic sign recognition. As a result, infrared spectral sensing can provide a more comprehensive survey of surroundings and provide more information for decision-making.