Most of the autonomous systems currently operating on road, and even those under development, use some or the other combination of video cameras as ¡®eyes¡¯ for observing the environment of the vehicle. While these are apt in most of the use case scenarios until now, they might not be enough for fully autonomous vehicles, which are poised to ply on roads in the near future. Reason being, the input from these is mostly in the form of a 2D map, which can pin-point the anomalies on the road but might not be enough to for a speeding vehicle and its corresponding interaction with moving objects in the vicinity.
So, as a solution, either you use too many video cameras, which collaboratively provide an in-depth imaging of the road and its objects. Or else, you come up with new and better technologies for the purpose. Purpose here solely being to eliminate the need of a driver to operate a vehicle. The alternative, or rather complementary, technologies invented so far include LiDAR, which makes use of pulsating lasers to measure range. Though in case one would want to stick to cameras, thermal imaging can act as the best solution at hand.
Thermal imaging, or thermography, captures the heat emerged from objects to map the surrounding based on that. Kind of like the movie ¡°Predator¡± where the protagonist sees everything in thermal vision. Now, to use that as an autonomous driving technology might seem a bit far fetched. But, know this, thermal imaging has since long been used for military purposes. Given such history, it can easily be deduced that the technology has come a long way since its inception and much advancement has been gained on it over time. So much so, that as of today, it is practically used in almost every field, including by firefighters to see through smoke or by maintenance engineers to find overheated sections of an element. The list of its applications is endless.
So, why not self-driving cars? After all, technology has proven its worth in every area of application and can be a huge boost to the autonomous systems. As a matter of fact, it has. When applied in autonomous systems, it was seen that the imaging mechanism had a huge advantage over the video cameras being used previously.
In essence, thermal imaging uses Infrared to map the surroundings based on heat emissions. This eliminates the dependency on light, meaning thermal imaging is quite advantageous over normal video cameras especially during low light conditions.
A major player in the thermal imaging infrared camera game is FLIR. The company has managed to come up with a range of infrared cameras that are meant to help autonomous systems for cars. The integration of the camera input with the autonomous system is made possible by a high-resolution Thermal Vision Automotive Development Kit (ADK) offered by FLIR. The results shows a wide contrast in comparison with regular video cameras. You can check out the difference and judge for yourself in the following video:
As can be clearly seen in the videos, thermal imaging infrared cameras deliver a much better result when it comes to mapping the road. Thus, if self-driving hardware needs to be upgraded to the next level, thermal imaging might have to be a crucial part of it, at least for now.