Vehicles with autopilot are equipped with a large number of different sensors for signal processing, which require significant computing power. Scientists at China's Tsinghua University have found a way to dramatically reduce the computing power of an object tracking system by limiting its input data to just one pixel.
Researchers at Tsinghua University have developed a method for three-dimensional tracking of fast-moving objects. This happens at unprecedented speeds with reduced computing power requirements and a single-pixel input limit.
"Our approach does not require the reconstruction of the image of the object to calculate its position, which significantly reduces the cost of data storage and computing power. So, to obtain 3D coordinates, you need only six bytes of disk space and time for calculation. By reducing computing costs and increasing efficiency, the cost of the equipment needed for high-speed tracking can be reduced, making the technology more affordable and opening up new opportunities for its application, — said Jihang Gen, head of the research group.< /p>
The detector developed by Chinese scientists allows you to track objects 200 times faster than traditional methods based on video processing — with minimum computing power. For this, geometric light patterns are projected onto the object and the brightness of one pixel is measured. Using complex mathematical calculations, you can determine the location and the object's trajectory — so far only one is tracked, but in the perspective there will be several objects.