3D machine vision is faster, cheaper and only needs one pixel

Self-driving cars or driver assistance systems rely on multiple sensors, backed by a lot of computing power. But is all that much power really necessary? A research team led by China’s Tsinghua University is looking to reduce computing needs with a tracking system… that only requires one pixel to do the job.

When people first started thinking about self-driving cars or smart security systems, the computer vision systems needed to make them work seemed like a simple matter: Plug a camera into a processor, mix in some algorithms, and voila. Pretty much like a human, processing images almost effortlessly with a pair of “biological cameras.”

Not even close. A human eye is optically simple, but the processing behind it is astonishingly complex: how a human uses this information and reacts to it in an almost reflexive way, activating any part of the body instantly. Copying that ability is a near-impossible task.

The Tsinghua team’s breakthrough involves developing a 3D method for tracking fast-moving objects while reducing computing requirements. The clever way to do this is not with a full image, but by relying on a single pixel.

“Our approach does not require reconstructing the object image to calculate its position, which reduces data storage and computational costs,” said research team leader Zihan Geng. “Acquiring a 3D coordinate requires only six bytes of storage space and 2.4 microseconds of computing time. This could reduce the cost of the equipment needed for high-speed tracking, making the technology more accessible and enabling new applications.

According to the team, with just one detector their technique can track an object 200 times faster than traditional video-based methods, without any prior information and with minimal computing power. It does this by projecting geometric light patterns onto the object and measuring the intensity of a single pixel. Using mathematics (so complex that we can’t explain it to you) the object’s position and trajectory can be calculated.

The team started out by relying on simulations, but then moved on to using a laser and a digital micromirror device (DMD) to illuminate a metal sphere moving down a curved spiral wire. The information in one pixel allowed the system to perform the necessary calculations. So far, the technique is being used to track a single object; the next step will be to move on to multiple objects.

“This technology could improve the sensing capabilities of technologies such as autonomous vehicles, enhance security surveillance systems, and provide more efficient monitoring and quality control for industrial inspection,” Geng said. “In addition, this high-speed localization technique can be used in scientific research, such as insect flight path studies.”

Happy drivers

Client testimonials

Lisa D.

Customer

Theuth has completely changed how I follow automotive news. Relevant articles, thorough reviews, and an active community!

Hanna A.

Customer

I never miss a live show. It's like having front-row seats at every exhibition or race, but from my couch!

Andrew R.

Customer

As a novice in the world of motorbikes, Theuth's guides and analysis have been invaluable for my first purchase.

Ignite the drive, explore the ride