Stanford scientists have developed an artificial intelligence camera that can recognise objects faster, and could be used to help autonomous vehicles better navigate through obstacles. The image recognition technology that underlies autonomous cars and aerial drones depends on artificial intelligence: the computers teach themselves to recognise objects like a dog, a pedestrian crossing the street or a stopped car.
The problem is that the computers running the artificial intelligence algorithms are currently too large and slow for future applications like hand-held medical devices. The researchers from Stanford University in the US combined two types of computers to create a faster and less energy-intensive image processor. The first layer of the prototype camera is a type of optical computer, which does not require the power-intensive mathematics of digital computing.
The second layer is a traditional digital electronic computer, according to the study published in the journal Nature Scientific Reports. The optical computer layer operates by physically pre-processing image data, filtering it in multiple ways that an electronic computer would otherwise have to do mathematically. Since the filtering happens naturally as light passes through the custom optics, this layer operates with zero input power. This saves the hybrid system a lot of time and energy that would otherwise be consumed by computation.
“We’ve outsourced some of the math of artificial intelligence into the optics,” said Julie Chang, a graduate student at Stanford.
The result is profoundly fewer calculations, fewer calls to memory and far less time to complete the process. Having leapfrogged these pre-processing steps, the remaining analysis proceeds to the digital computer layer with a considerable head start.
“Millions of calculations are circumvented and it all happens at the speed of light,” said Gordon Wetzstein, an assistant professor at Stanford.
In speed and accuracy, the prototype rivals existing electronic-only computing processors that are programmed to perform the same calculations, but with substantial computational cost savings. Researchers said their system can one day be miniaturised to fit in a hand-held video camera or an aerial drone. In both simulations and real-world experiments, the team used the system to successfully identify airplanes, automobiles, cats, dogs and more within natural image settings.
“Some future version of our system would be especially useful in rapid decision-making applications, like autonomous vehicles,” Wetzstein said.
