Oculii, a software startup that aims to improve the spatial resolution of radar sensors by up to 100 times, has secured new investment from General Motors. The new funding, which the two companies say is in the millions, comes just months after Oculii closed a $55 million Series B round.
Oculii and GM have been working together “for quite some time now,” CEO Steven Hong told TechCrunch in a recent interview. While he declined to specify exactly how GM plans to use Oculii’s software, it could be used to boost the capabilities of the automaker’s advanced hands-free driver assistance system, known as Super. Cruise. Hong added that the company was also working with a few other OEMs, including one on the ceilings table.
“When a company like GM says it’s great technology and it’s something we potentially want to use down the line, it forces the whole supply chain to consider that. and to work more closely with you to adopt the solution, the technology, in what they sell to OEMs,” he said.
The startup has no plans to build hardware for its automotive customers (although it does work with robotics companies for which the company makes sensors, a company spokesperson said). Instead, Oculii wants to license software to radar companies. The startup says it can take low-cost, off-the-shelf radar sensors — sensors that weren’t designed for autonomous driving, but rather for limited scenarios like emergency braking or assist. to parking – and use its artificial intelligence software to enable more autonomous maneuvers, Hong said. .
“We really think the way to deliver something scalable is through software, because software fundamentally improves with data,” he said. “Software fundamentally improves with better hardware in every generation that’s released. Over time, software gets cheaper and cheaper, much faster than hardware, for example.
The news is certainly bullish for radar, a sensor typically used for assistive capabilities due to its imaging limitations. But if Oculii can actually improve the performance of radar, which tends to be much cheaper than lidar, it could mean massive savings for automakers.
Tesla, the world’s largest electric vehicle maker by sales volume, recently removed radar sensors from its advanced driver assistance system, in favor of a “pure vision” approach that uses cameras and a neural network powered by a supercomputer. Hong said the radar that Tesla eliminated was very low resolution and “really didn’t add anything to their existing pipeline.”
But he doesn’t think the company would necessarily always rely on radar, if the technology improved. “Basically, each of these sensors improves [the] security case and brings you closer and closer to 99.99999% reliability. In the end, that’s the most important thing, getting as many reliability nines as possible.
The story has been updated to include clarification on the company’s hardware manufacturing.