Ouster’s new color LiDAR replaces cameras

For the past decade, the tech industry has been considering whether self-driving cars need LIDAR sensors, cameras, or all of the above. LiDAR company Ouster says it has found a new answer: putting both in the same sensor.

On Monday, the San Francisco-based company announced a new lineup of lidar sensors it calls “Rev8.” All of these provide so-called “native color riders”. These sensors can simultaneously capture color images and three-dimensional depth information, performing the functions of two sensors in one.

Ouster CEO Angus Pacala, who said the company has spent a decade developing it, was not shy about his ambitions for the new product lineup in an exclusive interview with TechCrunch, calling it “the holy grail of what roboticists have always wanted.”

“Throughout human history, we’ve been buying LIDAR sensors, buying cameras, trying to understand that combination with higher-level reasoning, and wasting tons of time on this,” he told TechCrunch. “And businesses are only half way there when it comes to coordinating and converging data streams.”

Ouster’s new sensor changes that equation, he said.

“The goal is to eliminate the need for cameras. There’s no reason why one sensor can’t perform both functions,” he said.

The Rev8 lineup comes at a dynamic moment for lidar companies. There has been a wave of consolidation over the years, including Auster’s acquisition of Velodyne and, more recently, the bankruptcy acquisition of Luminar’s assets.

At the same time, the sensor market is also experiencing explosive growth. Waymo and others have finally introduced working robotaxis and are rapidly expanding. Humanoid and industrial robot companies are raising investment capital and need sensors to perceive the world. Interest in this field is so high that startups like Boston-based Teradar are emerging and experimenting with the field in entirely new ways. (For Teradar, we use terahertz imaging.)

Pacara said color lidar, which combines pinpoint depth information with camera-quality image data, could be particularly valuable to robotics stakeholders. Auster said he worked with Fujifilm and imaging science company DXOMARK to understand “what it means to make a great camera.”

In fact, Pacara claims that Auster’s color lidar is “improved in many ways with its latest cameras” thanks to the way the company already designs and builds its sensors.

Ouster uses a so-called “digital LIDAR” architecture. Instead of analog approaches that require many moving parts, Ouster uses what is known as a single-photon avalanche diode (SPAD) detector to capture LIDAR information directly on a custom chip.

The company uses this same SPAD technology to capture color image data with its Rev8 sensor. Pacara said the new technology allows images to be captured with higher sensitivity than regular cameras.

“48-bit color, 116 dB dynamic range, equivalent to mega pixel resolution. Those are the gold standard numbers that make this a great camera. But it just happens to be delivered as a data stream that is pre-fused as a 3D colorized point cloud,” he said. “You can actually use the data as a camera stream, and that’s one of the strengths of this system, and depending on the forward thinking of the recognition team, you can use just the LIDAR data stream, you can use just the camera data stream, you can use a pre-fused data stream.”

Pacala said the company has already shipped samples to existing customers and is currently accepting orders. He is particularly proud of the OS1 Max sensor, which he believes is “the best long-range lidar in the industry,” he said. It can see up to 500 meters in all directions and is “significantly” smaller than other long-range riders.

“We had long-range LiDAR, but it wasn’t clearly better than anything else,” he said. “This is a huge leap forward for Ouster. I think it means we’ll start to see a lot more of it in high-speed robotic trucking and robotaxis applications. I think a lot of drone-related products will migrate to OS1 Max.”

Other new LIDARs built on the Rev8 platform will include OS0, OS1, and OSDome, according to the press release.

Ouster isn’t the only company to start talking about colored riders. Last month, Chinese company Hesai unveiled its own color lidar platform and said it would enter mass production by the end of this year. Other companies, such as Innoviz, have also pitched their own take on “color LIDAR” in the past.

However, most other players trying to “fuse” cameras and lidar sensors are essentially putting them in a box together, Pacala says. The approach taken by Ouster (and Hesai, to be fair) is to put the lidar and imaging technology on the same chip.

This will significantly reduce the amount of work Ouster customers have to do to understand competing sensor streams, and ultimately lead to those customers avoiding cameras altogether, Pacala said. At the same time, it’s cheaper and smaller than Ouster’s previous technology.

“This is something that fundamentally changes the value proposition of what we sell to customers from this point forward,” he told TechCrunch.