New color LiDAR sensors give machines human-like vision

For years, machines have navigated the world color-blind. LiDAR sensors – the laser-based eyes of self-driving vehicles, industrial robots, and inspection drones – construct exact 3D maps of their environment, however all the things is constructed of monochrome geometric shapes. Ouster’s new Rev8 sensor household goals to alter that, not by bolting a digicam onto a LiDAR unit, however by fusing coloration straight into each level of information the sensor captures.

Autonomous notion methods typically fall into two camps: camera-only arrays – just like the imaginative and prescient system Tesla makes use of for its underwhelming Full Self Driving tech – or a two-step sensor fusion method with LiDAR for exact geometry, a digicam for coloration, and a software program algorithm to mix them. That stitching course of introduces calibration errors, latency, and spatial mismatches – an issue that turns into essential when a robotic or car is shifting quick via a crowded road.

Rev8 eliminates that structure totally. Every level within the 3D map the sensor generates already carries coloration data in the mean time of seize, with no extra software program processing required. This, Ouster says, makes it the primary LiDAR with native coloration – although it is not competing alone for lengthy.

Only a few weeks in the past, the Hesai Group, the worldwide chief in LIDAR know-how, introduced a full-color platform known as 6D ETX. The Chinese language sensor takes a unique method: somewhat than merely including coloration to the purpose cloud, it captures six full dimensions of information – X, Y, and Z coordinates alongside reflectivity, velocity, and coloration – making it much less a coloration LIDAR and extra a multi-dimensional notion engine.

Ouster’s know-how is constructed round its new L4 chip, which embeds Fujifilm‘s coloration science – the identical experience behind the corporate’s imaging know-how – to ship hardware-level coloration processing. It integrates 42.9 GMACs of processing capability, detects as much as 20 trillion photons per second, and operates at 40 kHz with picosecond-level precision. These numbers are dense, however they imply a single sensor can now learn a site visitors signal, detect whether or not the automotive forward is braking by the colour of its brake lights or produce topographic maps with real-world coloration knowledge – all with out extra {hardware} or calibration.

Ouster

The flagship mannequin of the Rev8 household is the OS1 Max, a 256-channel sensor with a detection vary of as much as 200 m (656 ft) at 10% reflectivity – which means it could actually spot surfaces that take up many of the gentle hitting them – and as much as 500 m (1,640 ft) underneath optimum circumstances. Its area of view spans 45 levels vertically and 360 levels horizontally, and Ouster claims it doubles each the vary and determination of its earlier era, the Rev7.

The sensor handles a formidable unfold of lighting circumstances, from near-total darkness at 1 lux as much as 2,000,000 lux, roughly equal to intense direct daylight. Colour depth reaches 48 bits with 116 dB of dynamic vary – technical shorthand for the sensor’s skill to seize high-quality element throughout extraordinarily shiny and very darkish areas concurrently.

“Rev8 is probably the most superior household of LiDAR sensors ever launched and units a brand new commonplace in sensing,” mentioned Ouster CEO Angus Pacala. “With the L4 Ouster Silicon, we’re delivering on the promise of our digital structure to ship exponential enhancements in efficiency, doubling our core specs and concurrently introducing the world’s first native coloration LiDAR to offer machines 3D human-like sight for the subsequent period of Bodily AI.”

Ouster

Early adopters embody Google, Volvo Autonomous Options, Skydio, PlusAI, and Seegrid, amongst roughly two dozen corporations throughout robotics, automotive, and sensible infrastructure.

The launch additionally displays a broader strategic shift. In February 2026, Ouster acquired StereoLabs – a pc imaginative and prescient specialist – for US$38 million in money plus 1.8 million shares, signaling a transfer away from promoting standalone sensors towards providing a full notion platform. When coloration and geometry are fused on the supply, coaching AI fashions turns into considerably simpler.

“Rev8 is the foundational know-how that can enable prospects to maneuver from prototype to business manufacturing at scale, offering the reliability and affordability required to allow real-world autonomy throughout industries,” added Pacala. Rev8 sensors can be found to order (although we have no phrase on pricing for enterprise prospects), with shipments anticipated to start this quarter.

Supply: Ouster