Why native-color lidar is an architectural change
Traditional 3D perception in autonomous systems relies on two separate sensor types: lidar provides spatial structure (a point cloud with distance measurements), while cameras provide colour and texture. Fusing these data streams requires external calibration and temporal synchronisation — computationally expensive and error-prone under fast motion or variable lighting.
REV8 unifies both functions in a single sensor. The L4 Ouster Silicon chip integrates Fujifilm colour science and processes colour data on-chip, with hardware-enabled HDR. As Ouster describes it, every point is "born" with colour — spatial and visual data are captured simultaneously, with zero latency between geometric and colour signals.
Technical specifications reflect the scale of the change: 48-bit colour depth, 116 dB dynamic range, and operation across a 1 lux to 2 million lux range. This means full perceptual capability in both darkness and direct sunlight — conditions critical for robots operating outdoors and in transport infrastructure.
CEO Angus Pacala commented: "Perceiving the world in full context requires a combination of structure and colour. Rev8 is the first sensor to unify both. For the first time, a single sensor can understand road signs, interpret brake lights, or simply capture the richness of planet earth in survey-grade, colorized maps."
Technical specifications of the REV8 family
The REV8 family includes completely redesigned OS0, OS1, and OSDome sensors, plus the new flagship OS1 Max. All are based on L4 (128-channel) or L4 Max (256-channel) silicon.
The OS1 Max is the most capable sensor: 256-channel architecture with range up to 200 m at 10% reflectivity and maximum detection range of 500 m. With a 45° vertical field of view, the sensor targets long-range applications: high-speed autonomy, smart infrastructure, and heavy industrial use.
The L4 chip processes up to 10.4 million points per second and transmits 22.4 Gbps of data off-chip. A 40 kHz measurement rate with picosecond timing precision means REV8 can capture dynamic scenes with minimal motion artefacts.
All REV8 sensors are automotive-grade, with ISO 21434 cybersecurity certification and compliance with ASIL-B (ISO 26262), SIL-2, and PLd functional safety standards. Ouster declares a planned 10-year production life — significant for customers planning high-volume deployments.
The StereoLabs acquisition and platform strategy
In February 2026, Ouster acquired StereoLabs — a vision-based perception provider — for $38 million. The transaction signals a strategic shift: from pure-play lidar vendor (as Pacala himself described the previous model) toward a perception platform combining lidar and cameras.
Pacala was direct: "For 10 years, Ouster was a pure-play lidar vendor. The acquisition is to build a systems or platform business." He also emphasised that cameras and lidar are complementary for his customers, not competing: "It's a matter of using the right sensor for the job."
This approach differs from NVIDIA's strategy, which integrates AI models with perception software within the Isaac ecosystem without proprietary sensor hardware, and from companies such as Luminar Technologies, Innoviz, or Hesai that remain lidar-only. By combining its own silicon chip with StereoLabs stereo cameras, Ouster is building vertical integration from sensor to AI model inputs.
Customers and applications: from robots to infrastructure
Ouster named dozens of companies intending to deploy REV8 sensors: Google, Volvo Autonomous Solutions, Liebherr, Epiroc, Field AI, Flyability, Skydio, PlusAI, Constellis, Bedrock, Kässbohrer, Third Wave Automation, Burro, Seegrid, Gecko Robotics, Pratt Miller, AIM Intelligent Machines, Cyngn, Freefly Systems, ATI Robotics, and SwarmFarm.
The list shows clear sectoral diversification: autonomous heavy vehicles (PlusAI, Volvo), inspection drones (Flyability, Skydio), mobile logistics robots (Third Wave Automation, Seegrid, Burro), security systems (Constellis), infrastructure inspection (Gecko Robotics), and agriculture (SwarmFarm).
Pacala framed the market broadly: "Physical AI is something with the AI component in the software layer, fed by a suite of advanced sensors and being run on a computer like an NVIDIA edge GPU. We're selling picks and shovels — a family of products across a variety of platforms. The application could be on a traffic light, a humanoid robot, a car, or a delivery robot."
Why this matters
Native-color lidar addresses a problem that autonomous perception systems have faced for a decade: how to unify geometric and visual data without calibration compromises. Existing solutions — camera plus lidar with software fusion — are effective but complex. REV8 moves fusion to the level of sensor physics, reducing latency and eliminating calibration errors.
For companies building physical AI models, training data quality is directly tied to sensor quality. Ouster emphasises that REV8 can collect data for both training and production deployment — the same sensor used at prototype and at scale. This reduces costs and simplifies the data pipeline for companies such as Google and Skydio.
Automotive certifications and a 10-year production lifecycle signal that Ouster is positioning REV8 not as a niche product but as infrastructure for mass autonomy deployment across the 2026–2035 decade.
What's next?
- REV8 OS sensors are available to order now and shipping in Q2 2026.
Tracking early deployments by Boston Dynamics and Google will indicate how native colour affects perception model performance in real-world conditions.
- The planned ASIL-B certification (full ISO 26262) has not yet been finalised — its timeline is critical for automotive customers.
Sources
- The Robot Report
Ouster releases REV8 OS sensor family with native-color lidar - https://www.therobotreport.com/ouster-releases-rev8-os-family-native-color-lidar-sensors/ - Ouster
Rev8 OS downloads and specifications - https://ouster.com/downloads - Ouster
ouster.com - https://ouster.com/
