Luxonis OAK-D

A stereo depth camera from Luxonis with a built-in AI chip, letting neural networks run directly on the camera.

OAK is Luxonis’s camera product line, with the OAK-D as its most common model: two global-shutter monochrome cameras do stereo depth sensing (a 7.5 cm baseline, with an official ideal range of about 0.8–12 meters), a color camera sits in the middle, a 9-axis IMU is built in, and it connects to a computer over USB. Its distinguishing feature is an onboard compute chip (the RVC2 platform, rated at 4 TOPS, with 1.4 TOPS available for AI), so depth computation and neural networks like object detection can run entirely on the camera, sending only the results to the host — well suited to small robots with limited onboard compute. The newer OAK 4 series switches to a Qualcomm QCS8550 chip (the RVC4 platform, 48 INT8 TOPS) with a 6-core ARM CPU running Linux, letting it operate independently without a host computer. Its companion open-source library is called DepthAI (written in C++, with Python bindings, MIT licensed), with an official ROS driver, depthai-ros, also available.

ExampleMounting an OAK-D on a low-cost mobile robot, the official ROS driver publishes color and depth images to ROS topics while an object-detection network runs on the camera itself, so the host only receives detection results, saving onboard compute.

Also called

OAK 4, OpenCV AI Kit, DepthAI Camera

Related

Depth Camera · Stereo Camera · RealSense Depth Camera (D435i / D405) · Orbbec Gemini 330 Series · Stereolabs ZED · On-Device / Edge Deployment

Sources

Luxonis Docs: OAK-D
Luxonis Docs: RVC4 平台（OAK 4 系列） (Chinese)
GitHub: luxonis/depthai-core

As of

2026-09