Intel® Robotics Open Source Project (Intel® ROS Project) to enable the object detection, 2D location, 3D location and tracking with GPU or Intel® Movidius™ NCS optimized deep learning backend, and Intel® RealSense™ camera under ROS framework.

The relationship among ROS packages are: alt text

Installation Prerequisites

Software environment: Ubuntu 16.04 and ROS Kinetic software configuration. 3D camera ROS node. (e.g. Intel® RealSense™ Camera node).

Hardware environment: RGBD camera (e.g. Intel® RealSense™D400 serial cameras).





Object Analytics ROS node is based on 3D camera and ros_opencl_caffe ROS nodes to provide object classification, detection, localization and tracking via sync-ed 2D and 3D result array.


ROS node for object detection backend.


ROS package for object related message definitions.


ROS node for object categorization backend with Intel® Movidius™ NCS stick.


ROS node to publish mark tag of objects based on information provided by ROS object analytics when building map.


ROS node to provide visual inference and network optimizer with Intel® CPU, GPU, VPU and FPGA.

Report a Bug

Use above github projects to report bugs or submit feature requests.

Contributing to Intel ROS Project

The Intel® ROS Project is developed and distributed under Apache 2.0 license. Follow the general ROS Contribute process to submit your code.


1. Intel® RealSense™ 3D camera is recommended as camera input, and the ROS packages are:


Underlying library driver for communicating with Intel® RealSense™ camera.


ROS Intel® RealSense™ camera node for publishing camera data.

2. Intel® OpenVINO™ toolkit open source at dldt. Binary package can be got from OpenVINO toolkit.

3. Intel® ROS2 packages are also provided, please check ROS2 Intel Projects for more details.

Wiki: IntelROSProject (last edited 2019-04-29 06:30:40 by cathyshen)