ZED 2I (RGB-D Camera)
General Description
The ZED 2i camera, developed by Stereolabs, is a high-performance stereo camera designed for 3D depth perception and spatial computing. It features dual 4K stereo cameras, an onboard IMU (Inertial Measurement Unit), and a depth sensor, providing accurate depth maps, object tracking, and spatial awareness.
It provides a range of advanced features, such as real-time 3D mapping and visual SLAM (Simultaneous Localization and Mapping), enabling the camera to track and map surroundings with minimal external hardware. It can detect objects, recognize surfaces, and understand spatial relationships within a scene, which is highly valuable for autonomous navigation and interaction.
The module has been developed for the ZED 2I camera deployment in two different architectures: x86 and arm64 (for Jetson devices).
| Resource | Link |
|---|---|
| Source code | Link to source code |
Contact
The following table includes contact information of the main developers in charge of the component:
| Name | Organisation | |
|---|---|---|
| Ander Gonzalez | ander.gonzalez@ikerlan.es |  |
| Daniel Sedano | dsedano@ikerlan.es | |
License
Open source - BSD
Technical Foundations
WIP
Integrated and Open Source Components
Overview
The deployment of the ZED camera and its data acquisition required the usage of the GPU of the unit that the ZED camera is connected to.
This module requires the combination of NVIDIA drivers alongside the proper CUDA toolkit and the ZED ROS2 wrapper for the deployment of the camera.
The CUDA toolkit and the ZED 2 wrapper are available inside the Docker container that deploys the ZED 2I camera drivers, but the host that uses this container must have installed the proper NVIDIA drivers for its correct usage.
Pre-existing Components
NVIDIA drivers
Source
Description
The CUDA version that is used in the module is 12.1. In order to use it correctly, NVIDIA drivers 525.x or above must be available in the system for a suitable compatibility.
These drivers are software packages that will enable the communication between the NVIDIA GPU and the operating system of the computer.
Purpose in AI-PRISM
These drivers are required for the communication of the ZED 2I camera and its deployment in a robotic environment for the data acquisition of the surroundings of the robotic application.
License
NVIDIA CUDA toolkit
Source
Description
NVIDIA CUDA is a parallel computing platform and programming model that allows the usage of a NVIDIA GPU for general-purpose computing tasks.
CUDA provides software tools, libraries and APIs that enable the acceleration of the workloads of applications.
The CUDA version employed for the ZED deployment is 12.1.
Purpose in AI-PRISM
The CUDA toolkit is required by the ZED camera for its deployment and in the data extraction and processing of the robotic environment.
License
ZED ROS2 Wrapper
Source
Description
The ZED ROS2 Wrapper provides the required packages for the integration of the control and data acquisition performed the Stereolabs ZED cameras with ROS2.
Purpose in AI-PRISM
The purpose of the component is to integrate the data acquisition performed by the ZED 2I camera with the ROS2 framework for its further processing.
License
How to install
WIP
How to use
Once the component has been installed and deployed, the module will publish ROS2 topics providing RGB and depth information captured by the selected camera.
These topics shall be used by other componentes deployed in the cluster for the sensing of the surroundings of the robot and further processing.