# Canonical Workshop Environments [Canonical Workshop](https://ubuntu.com/workshop/docs/) creates reproducible development environments as LXD system containers. The Robotics AI Suite provides a Workshop definition for each component: one command creates a container with the suite and that component installed, and passes the host's Intel® GPU, Intel® NPU, cameras and desktop through to it. Use workshops to: - Try a component without installing the suite on the host. - Run ROS 2 Jazzy (Ubuntu 24.04) and ROS 2 Humble (Ubuntu 22.04) side by side on one Ubuntu 24.04 host. - Connect a ROS 1 Noetic application to the suite through `ros1_bridge`. - Keep each component's dependencies separate, and rebuild an environment from scratch at any time. Inference in a workshop runs at the same speed as on the host: in the [Object Detection](object-detection.md) tutorial, OpenVINO™ throughput in the workshop was within run-to-run variation of the host's on the CPU, GPU and NPU. ## Available workshops Every ROS 2 workshop has a Jazzy definition, ``, and a Humble definition, `-humble`. | Tutorial | What it runs | Extra hardware | Jazzy | Humble | |---|---|---|---|---| | [Object Detection](object-detection.md) | OpenVINO™ object detection on CPU, GPU and NPU | - | Yes | CPU and GPU | | [Wandering](wandering.md) | Autonomous exploration and mapping in simulation | - | Yes | Yes | | [ITS Path Planner](its-planner.md) | ITS global planner for Nav2 in simulation | - | Yes | Yes | | [FastMapping](fast-mapping.md) | 3D and 2D mapping from a recorded depth bag | - | Yes | Yes | | [Groundfloor Segmentation](groundfloor.md) | Ground and obstacle segmentation from a recorded depth bag | - | Yes | With a workaround | | [ADBSCAN](adbscan.md) | Obstacle clustering from a recorded laser scan and point cloud | - | Yes | With a workaround | | [Collaborative SLAM](collaborative-slam.md) | Two visual SLAM trackers with GPU feature extraction | - | Yes | Yes | | [RealSense™ Camera](realsense.md) | A live RealSense™ camera through the camera interface | One RealSense™ camera | Yes | Yes | | [Multi-camera AI Demo](multicam.md) | YOLOv8 on two RealSense™ cameras at once, on the GPU | Two RealSense™ cameras | Yes | Yes | | [Robotics System Profiler](benchmark-framework.md) | ROS 2 KPIs in Grafana, monitoring another workshop | - | Yes | Yes | | [ROS 1 Bridge](ros1-bridge.md) | A ROS 1 Noetic camera, `ros1_bridge`, and detection on Jazzy | - | Yes | - | All workshops were tested on an Intel® Core™ Ultra X7 358H processor with Intel® Arc™ B390 graphics and robotics-installer v2026.3.12. The results in these tutorials are examples from that system. :::{note} Individual results may vary. ::: ## Where to start ::::{grid} 2 :::{grid-item-card} **Getting Started** :link: getting-started :link-type: doc :link-alt: clickable cards Install LXD and Workshop, check the GPU and NPU in a container, and launch your first suite workshop. ::: :::{grid-item-card} **How Workshops Work** :link: how-it-works :link-type: doc :link-alt: clickable cards SDKs, device interfaces, Jazzy and Humble pairs, and how to add a workshop for another component. ::: :::{grid-item-card} **ROS Distributions and Interoperability** :link: ros-distros :link-type: doc :link-alt: clickable cards Humble workshops, mixing Humble and Jazzy on one ROS domain, and reaching workshops from the host. ::: :::{grid-item-card} **Troubleshooting** :link: troubleshooting :link-type: doc :link-alt: clickable cards Fixes for device, display, proxy, storage and ROS problems in workshops. ::: :::: The workshop definitions are in the repository next to each component's source, in `src/components//workshop/`. The shared parts, and the workshops without a component of their own, are in [`src/workshops/`](https://github.com/open-edge-platform/robotics-ai-suite/tree/main/src/workshops). :::{toctree} :hidden: Getting Started How Workshops Work Object Detection Wandering ITS Path Planner FastMapping Groundfloor Segmentation ADBSCAN Collaborative SLAM RealSense Camera Multi-camera AI Demo Robotics System Profiler ROS 1 Bridge ROS Distributions and Interoperability Troubleshooting :::