本课程聚焦ROS 2从笔记本到野外集群的工程化运维，涵盖Docker多阶段构建、CI/CD加速、ARM交叉编译、systemd自愈及A/B分区OTA回滚，助你安全管控成百上千台机器人生产堆栈。
原始标题：ROS 2 Robotics DevOps: Docker, CI/CD and Fleet Deployment
本课程专注于解决 ROS 2 从“笔记本运行”到“野外机器集群量产部署”之间的工程化与运维（RoboOps）痛点。课程采用完全数据驱动、实操演示的方式，深入讲解如何通过多阶段 Docker 构建与层缓存优化将镜像体积和编译时间缩减近半，并解决容器间 DDS 网络的通信发现难题。同时，涵盖了在 CI 阶段利用缓存加速底层编译，以及在无机器人硬件时，通过单元测试、launch_testing 和 ros2 bag 进行仿真回归测试的完整闭环。
针对嵌入式端与现代化机队管理，课程深入解析了 ARM/Jetson 平台的交叉编译成本对比，并展示了如何利用 systemd、硬件看门狗与守护进程确保机器人掉电重启后的系统自愈能力。最核心的价值在于其工业级的远程运维方案：通过 A/B 双分区或安全的 OTA 机制实现自动验证与失败回滚，防止远程升级导致机器人“变砖”；同时搭建起云端指标与日志监控网络，配合动态密钥与配置管理，让工程师无需亲临现场即可高效、安全地管控成百上千台机器人的生产堆栈。
Published 10/2026
Created by Frank Robotics Lab
MP4 | Video: h264, 1920×1080 | Audio: AAC, 44.1 KHz, 2 Ch
Level: Intermediate | Genre: eLearning | Language: English | Duration: 89 Lectures ( 6h 55m ) | Size: 7.4 GB
Containerise ROS 2, build it in CI, cross-compile for ARM, boot it with systemd, update a fleet and roll back safely.
What you’ll learn
⚡ Write a ROS 2 Dockerfile that builds a real colcon workspace, and know why each line is in that order
⚡ Cut image size and rebuild time with multi-stage builds and correct layer ordering, measured not guessed
⚡ Run a multi-node stack with Compose, and get DDS discovery working across containers
⚡ Build and test ROS 2 in CI, including caching the underlay and reading a failed run
⚡ Test nodes with no robot attached: unit tests, launch tests, faked sensors and bag regression
⚡ Cross-compile for ARM and Jetson, and know the real cost of QEMU versus native runners
⚡ Start a robot stack at boot with systemd, with restart policy and watchdogs that behave on a power cut
⚡ Ship an over-the-air update to a fleet, verify it, and roll it back automatically when it fails
⚡ Get metrics, logs and alerts off a robot so you can tell which one is broken without driving to it
⚡ Manage per-robot config and secrets across a fleet without hand-editing machines
Requirements
❗ You can write and run a basic ROS 2 node in Python or C++
❗ Comfortable on the Linux command line
❗ Docker installed; everything else is installed on screen
❗ No robot hardware required: the whole course runs in simulation and containers
Description
This course contains the use of artificial intelligence.
Most ROS 2 courses stop at the point where the code runs on your laptop. This one starts there. It is about everything between a node that works on your machine and a fleet of robots running that node in the field, where you cannot plug in a keyboard and a bad update means a van journey.
Nothing here is a slide of advice. Every claim is run, measured and shown. You will watch a naive ROS 2 image go from897.5 MB to 492.1 MB with a multi-stage build, see a cold build drop from21.55s to 9.35s, and see why one line in the wrong order turns a1.79s rebuild into a9.39s one.
What the course covers, in nine sections
✨Containers for ROS 2 – Dockerfiles that build a real package, layer caching, multi-stage builds, base image choice, tags and digests
✨Compose and multi-node systems – several nodes as one stack, DDS across containers, host vs bridge networking, domain IDs, recording bags
✨CI for ROS 2 – colcon in CI, caching the underlay, matrix builds, linting, build and push, branch protection, reading a failed run
✨Testing – unit tests with no robot, launch testing, faking sensors, bag regression, timing and deadlines, flaky tests, coverage
✨Cross-compiling – x86 to ARM, buildx and QEMU, measuring the penalty, native runners, multiarch manifests, Jetson and CUDA
✨Boot and systemd – unit files, restart policy, start order and hardware, logs and disk, device permissions, watchdogs, resource limits
✨OTA and rollback – pull-based updates, staged rollout, health checks, automatic rollback, updating mid-mission, migrations, signing
✨Fleet observability – which robot is broken, metrics off the robot, Prometheus exporters, dashboards, alerting, log shipping, bag triage, remote access, post-incident
✨Multi-robot – one robot to a fleet, Open-RMF, fleet adapters, per-robot config, secrets, fleet-wide upgrades, what it costs
What this course is not: it is not a Docker fundamentals course that happens to mention robots, and it is not a slideshow of best practices. If you have never written a ROS 2 node, start elsewhere and come back.
Who this course is for
⭐ ROS 2 developers whose code works locally and now has to run on a real robot
⭐ Robotics engineers who have been handed deployment and have no DevOps background
⭐ Teams moving from one robot to a fleet and hitting update, config and observability problems
⭐ Backend or DevOps engineers moving into robotics who want the ROS 2 specifics
此处内容需要权限查看
会员免费查看
