Structured ROS2 course vs just building stuff and googling as you go?

Oct 13, 2026 Last reply: 14 hours ago 7 Replies

First reply

RunnerBramble_1984

I went the project route and don't regret it, though I'd tweak how you do it. Two years ago I bought a secondhand TurtleBot clone off a local marketplace and set myself the goal of making it drive to my kitchen and come back. First month was rough, lots of staring at logs and…

Read the full reply

So me and a couple of friends are finally getting around to learning ROS2 (Jazzy) and we're torn on how to start. Part of me wants to just buy a proper online course and follow it start to finish so we don't miss fundamentals. The other part thinks we'll learn faster by picking a dumb little project and figuring things out as we go, even if it's messy.



For anyone who's done a course recently, did it actually hold up? I've seen some that are clearly recorded on older versions and half the commands don't match anymore, which seems rough for beginners who won't know what's outdated vs just broken. Also curious wether people found video courses worth paying for at all, or if the free docs and tutorials were enough once you got past the install headaches. We're mostly intrested in drones eventually, if that changes what path makes sense. whould love to hear what actually worked for people, expecially if you went the self-taught route and don't regret it.


I went the project route and don't regret it, though I'd tweak how you do it. Two years ago I bought a secondhand TurtleBot clone off a local marketplace and set myself the goal of making it drive to my kitchen and come back. First month was rough, lots of staring at logs and rebuilding my workspace because I broke somethig, but that's exactly where the learning happened. A course would never have taught me to actually read an error message properly. That said, I did borrow a course from a friend partway through and honestly the recorded stuff ages fast. The API changes between releases, and even tutorial code on the official docs sometimes lags a version or two. So don't pay for anything until you've checked when it was last updated and whether people in the comments are complaining about broken commands.

For the drone angle, be warned that's a big jump. Aerial stuff adds flight controllers, simulators like Gazebo become basically mandatory since you can't crash-test cheaply, and the safety rules around flying anything with real props are stricter. I'd get comfortable on a wheeled robot or even just the sim first, then look at something like PX4 once the fundamentals click.

If your group wants structure without paying, pick one shared project and assign each person a piece, like one on sensors, one on navigation, one on the launch system. Comparing notes weekly kept me honest when I did this with a buddy.

I did it the other way around and burned a solid month for it. Last spring I jumped straight into building a line-following robot with a couple of buddies, no structured learning at all. We spent most evenings fighting build errors and YAML files we didn't understand, and the worst part was never knowing if a tutorial was wrong or if we were. We nearly quit twice. Then one of us quietly worked through a paid course over two weeks of lunch breaks, came back, and suddenly our project made sense. The difference was night and day. For your situation I'd say the course matters even more because of the drones part. When something goes wrong with a flying thing, you can't just unplug it and poke around, so you want the fundamentals solid before anything leaves the ground. Two things I'd check before paying: make sure the course was recorded or updated for Jazzy specifically, and look at the date on the most recent lesson, not the sales page. Older courses recorded on Humble or Foxy will waste your time on install steps that changed completley. The free official tutorials are genuinely good, but they read like reference material, they assume you already know what you're looking for. A course that walks you through in order fills that gap. Do both eventually, but start with structure.

Self-taught here, and the thing that actually got me unstuck was stumbling on Josh's Articulated Robotics video series where he builds a whole robot from scratch in sim. Watching someone set up a workspace and debug live was the missing piece, docs alone weren't doing it for me. I've never owned real hardware, just ran everything in Gazebo on a laptop, and by the time I wanted to try drone stuff I alredy had the basics down. The paid course I sampled later felt slow compared to what I'd already pieced together. So yeah, messy project plus one good free video series worked for me.

Hybrid approach worked for me. I run a small automation lab at work, maybe 6 robots on the bench at any time, and when we moved to ROS2 I made everyone do the official beginner CLI and client libraries tutorials first, about a week each, then threw them straight at a real task. The tutorials are dry but they cover the stuff you'll waste days guessing at, like how nodes and topics actually fit together. One thing I'd add for your group: pick a version and stick to it. Half our early headaches came from mixing instructions written for diffrent releases. Once you're past that, the free material really is enought, paid courses mostly just repackage it.

Okay so when you all say you ran everything on a simulator, what does that actually look like? Like is it a fake robot moving around on your screen, or do you still need some real parts plugged in? I tried one of the free starter tutorials last weekend and my old laptop chugged so hard on the 3D window that I gave up and just read the text instead. Wondering if I need a better machine before I even start, or if that's just how it is for everyone at first.

Started on ROS1 back when the only options were a printed book and praying the community wiki hadn't rotted, so any of you kids complaining about outdated videos would've wept. What I'd add is don't sleep on the Q&A archives. Nine times out of ten your wierd launch file error was asked and answered years ago by somebody grinding through the same thing. Also, for drones later on, spend time now learning to read node graphs and inspect topics from the command line. When a drone misbehaves in sim you'll be living in those tools, and there's no video course that does that digging for you.

Update: we took the hybrid idea from the lab manager and ran the official beginner tutorials first, about a week in and honestly less painful than expected. Also started watching the Articulated Robotics build someone mentioned, seeing the debugging happen live is clicking in a way the docs weren't. Question though, for the project-first folks, did you set a hard deadline for when to stop fiddling and ship something ugly?

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required