Autonomous mapping drone

How hard would it be to build a drone that can autonomously map an indoor space?

1 Like

A bit harder than building one that can semi-autonomously map an indoor space.

4 Likes

In addition to “drone,” you would need…

  1. Some method of obstacle avoidance and station-keeping, either LIDAR or ultrasonic
  2. Some method of recording the drone’s position at a better resolution than your intended map
  3. Some method of scanning / mapping the space - a scanning laser is pretty popular

This is on top of a controller to fly flight plan… Which, you haven’t mapped yet, right?

There was an article recently (last 6 months I think) in one of the print drone magazines (Drone 360 or Rotor Drone) about scanning/mapping drones that might give you some starting points.

Why a drone? Seems like a ground-based robot might be better, since it eliminates a lot of the navigation issues, especially the “can I bump into something” issue.

1 Like

Much better visibility. No chairs, desks, file cabinets, trash cans, or other obstacles to try and interpret.

You could do this by building a rover with LIDAR that can roll around will give you a bunch of mapping data. I think you would need a navigational estimator to help you localize the robot given the LIDAR data, and then use that to help process the LIDAR into a 2D-ish map.

https://www.amazon.com/Probabilistic-Robotics-Intelligent-Autonomous-Agents/dp/0262201623/

Incredibly hard is the short answer. Though easier than it was five years ago.

thanks if you remember the title will you put it in here?

Mapping drones didn’t work out so well in Prometheus…

1 Like

that is what I want to build!

Making a push cart with cameras and doing the mapping with fiducials and photogrammetry is far more achievable with out a few spare PHDs in your back pocket. That said it would still be really hard and require alot of manual work to get it look right.

https://developer.valvesoftware.com/wiki/Destinations/Advanced_Indoors_Photogrammetry

everybody is a debbie downer!

I feel like I’ve seen roombas setup for this specific purpose. And they do it to learn the home layout to optimize cleaning instead of just bumping into the room every time.

You’re kidding, I know, but… when you start with “how hard would it be to ____?” people tend to answer with how hard. And if building an autonomous mapping drone were easy, DJI would have one in Best Buy for $999.99.

How about posting some more info about what you are envisioning, because I think it’s a cool project.

  • What type of spaces are you looking to map? You say “indoor” but are we talking homes, office buildings, stairwells … lots of things are “indoor”
  • What type of resolution/accuracy are you looking for in your map?
  • How autonomous? Like, put drone inside door and hit a green “go” button on your phone? Or is a little bit of preparation/intervention allowable?
  • What’s your budget?

Honestly, if you’re talking about human-made structures, a tracked vehicle with a rotating laser scanner is probably the easiest, most robust solution, with the longest run-time. You’ll have plenty of challenges with navigation and “not bumping into stuff.” Fancy it up by putting the scanner on a telescoping mast, or add cameras to build a visual skin to put on the scan.

I think the higher-end roombas do this. Our mid-range roomba is a moron.

1 Like

It’s super easy. Why haven’t you done it yet?

I want to be able to place a drone down press a button and it start flying around until it maps all open spaces, then return home and shut off, save the data to a SD card and then use some kind of stiching software to create the map.

great video thanks so do you want to build the drone with me?

Frank, this was a fantastic link, thank you.

I think this is doable. It’s going to require a lot of research and time but doable. If you decide to do it I would like to help.