Skycam assembly gallery and robot debugging

I sold a record number of Drawbots today. Awesome possum! To celebrate, here’s a little bit more about the Skycam I’m going to be bringing to the Vancouver 2012 New Forms Festival, September 13-16. The Drawbot will also be there.

I assembled the Skycam a few days ago. It’s two Drawbots put together with a bit of modified software on top. I assumed I would set up four steppers around a clock face. Each stepper would be at 1:30, 4:30, 7:30, and 10:30. An X+ movement should be towards 3:00 (Berlin), a +Y towards 12:00, and a +Z towards the ceiling. Sounds do-able, but it’s a whole new world of problem solving.

I knew from the start the steppers would be mounted backwards or something. I wrote some code to “jog” just one motor so I can identify which is which and then switch their positions in software. Each motor gets assigned an XYZ coordinate. When my Java program says “Go to XYZ” the Drawbots can figure out the rest on their own.

I also knew that I had to wait for a movement to complete before starting another or accuracy would suffer. That was challenging, and the first time I wrote Listener pattern in Java. Thank you to everyone in freenode#java for their help!

Re-centering a four-string device is a lot harder than a two-string device. I put the origin in the middle of the floor. I still have to manually turn the bobbins which is my least favorite part.

I broke the strings three times already. Lesson learned: run the Skycam with the emergency stop in my lap. Also, disconnect three of the four strings when I’m done with the robot for the day so people can’t get snagged on it by mistake.

So how’s the actual movement, when everything else is said and done? Well…so far the results I get are really confusing. They have to be seen to be described. Anyone interested in a video? Comment now or forever hold your peace.

For a complete description of the build process, read the descriptions and titles of each picture in the gallery. I make no apologies: I have Burning Man on the brain.

If you’d like to build one of these Skycams, let me know. I can package up all the parts if you can get the 2x4s and the drywall/wood screws.

Tags: , , , , , , , , , , , , ,

12 Responses to “Skycam assembly gallery and robot debugging”

  1. Evan Jones says:

    I’d love to see video! The original Hektor.ch had horribly slow motion planning, but ended up with really smooth paths that minimized pendulum issues. I’m curious if you think an approach like that would help or if your acceleration code takes care of most of that already.

    Also, any thoughts about a center weight with bump sensors for self calibration. Seems like it would make setup a lot easier if you could just toss the four motors out wherever and have a calibration routine figure out where each was. I’ve done some preliminary math for a 3-motor version, but haven’t followed up on it much.

    • Dan says:

      actually, i’m working on a new motor mount that will incorporate the switches needed for self-calibration on both the drawbot and the skycam. What I really need is a $500 mendelmax and a $2000 copy of solidworks.

      The drawbot runs at about 2500mm/minute with very fast turns and no pendulum problems to speak of. I didn’t have to write any fancy motion code, either. I tried that for about 6 weeks and it was a dead end that just made things more frustrating.

      Three steppers sounds easier to calibrate, but how often do people use triangular rooms?

  2. [...] With the Olympics on there are a lot of really great camera shots shown during the events. One of the best is the overhead view, which is provided by a camera suspended between cables. It’s not new for the Olympics, SkyCam has been around for over twenty years. What is new is [Dan Royer's] attempts to build his own aerial camera setup. [...]

  3. [...] With the Olympics on there are a lot of really great camera shots shown during the events. One of the best is the overhead view, which is provided by a camera suspended between cables. It’s not new for the Olympics, SkyCam has been around for over twenty years. What is new is [Dan Royer's] attempts to build his own aerial camera setup. [...]

  4. [...] With the Olympics on there are a lot of really great camera shots shown during the events. One of the best is the overhead view, which is provided by a camera suspended between cables. It’s not new for the Olympics, SkyCam has been around for over twenty years. What is new is [Dan Royer's] attempts to build his own aerial camera setup. [...]

  5. [...] With the Olympics on there are a lot of really great camera shots shown during the events. One of the best is the overhead view, which is provided by a camera suspended between cables. It’s not new for the Olympics, SkyCam has been around for over twenty years. What is new is [Dan Royer's] attempts to build his own aerial camera setup. [...]

  6. Marshall says:

    I was once lucky to see a skycam up close and my father was able to see and use the controls. the mounting of the cables on the SkyCam is quite different than what you have. The SkyCam has and central structure that has a square near the middle or top. the cables connect to the corners of that square. Also they use flight cable. high tensile aluminium or steal cable. Usually the first. Also the control system is extremely complex. There are 3 operators. You have one person on the camera control. They control the zoom, pan and tilt, focus. The second operator controls the mount on which the camera is on. the left, right, up, down. the 3rd operator is the computer system itself. there are sensors on the mount itself and tracking sensor near the spools. the computer overrides the mount control if a obstacle is detected. The camera control is fairly easy, the mount is the hard part. to move right, two spools pull in while the other two release. Going diagonal you then have 1 spool pulling in and 3 releasing. For up and down that is easy. its either all pull in or all release. Hope that helps you out

  7. Marshall says:

    One thing I forgot to put in there. The way the cables are done. The spools have the cable exit point aimed at the center of the field. The cables also come out like that of a fishing real.

  8. [...] I found a skycam project (featured on hackaday) created by Dan Royer from Marginally Clever and a fellow member of [...]

  9. James Newton says:

    You might be interested in this 3D printed computer controlled winch:
    http://www.thingiverse.com/thing:161157

  10. justkristin says:

    […] you can build one. […]

  11. […] you can build one. […]

Leave a Reply