Jan. 23rd, 2022

Pixar Lamp video part 5: hello 6DoF + under-actuated.

Why deal w/ friction when you can deal w/ air resistance instead?


Instead of fully going back to the drawing board — we had a long conversation with my old suite-mate Ronak last night and we came up with a few ideas that I’ll list here before picking one to do.

  • First, thought the bearings might be going the wrong way. If we rotate the bearings 90 degrees and extend the lever arm, we may be able to get the lamp to stand “straighter” and also ride smoother depending on the spacing of the bearings. 

  • Second, we thought deeper groove v-bearings might help keep the cable inside the bearing’s path.

  • Adding a second ratchet to increase the overall system tension in both sides since due to friction the right side is higher in tension than the left.

But we came up with another idea because a question was posed “would the lamp only work in 1D or would it work in 2D too?” Meaning could we somehow get the lamp to jump both horizontally and vertically? This was an interesting question because it might be harder than getting the lamp to jump in 1D, but it would be so much cooler so if we’re re-designing a bunch of things, we might as well re-design it to be cooler! So here’s the initial thoughts on what we need.

  • A linear rail with a carriage or some sort of zipline cart that can travel along horizontally

  • Bungees hanging from the rail to the lamp

  • Some sort of control on the horizontal motion

  • A much grippier foot 

So we haven’t fully worked out that plan yet but what we do know is we can test the four bungees from the ceiling because that was one of our original ideas except this time we have a slightly better idea of how to do that. Let’s just neglect friction this time (because that went so well the last time).


So here's we're going to try temporarily, we're going to hang four bungees from the sprinkler pipe and the tie downs we installed and tie para-cord between them and the lamp itself. There's four and if we tension them properly we might just get enough of a constrained up and down motion with springiness for this to work properly.

The goal is to test the bungees stationary and if this works well we can move onto the linear rail that lets it move in 2D.

Here's an initial test of the four-bungee bungee system instead of the cable linear rail. The biggest concern with this was the constraining of the lamp to move up and down without meandering too much to each side or changing its angle. There's also some concern with the resonances that can occur between the four springs the mass and the impedance controller.

But based on this video, it actually seems to be working ok and this is with minimal thought given to tensioning. We might need a slightly more complex control scheme but for now this seems pretty ok as a start! Welcome to under-actuated robotics.

Looking at the video it seems like we can estimate the lamp-bungee resonant frequency to be around 1Hz (definitely need a more accurate number so we can put a timer on the screen next time). The next steps would be the following.

  • Fix the tensioning in the system just a little

  • Put a block under it and try to get the foot to drive it at resonance and measure the response

  • Add some rubber pad to the foot

Here's a better video we can use to estimate the resonant frequency of just the lamp and bungee, we can see that there are three oscillations between 1:16:59 and 1:19:45 which is near as makes no difference to 3 seconds (it's 2.9 something) which means 1 cycle/second or 1Hz was actually a fairly accurate guess. This video was recorded just on an iPhone slow-motion camera and then the timing is just from the google stopwatch at full screen in the background (thanks to Elijah in my lab for showing me this technique). 

Now let's plug in a large 6s lipo (~24V) and get it to jump! We're abandoning the Dark Power Pro supply because it hasn't been working since we went unstable. Our guess is we regen'ed too hard into it or something.