Over The Edge: abusing "toy" hardware in new (wet) ways

Someday, the hardware police are going to come and take away my toys, citing me for criminal abuse. Until then, I'll continue to see what an NXT can handle... and what I can learn in the process.

On a recent trip to Niagara Falls, I was fascinated again with this huge, impressive natural feature. And thinking about the foolhardy souls that over the years have tried to survive going over the falls in a barrel, I did what was natural for me - thought about the NXT doing it. Of course, getting permission to throw something over Niagara Falls is unlikely to happen... and it's dangerous, and retrieving it would be extremely difficult. But I could certainly do some test runs... so was born the "NXT in a barrel".

I used an NXT slid into a 4" PVC pipe segment, sealed tightly at either end with removable closeout plugs. To keep it from bouncing around inside, I used a hot-wire cutter to shape foam "bulkheads" to support and lock the NXT into place within the tube, and attached a 50' line to retrieve it after. Then I went to a small local dam, started it, sealed it tightly, took a deep breath... and threw it over.

The NXT logged two sensors 30 times a second, a 3-axis accelerometer from Mindsensors, and the LEGO sound sensor. Above the dam it floated tranquilly, quietly, and approached the edge. During the fall, the sound level went up only a little, but the accelerometer clearly registered a freefall as it went over the edge. Then there's a sharp acceleration as the pipe hits the rocky bottom end-on, and the noise level suddenly spikes up, as water crashes down on top of the PVC pipe and the unit is tossed and turned at the sloped, rocky base of the falls. Then suddenly it quiets down, bouncing in the tailrace in the fast-moving water below the waterfall.

All in all, this gave me yet another chance to "live vicariously" and understand something that I have no (safe) way of experiencing myself. Could I scale this up? Certainly (& I'm planning to, in a couple of different ways). Could this be used as a teaching tool? Absolutely, in so many ways. Total cost beyond the NXT was maybe $15 total for the plumbing, and programming was very easy.

Niagara Falls it's not... but it's as close as most of us are likely to get. Now excuse me, I've got to find a bigger waterfall...


222Doc said…
LOL Brian taking it to limit, hey Niagara! I can see it soon on the News, man in custody for sending a unmaned unknown devise over the falls today. Home land security is investigating this device as the HLS robot recovers it from the deep.
Brian Davis said…
Which is exactly why I *DIDN'T* do this at Niagara, and why I *WILL NOT* encourage people to do this at Niagara. Safety first... which includes following local laws. :)
I_Fixed_It_44 said…
Wow,Brian Your Really Pushing the Limit.0Next You Should Make A Submerine
I knew as soon as I saw the picture of the pipe bobbing in the water who posted this article. Scrolling further down only confirmed it. You crack me up Brian.
Martijn H. said…
O my... what ways of MINDSTORMS-(ab)use haven't you tried?
Were do you keep getting these ideas? =)
No video? such a shame...
although I'm not sure I can handle the picture of an NXT going over the edge...Horror:D
Good job!
Jordan Gray said…
I can't figure out how to register in the NXTstep forums... When I click the Register link (http://thenxtstep.com/smf/index.php?action=register), it just tells me "An Error Has Occurred!
Sorry, registration is currently disabled."

When/How can I make an account?
AS said…
Very cool thing you did there.

I don't think I fully understand your graph though. You plot 'x fore/aft' and 'Net Accel'. I assume this data is somehow computed from the 3 axis accelerometer data, but could you give some more insight in what they exactly are and how they are computed?
Brian Davis said…
Martijn: Actually, the picture of Annie "on the brink" is a single frame from a video, but it really doesn't show much, as it's hard to follow a white cylinder as it bounces unpredictably down white water. As to what I haven't done as yet... I'm not out of ideas yet :).

AS: You're completely correct, it's a 3-axis accelerometer. To get the net acceleration (dark blue dataset), I take the data from each axis and square it, add it together, and take the square root: Net = sqrt( xaccel^2 + Yaccel^2 + Zaccel^2). That will be a measure of the total acceleration on the robot, regardless of how it happens to be oriented at the time. The "fore/aft" trace is the data from just one axis of the accelerometer - the one that was directed along the long axis of the cylinder. For this particular run, the cylinder went over almost exactly "bow first", and hit the bottom end-on, so the acceleration along that one axis was very distinctive. That didn't happen for every run (I did about 6), but for this run it was a very "clean" signal.

By looking closely at the 3-axis data, you can see when the cylinder rolls as well, but for such short runs and a well-balanced cylinder, that usually doesn't show much. Longer runs in a wave pool (OK, my in a pool making waves) need to be done next.

Popular Posts