Cardboard Hovercraft Robot

Quite the 180 from my last post, due to a lot of things finally going my way.

Before I get to the new superpower I've acquired, let me touch base on the fabric. I decided to try a different fabric, something called ripstop nylon, and man is it awesome. It's thin, light, flexible, but at the same time strikes me as being tough. For reference, this is the same (or similar) fabric that used in kites and camping tents (I think). Definitely much better than the previous stuff.

Onto the sewing. Out of options for binding the edges of the hovercraft's skirt pieces, I finally decided to learn how to sew. Fortunately, I managed to find someone at Techshop who was willing to show me the basics.

Y'know when you picture how hard something is in your head, simply because you don't know about it at all? Yeah, that was sewing for me. After I actually started, not only did I pick it up quickly, but I found it relaxing and fun! I can now theoretically make my own clothes (this is what my mind does...)! I still like seeing the face on people when I would tell them that I'm sewing a skirt....

Anyways, back to the skirt. Sewing was just what I needed! It took a few tries to get it right, but once I got it figured out, the skirt inflated wonderfully. And here's the kicker: now that it's working, not only can I hover on asphalt without issues, but I'm using even LESS thrust than I ever have so far! I'll likely tweak the design a bit more, and I still need to find a way to bind it to cardboard (tape isn't doing it), but I'm hopeful once again!

Well crap.

I've been trying several methods for binding fabric to cardboard with limited success (magnets didn't work again...), and have even been hitting a roadblock with binding the corners together.

I'm trying to make these skirts as easy to make as possible such that I can simply create several of them and swap them relatively quickly if worse comes to worse. So far, my usual methods for binding the shape both to itself (fabric to fabric) and to the hovercraft (fabric to cardboard) have been failing me. I've been having limited success with double-sided scotch tape, but I know it won't do. Surprisingly enough super glue, gorilla glue, and even putting a torch to the fabric didn't work. I did find it amusing that hot glue, out of all things, seemed to work the best (but still not good enough for my preference).

The good news is that it's easy enough to cut the fabric on the laser cutter and get some really nice cutouts (fail fast, fail often), although it looks like the fabric that I'm currently using isn't necessarily the best choice. It's quite heavy (think half the thickness of denim pants), and is very hard to shape and in general work with.

While I do still have a month until the AVC, I'd much prefer to be hardware-complete sooner rather than later. I'm just not sure when this'll happen.

Up until now, I've only been talking about the Hovercraft as an entry for the Sparkfun AVC, but not necessarily going into too many details about that. This update is mainly to clear up a few things.

First off, when I decided to turn this into an AVC entry, I already knew that it was an excellent candidate for Non-Traditional Locomotion. In fact, I think that's what some of the CURM robots were going to do as well, but I'm not sure.

As time has gone by, I've also been trying to get this design to fit within the dimensions for the Micro/PBR category as well. It's requirements mean that I need to fit within a 4"x6"x10" box, which while I'm not quite there, I'm close.

One of my "bag of tricks" involves using Bluetooth for my communication between smartphone and firmware. I looked through the rules, and while it does not allow wireless communication during the match, I was able to get a clarification from the Sparkfun peeps saying that my specific design was alright, since the communication did not extend past the robot. They did recommend against using Bluetooth since the spectrum would be hella-crowded, but since my application has the antennas only centimeters away, I think I'll be fine. I did want to use USB, and did manage to get it working at one point, but since time is getting short, I don't think I can implement it and log enough hours with it to be comfortable with it.

The second trick was my choice of Smartphone. I've owned an HTC One S before until I lost it to a swimming accident (ziplock bags can't stop water. Who knew...). What I love about it is that for its generation, it's quite the powerhouse while at the same time being small. Although there are faster and more powerful phones available on the market right now, I'll gladly take the performance hit as a tradeoff for a smaller size.

As of right now, the Hovercraft is a bit too big to fit in the dimensions. There's simply too much stuff to fit it all in with the way that I've arranged things, but I do have one final trick.

It doesn't look like I'll have enough time to use Computer Vision in the project (walk before running), so my last trick will involve rotating the phone 90 degrees, and securing everything on the outside wall of the hovercraft.

Oh, and I've also started looking into making a skirt for the Hovercraft. Hoboy...

I have two tidbits of good news to share!

First, I have a Hovercraft Controller App! This makes driving around the Hovercraft MUCH more easier and fun to test and drive around. This is ideally the point I would have liked to been at at AMMF, but oh well. It's easy enough that I can give my phone to friends, and they can drive the Hovercraft around (unfortunately, that's still not that easy to do).

The second piece of news is even better! Yes, better than an App. Version 3.1 of the Hovercraft is working! I modified the Hovercraft to have a fixed base, and it seems to be holding up well. The asphalt doesn't do too much damage (from what I can tell), but it works!

There's still more I'd like to fix and improve before the Sparkfun AVC. For one, the hovering isn't consistent, which causes the Hovercraft to get stuck every once in a while.

There is also a larger issue that needs to be addressed. While the Dual EDFs ended up working better, the response time for the EDFs isn't what I'd like it to be. This could be a side effect of my command set, or the Android App, or the EDFs themselves, or a combination of everything above. Also, there are still times when an EDF simply cuts out, which I'm not sure if this is a firmware issue or hardware issue.

It's looking good, but there's still more to do.

Well darn. It seems that the magnets aren't enough to keep all of the air from escaping. Looks like I'll need to change this into a fixed chamber design. This means I'll need to really test to see how much damage the cardboard takes on asphalt.

I've got a new design, and it shows promise. I've altered a number of things based on what I've learned, and as a side note, the cardboard change was SO worth it!

Changes:

- Single hover EDF. Addresses rotational torque issue.

- Magnetic chamber. I'd like to be able to change out the bases as they wear and tear.

- New chamber design. Instead of a single big hole on the bottom, there are now six small holes. Still using electrical tape for sides of chamber.

- Removed Servo from design and instead opted for Dual EDF design. The EDFs are tilted out at 30 degrees to make turning easier.

Still need to test it, but I'm hopeful!

Well phooey. Now that I have better control over the hovercraft, I decided to give it its first test on Asphalt since that's what the Sparkfun AVC course is made of.

Unfortunately, the current design is a no-go. While I can get the hover EDFs spinning plenty fast, the Hovercraft is simply generating too much rotational torque for the current design to counteract for.

Will definitely be going back to the drawing board.

In my bout of testing over these past few weeks, I've discovered a number of issues with my BoosterPack that I'd like to address, and a few things I'd like to tweak.

One of the things that I'll be updating that's relevant to the Hovercraft is the ability to plug the battery directly into the BoosterPack, allowing me to get real time current consumption. I'm not able to do this right now since the current shunt resistor can only handle about 40A. Each EDF can top off at 30A, which is something that I'd rather not test. The plan for the next board will be to handle 100A at 24V.

Where does the time go? Just like that the Austin Mini Maker Faire (AMMF) has come and gone, and it was a blast!

Unfortunately, I didn't have time to get to the Hovercraft App since tackling the firmware issues took longer than expected. For most of the event, the robots were display only since I didn't have a good way to control them. Since the venue was crowded, I didn't want to risk having people step on the bots, or have the bots lose communication and simply walk off.

Although there were a few disappointed faces that I knew I just couldn't appease, I was able to throw together an app by the last hour to control both robots. Sure enough, the Hovercraft, with very little resistance to the concrete floor, was very difficult to control. I couldn't get the finesse to have it go in a straight line, and if it spun out of control, I could very rarely bring it back to control.

This is the first time that I gave the Hovercraft design a good breaking in, but I'll need to do something else. As it stands, I have to constantly tell the thrust EDF to spin clockwise to counteract the torque being created by the hover EDFs. I'm not exactly sure what I could try, but as it stands, I don't need that much thrust to get the craft to hover. I think I can get away with only one hover EDF with a better chamber. I'm also thinking that getting rid of the servo and trying differential thrust control for steering might work out.

I may have had an ulterior agenda for the hovercraft. Back when I was in college (a little more than a year ago), I was a part of the IEEE Robotics and Automation Society at UT Austin, and we would make robots for various competitions (this is actually how the Robotics BoosterPack got started. the department was switching over to Stellaris Launchpads, and we wanted to update our Robotics controller to keep in line with the Electrical Engineering's curriculum. Out of the three designs that were created for consideration, mine was a tad bit too complicated to make without assembly services, but I've still been working on it ever since).

Well, as luck would have it, there was a small informal competition that started being held called the Convention for Unconventional Robotic Movement (CURM), in which all you needed to do was go straight in a unique and unusual way. The only real limitation was that you couldn't use wheels. They didn't place a restriction that you needed to be a student, and since they're a fun, smart group of kids, I decided to enter and have fun.

Fast forward to the competition, and now I have a semi-working cardboard prototype! At this point, I have a very basic test app for the BoosterPack that isn't necessarily geared for Hovercraft control, but it'll do.

I've also made progress on the cardboard hovercraft, which works quite nice indoors (aka concrete and tiles). I created a chamber out of cardboard and addressed the perforation issue by wrapping electrical tape around the chamber. It works, but I'd like something better, if possible. I'm still using the dual EDF design to provide hover thrust, and ziptied a third EDF to a servo on top to provide steering capabilities. It get's the job done, but again, I'd like to find a better solution.

During the competition, it performed well enough (sense a trend?) since the distance needed to finish was only a couple of feet. Control is still a huge pain to do manually, which is something I'd like to address later. Good times were had, people OOhed and AAhed, and as far as I could tell, everyone had fun. Proof of concept success!

As of right now, there are ALOT of firmware bugs that need to be fixed. I also want a better way to control the Hovercraft, but that'll have to take a backseat until I get the firmware bugs swashed.

One week until the Austin Mini Maker Faire!