-
Polycarbonaut Mark 1:
Building an aluminum and polycarbonate robot chassis
My current "big" robot project is a low-cost omnidirectional vision system for hobbyist robots - something you might put on your sumo or fire-fighting robot. Wanting to get straight to the brains, I ordered a Terminator sumo kit from Lynxmotion to act as the mechanical platform. I was so impressed with their polycarbonate-and-aluminum kit that I went straight to designing my own scratch-built version. I needed an opponent to test my sighted sumo bot on, after all!
Well, I've ended up spending a lot of time on the Polycarbonaut Mark I, and virtually none on the vision system, so I thought I should share my experience with others. I'll focus on construction techniques here, leaving the more interesting topic of chassis design to the reader.
This robot is really not intended for the kind of combat that would involve coming at it with a chainsaw, but it is pretty sturdy. Also, like the Lynxmotion kit, this is just a base to build on. I leave the scoop, or whatever other application-specific hardware is needed, up to you.
Update: Part II is up.... come check it out here.
Tools
The tool selection I used isn't huge : hacksaw, metal file, cordless drill with bits for drilling metal, utility knife, straightedge, clamps, pliers, screwdriver, and a metal tap.
I ended up buying the tap and tap wrench (shown assembled here) from the hardware store so I could screw straight into the aluminum parts, instead of having to play around with nuts. They weren't very expensive, and it's the kind of thing you can easily find uses for. The tap I got included a drill bit, which is important since the size of hole you need for a proper thread tends to be both finnicky and hard to find in a standard drill bit set. I decided early on to only use one size of screw for the whole chassis (8-32), so that I wouldn't need more than one size of tap. I'll go over how to use the tap later on.
Materials
Polycarbonate ("Lexan"
or "Perspex") sheets are available at my local hardware store (The Home Depot on Saint-Antoine Ouest has them - others probably do also). I think people use them to cover windows in the winter. Anyway, good news for me, since it's the key ingredient here! Some more shatter-prone plastic sheets are also available, but early attempts at machining these left me to believe that any robot made out of the stuff would turn into tiny-little-robot-bits pretty quickly.The polycarbonate I got is in sheets of 11"x14" x 0.093" thick. This is thinner than the stuff Lynxmotion uses, but it's provided enough strength so far. I think that larger robots, or more extreme chassis designs, would require either thicker polycarb or reinforcement.
The polcarb comes covered with protective plastic on both sides - I kept this plastic on for as long as possible in the construction process, to keep the polycarb protected, and because it offers a handy surface to mark up without messing up your final product.
The aluminum in the Polycarbonaut is in two forms: aluminum bar for the main structure, and a 90-degree angle bracket for mounting secondary parts, like the shelf that's going to hold the electronics. The bar is 1/2" square, and the bracket is 3/4". These are also from my local hardware store, but I understand you can get scrap aluminum from plenty of cheaper sources.
Techniques : Polycarbonate
Working with polycarbonate is surprisingly easy - I'd say it's much more forgiving than wood, for example... it doesn't split, splinter, or crack the way wood does.
All the panels on the polycarbonaut are simple in shape, so I didn't have to make any fancy cuts. I want to try some more complex shapes in the future, but for now everything was done by scoring the plastic with a utility knife, and snapping it. It takes a surprising amount to get this stuff to snap... I wore safety glasses.
Scoring: clamp a straightedge along the line you want to cut. I keep the "exit edge" - where my blade ends up coming off the plastic when I score - hanging off the edge of my workbench, so I don't repetitively mash my blade into the workbench. It's probably a little more dangerous this way (don't put your fingers there!), but hey, we are building killer robots here, right?
Snapping: you need to score pretty deep before you can easily snap this stuff. When you do, it helps to use some blocks of wood to apply even pressure along your score. For big pieces, I've even gone so far as to clamp the main piece to the bench with one block of wood, while applying pressure to the overhanging piece with another.
After snapping the piece, there may be some irregularities in the edge. A quick filing with a metal file seems to fix it up just fine.
Holes are easily made by drilling with wood or metal bits. Always use a hole that's large enough for your screw to slide in - don't try to tap the polycarb, as this will weaken it; also, the plastic shrinks/grows quicker than the metal screws, so a hole that's too small might shrink around the screw and crack.
I haven't surfaced my polycarbonate panels yet. It seems to scratch pretty easily, and I hear that orbital sanding gives a nice, ash-coloured scratch-resistant finish. I'll update this page when I try out an orbital sander on the stuff. Another option is to paint with special plastic paint, available at hobby shops.
Techniques : Aluminum
The aluminum bar is easily cut with a hacksaw - aluminum is actually a pretty soft metal. To get a half-decent finish on the cut surfaces, a lot of filing was required. I remember using a special incremental metal planing tool in shop class when I was younger which rendered this task trivial, but since I don't own one of these the filing took a lot of time.
There's also some sort of dull coating on the sides of the aluminum bar which is difficult to file - you can see it in the pic below beside a section that I filed. The filed part is shiny, the part beside it is dull. After spending the time to file this stuff off on the first aluminum piece I made, I decided to leave the others. If anyone knows of a good way of dealing with this coating, let me know.
Quentin wrote in and let me know that the mystery coating is hydrated aluminum oxide (corundum) produced by anodizing. He says "There's no magic to removing the oxide layer - mechanical abrasion with sandpaper or a file is the answer. An orbital or belt sander will work. You can get a beautiful mirror surface with a felt wheel and polishing compound. It's a lot of work though."Alternatively, rather than beat it, you can join it, by anodizing the freshly cut surfaces. Anodizing protects the aluminum with a hard shell - you can anodize your own pieces with a fairly simple-looking tank setup and the right chemicals, and you can even colour your anodized aluminum! Do a google search for "anodizing aluminum" for more on this process.
Special drill bits for drilling metal are recommended for drilling the aluminum, and the use of a lubricant will speed things along - I've read that kerosene will work, but I picked up a bottle of special metal cutting fluid. Drilling a straight hole can be tricky without a drill press, but with some creative clamping it's possible to get a pretty good result. Making a small indentation with a punch (or any other sharp tool, such as a newly purchased tap) will help to start the hole at the right spot. I drilled all my holes straight through, so that when I tapped them I could clean them out more easily. Keeping little bits of loose, conductive metal around electronics and plastic gear sets is a bad idea.
Tapping was a little time consuming, but strangely satisfying. The tapping tool basically looks like a screw with some channels cut into it for metal bits to collect in. The T-shaped tapping wrench is definitely recommended, as it allows even pressure to be applied without any side-forces (which might snap the tap). Tapping is done by screwing in the tap a couple turns, or until turning becomes difficult, then screwing it back to clear the freshly removed aluminum bits, and repeating. For the bar I'm using, I had to completely remove the tap and clean off the aluminum bits a couple times for each hole. A drop or two of cutting fluid also helped the tapping process along. Because my holes go all the way through, I was able to use a pipecleaner to clean out all the extra aluminum bits.
Because the metal is so thin, the angle bracket is easier to cut, drill and tap than the bar. I was surprised how strong my tapped bracket ended up being - I expected to use bolts, but it turns out I didn't need them.
The Polycarbonaut
With the techniques described above, construction of the polycarbonaut was pretty straightforward. I had a couple extra geared motors, ball casters, and wheels sitting around from a previous robot, and tailored some of the dimensions to suit them. I started with the base:
I wanted to keep the robot symmetric, so I put the motor shafts right in the middle, front-to-back. For balance, the robot has two casters, centered left/right, towards the front and rear.
I also wanted the robot to be useful as a 3kg-class sumo bot, so I kept the front-to-back dimensions well under 20cm, leaving room for a front and back scoop. The wheel-to-wheel dimension is almost exactly 20cm.
Mounting all this up was pretty straightforward. At this point, this is actually a workable robot... well... it rolls when you give it juice, anyway!
Next, I wanted to create angled front/back panels, so I needed to make some angled mounting brackets out of the aluminum bar. The angle cuts weren't any trickier than a 90-degree cut, and a whole lot of filing later, they looked pretty pro. One snag here is that I originally intended to make the angle around 40 degrees, instead of 45. Big mistake - it takes a whole lot more work for nothing, since when you cut one 45 degree out of bar stock, you automatically have your next 45 sitting there ready to go. When you go with 40 degrees, you have a 50 degree waiting to go... and need to cut again to get it down to 40. I ended up sticking with 45 degree-ish angles. The polycarb is flexible, so the sligh errors in the angles generally go unnoticed.
The next step was a bit trickier: I needed to drill a hole, with a hand-drill, into the angled surface. For this, I created a special mounting bracket that would hold the angled face flat horizontally:
This same rig turned out to the handy for drilling the other holes, too - by mounting the part the other way around, I could use it as a vice. One tricky bit with the brackets was making sure the holes didn't intersect - in fact, some of my holes do intersect, on the bottom surface, which I did because I thought it would look cool, but which also means I need to use short screws. When I make the top brackets later on, I won't repeat this mistake.
Tapping goes off without a hitch. Lots of greasy aluminum bits everywhere. Drilled some nice holes in the base - transparent polycarb means you can align your drilling exactly!
Having screwed the four brackets to the base, I set about cutting front and back panels. I purposely let them overlap over the edge of the base, so as to provide more strength. The original front and back panels were too tall - I ended up trimming them.
Next came the sides. Having the rest of the robot assembled let me trace the perfect shape for the sides, including any slight errors in my brackets' angles. I made sure to overlap the side with the edges of the front/back and bottom panels, to increase the strength of the whole.
The bulk of the work for the side panels wasn't with the cutting and drilling, though, but with the motor shaft holes. I tried drilling holes for the shafts in one go - that is, I stuck my biggest drill bit in the tiny punch mark I made with my tap, and proceeded to drill a hole that was way off center. In retrospect, it's way easier to drill a small pilot hole (with your punched divot as a guide), then use the pilot hole as a guide for the bigger bit. Sounds like more work, but in the end you get a centered hole.
So, sitting with holes off center, and not wanting to waste my beautiful perfectly-angled side panels, I put my dremel to good use and cut/ground the shape you see below:
I don't include the dremel in the list of tools needed, since if I hadn't messed up, I wouldn't have needed it. It is mighty handy to have around, though. Next came 4 more mounting brackets. This time I didn't intersect any holes.
Now come the two side-rails for the shelf...:
Finally, we find out what the angle bracket is for! I'm planning on using IR sensors mounted in the middle of the robot, looking outwards, so I'm leaving most of space in the bottom of the robot unused. The electronics will sit on a polycarb shelf that runs between these two angle brackets.
Now all I need is a top and a shelf:
I had intended to trim the side-panels to be flush on the top, but the "fins" look kind of cool, and provide a useful way to handle the bot. I'll probably file down the sharp corners on the bottom surfaces of the side panels, though - they look cool, but will probably snag stuff.
Next I want to mount the batteries and sensors below the shelf, and then build the electronics onto the shelf. Finally, I might consier finishing the surfaces of the robot... although it would be cool to keep the transparent sections so my sensors can see through the polycarb, instead of having to cut holes.
Here's what the Polycarbonaut might look like if it was available as a kit (minus the shelf and top, since I didn't have those done when I took this pic):
Lessons learned:
- Pilot holes are good.
- Filing sucks, but makes things oh-so-pretty!
- Once you've chosen an angle for your front/back panels, don't change it to save time! (Conversely, chose your angles wisely!)
- The 8 bar-stock brackets on this robot took me more time to make than anything else. In retrospect, I'm thinking that angle brackets and hinges might have worked just as well or better, and taken far less work.
- I used size 8-32 screws for the whole robot - something smaller would have probably been strong enough, and been better on weight.
Finally, for those who are interested, the motors and rollers that I used on the robot are both available from HVWTech (Tamiya 6-speed gearset (SKU# 22050), and Ball Casters (SKU# 23120)). These parts cost more than the chassis itself, but thankfully I had them lying around from a previous robot.
Hope you found something useful in this - I figure, spending this much time on something must have generated some sort of useful information! If there's any demand, I'll post another guide as construction continues.
Update: Part II is up.... come check it out here.
Happy building!
Don