SumoBot TP: Difference between revisions

From Hive13 Wiki
Jump to navigation Jump to search
mNo edit summary
Line 171: Line 171:

Revision as of 10:05, 12 April 2011

Guard 20110403 1.jpg

This is the wiki page for a mini-sumo bot TP is building named 'Guard'.

Looking forward to our upcoming Mini SumoBot Competition - April 16th, 2011...


LVL1 in Louisville challenged Hive13 to join them in a Mini SumoBot competition early last fall (Oct 2010), and from what I hear, the folks on the House Bot ('Mr. Plow') team put in quite a showing! This has inspired me to try to build a bot to 'play' with Mr. Plow. By play, I of course mean "provide as much competition as I can."


I really do intended to put some Mini Sumo background info here (really). I think the Mr. Plow page has links to some background stuff.

For now, I'll just note that

  • bot must fit into a bounding box 10cm square (height does not matter from the point of view of the rules)
  • bot must weigh no more than 500 grams
  • bot must be autonomous (no remote control)
  • bot must have a delayed reaction (at least 5 seconds) to pressing of it's 'Start' button
  • bot must look cool
  • bot must win (often)

(The last two only apply to MY bot)


  • 2 Sharp 10-80cm IR range finders
  • 4 Sharp 4cm IR proximity sensors
  • 3 IR reflectance sensors
  • 1 momentary push button (Start)

(8 digital, 2 analog)


Basically, two DC motors in a gear box driven with H-bridges. Each motor will require 3 outputs (A,B,pulse).

I read somewhere that the tire hubs and the gearbox drive shafts will need to be modified to get it to fit within the 10cm bounding box. Turns out that was completely true - wish me luck.

Yes I know that the gearbox came with a set of motors, but they need to be replaced to match up with the driver I chose. Don't you worry - the replacement motors are every bit as cheap as the originals.


For my controller, as usual, I will be using a PIC16F688. "But wait, TP! That thing only has 12 I/O pins!" (10 sensor inputs + 6 motor outputs + 2 feedback LEDs = 18 I/O) Huh. Well, that's what they make shift registers for. I just love these little PICs too much. Irony Not Missed: adding the shift registers will take up more space that using a larger PIC with sufficient I/O pins.

I will be using a straight forward, pre-determined decision routine - no fancy AI crap here! (At least for now... this chassis may replace the current platform for SGRC4)

Current Status - Active

  • Wiring complete. Software debugging in progress. Visual appearance also in progress.

Work Log

Nov 8, 2010

I have actually been rolling this around in my head since the announcement of the LVL1 competition in Louisville. Finally got started with breadboarding the control system over the weekend. I have a box full of parts picked out on that I plan to order some time late next week. Hopeful I can get this bot into the ring before the year is out.

Jan 23, 2011

So much for the plan back from November!

Current Status:

  • Basic shift I/O of the control system functional
  • Parts have been ordered and should arrive mid week!

I have been kicking around ideas for wrapping the bot with a fiberglass shell. Still a thought in progress...

Jan 24, 2011

Or maybe already Jan 25...


In anticipation of the arrival of my parts order, on Sunday night (yesterday for me, right?) I decided to screw around with the control code I wrote a couple months ago, seeing as how it's in assembly and everything. Ya know, dust off the breadboarded circuit that's been patiently waiting all this time and make sure none of the wires have been accidently pulled out (several resistors were missing, actually): and my usually flaky PIC programmer just refused to work at all. I mean... really? Now? Flaky was annoying, but at least it worked more often than not. Well, it did work sometimes :)

It's always something, isn't it? I need to get one of those "I void warranties" t-shirts...

After many, many hours of combing websites and trying every combination of (computer / OS / IDE / complier / programmer software) I could cobble together without spending any money, I decided to take the programmer apart and look at the under side (bye bye warranty seal). No dice. Not sure what I was hoping to find. ANYWAY - at some point, for reasons not quite clear to me, I took apart the ZIF connector (woah - two screws and no sticker) and upon re-assembly I found that it held the chip much tighter, and of course the programmer now works...


So, hours wasted on a loose-gripping ZIF connector... but at least now my PIC programmer fires every single time!

I am SO glad I sprang the extra 5 bucks for the two-day shipping - I can't wait!

Jan 26, 2011

Parts arrived. Assembled the gearbox and tires, and started working on the layout. As expected, the drive shafts will indeed need to be ground down to stay within the 10cm box. The gearbox itself, however, takes up a lot of room regardless - where am I supposed to stick the stupid 9 volt?! I was hoping it would fit between the tires, underneath.

I don't think I am going to end up with the profile I wanted, but I am still hoping to keep the top from being no higher than the tires.

The plan for now is to concentrate on the physical construction of the bot. I have the I/O system breadboarded out, and while it is not completely 'ready to go,' it is close enough that I know what I'm working with as far as components and what-not. I have been mostly waiting for the parts so I could try to pack everything into the space desired - then I'll finish the software.

Guard title start.jpgGuard 20110126.jpg

Jan 27, 2011

Slow goings... stopped by the Hive tonight and cut down the wheel hubs and drive shafts - a hack saw went through it like butter - total width of the wheels is now approx 9.7cm. Also tried my hand at cutting some high-density fiberboard for building out the construction. Framing this thing is gonna take some doing...

Jan 29, 2011

Total sensor fest. Got all of the sensors mounted. Many will need adjusting later on, but for now this is enough to move on to wiring and placing the other electronic components.

Guard 20110129.jpg

Feb 02, 2011

Finished soldering the upper control board. This board contains the voltage regulator, microcontroller, output register, and motor driver. Looking at this picture, even I have trouble seeing all the work that went into it. Let's just say that I'm tired of working with perfboard, although I say that knowing there is still the lower board with the input register to make.

Guard 20110202.jpg

Feb 03, 2011

With the upper control board completed, it's time to test it! Had to rewrite most of my basic control code to include the intended search routine. For some reason my camera limits me to 20 seconds of video, so here are two (really) quick clips showing it moving about in search mode (no sensors are currently attached or even supported yet for that matter).

Feb 10, 2011

Still working on basic electronic connections, when I can. Soldered the input board tonight, but did not get far enough to test it yet.

Guard 20110210.jpg

Apr 03, 2011

I haven't posted any progress lately, but I have, in fact, been working on this guy...

The past week or so was spent in R&D for turning a screen wireform into a solid object (fun, if not mostly disappointing), and making wiring harnesses (boring, but more or less satisfying). I ended up completely abandoning the high-density fiberboard. The pink hue of the frame is from the activator for the Bondo; can't wait to paint :) I would share how I ended up building the frame, but will wait until I refine the process a little more (still planning to build a cover). If it seems like there is a lot of wiring going on, well, there is. Guard's got 9 sensors with three leads on each one, so...

The battery I've got is starting to die, but it seemed like things were working well enough before that. I am intentionally trying to draw as much power as a 9V will give me, so I may end up with a pocket full of batteries on fight night :) Still troubleshooting the long range IR sensors, but feeling generally good about the system.

A little concerned that my powertrain is not generating enough torque - although I have not been able to create a situation where the tires stall - they slip easily on all the surfaces I have. No clue what the current weight is, but I'd guess just about 300 grams.... brainstorming ways to add weight. Also, my black test surface is so shiny that it creates intermittent false positives for the line sensors (at least that's what I hope is happening).

Really looking forward to doing some tests down at the Hive.

Guard 20110403 1.jpg Guard 20110403 2.jpg

Oh, and I discovered that the top of my stove has the best lighting in the house for taking pictures...

Current Issues

  • Long and short range IR sensors have some strange conflict - may be due to low battery power.
  • Best guess is that my test surface is too shiny (you can see your reflection in it) for proper line sensor operation, although they seem to be working, more or less.

Next Steps

  • Ring test down at the Hive
  • Tweak, tweak, tweak (oh, and work on the bot, too, right?)
  • Construct totally amazing cover
  • Add enough weight to bring bot up to the edge of competition limits
  • Kick butt in sumo ring