post-defcon impressions, ideas for next year (from winning team)

[theamk]

Hello,
I wanted to share some impressions from this year's DefconBots contest. We had some very great teams attending this contest.

First of all, I wanted to correct Kallahar. During the awards ceremony, he mentioned "a couple of thousand dollars" as a robot price. I am not sure about other teams, but my robot was less than $600. My generation 1 robot (which failed last year because of simple mechanical problem) was less than $500, and it was performing as good as this one. I could probably design an even cheaper robot.

In the next few days, I will be posting a list of lessons learned while designing, building and debugging my robot. If there is enough interest, I am ready to post all the information about hardware and firmware.

As for future contests, I have thought that this year we had many good competitors, so the rules should be changed minimally. If some change is required, I would recommend adding weight and/or dimensions limits to the robots.

Mikhail

[wsbpress]

Re: post-defcon impressions, ideas for next year (from winning team)

I also have some comments and suggestions but before that let me say congratulations to Mikhail on his victory and also thanks to Kallahar for the time and effort he put in to running the competition.

I think most would agree that my robot (Team Octopi) would probably be the most expensive to reproduce primarily because of the AC servo motor and driver that we used to rotate our turret (the "pan" in pan/tilt). However those parts were donated to our team, and not from our university either. We received no financial support or technical help from our university other than I was allowed access to the student machine shop (not CNC) and I used this project as my senior project, as did my three teammates. My out of pocket expenses were really not that much, I bought the geared stepper motor that we used from ebay for less than some of the nice RC servos. My greatest expense was in the time I spent on the mill in the shop and the effort I put in to learning all the stuff that I needed to build a bot for the first time. The aluminum stock and nuts and bolts and stuff were offered to me at a discount also. I was resourceful because I was a poor college student, I had limited cash but tons of time (and no social life or girlfriend at the time).

Anyway, enough of that, I have just always been a little frustrated with the misconception that my team had university backing and spent a bunch of cash on our machine.

That isnt really where I wanted to go with this reply. What I wanted to suggest was the possibility of having two categories for next year's competition. One that uses a hardware kit and everyone used the same hardware as Kallahar has suggested. And then have an Open class or whatever that is similar to what the contest has been for the last three years where there have not been limitations on hardware. That way those that want to use the hardware kit and compete with less of a time/$ investment can, but those who have already put their time, effort, and $ into their existing bots can compete with them again as well.

Having said that I probably should mention that my team will still not be competing at Defcon next year. I do really enjoy the contest though and I have enjoyed meeting those of you who have competed over the last three years. I hope next year's competition will be the best one yet.

-wsbpress

[kallahar]

Re: post-defcon impressions, ideas for next year (from winning team)

Thanks for everyone for competing :) I was pretty burnt at the ceremonies, so sorry if I misstated anything. I know that when building the kit I had a hard time getting the cost down. You can always save money by using used parts, etc but to create a "sellable" version gets expensive quickly. Props for building such excellent bots on a budget!

[cfinger]

Re: post-defcon impressions, ideas for next year (from winning team)

I would guess the 'thousands of dollars' came from my set up, because, well, it does cost thousands of dollars.

However, there are many things that were added for paintball, that add no functionality to the defconbots contest. (Waterproofing, anodizing, a paintball gun, LCD, onboard computer, enclosed camera). In fact, many of these hurt us at the competition... we couldn't get reliable fills on the co2 tanks, the camera wasn't shielded and picked up lots of wifi noise, also I had no gui when setting things up, because we're working on keeping everything on the turret's computer.

Last year I put about $350 into my robot, and shot them down in 36 seconds. Considering my trip to defcon cost $700+, thats pretty cheap. The university didn't provide anything either, except the machine shop key. Mine was still 99% PVC, so you probably could've cut it all w/ woodworking tools or dremel. Personally I like seeing what people hack together for their bot, but I understand that most defcon attendees are more software oriented.

[Queeg]

Re: post-defcon impressions, ideas for next year (from winning team)

I like ping pong!

[astcell]

Re: post-defcon impressions, ideas for next year (from winning team)

I wonder where the point of diminishing returns is located. Can you spend 3x the $ and get 3x the speed? How bout 10x? Assuming someone has $10,000 to spend on a robot will they do 20x better than a $500 robot?

If things get that busy we can always have an award based on effectiveness per dollar.

[TheCotMan]

Re: post-defcon impressions, ideas for next year (from winning team)

I wonder where the point of diminishing returns is located. Can you spend 3x the $ and get 3x the speed? How bout 10x? Assuming someone has $10,000 to spend on a robot will they do 20x better than a $500 robot?

If things get that busy we can always have an award based on effectiveness per dollar.

But then you have to "trust the client" and trust contestants to provide accurate information on what they paid for each item, or you have a master list of "real prices for all parts" and then deny your contestants any capacity they may have to "wheel and deal" when buying parts, or find ways to get the parts they need from other equipment which was thrown away, and by recycling (through explicit permission, or "dumpster diving") get parts for much cheaper than if they were new.

Kallahar and other people working on this have tried to address the above problem before, but as you can see, it is a difficult one to solve for a contest at Defcon.

What do you mean a portable nuclear reactor disqualifies me because of cost!? I got this off craigslist for free!

heh-heh.

[astcell]

Re: post-defcon impressions, ideas for next year (from winning team)

Go with MSRP on the parts then.

[YenTheFirst]

Re: post-defcon impressions, ideas for next year (from winning team)

You would probably also have to factor in how long they worked on it, and that could easily be lied about.

[TheCotMan]

Re: post-defcon impressions, ideas for next year (from winning team)

... or you have a master list of "real prices for all parts" and then deny your contestants any capacity they may have to "wheel and deal" when buying parts, or find ways to get the parts they need from other equipment which was thrown away, and by recycling (through explicit permission, or "dumpster diving") get parts for much cheaper than if they were new.

Go with MSRP on the parts then.

That is kind of what a 'master list of "real prices for all parts"' would be, and would then include the other problems listed above.

[wsbpress]

Re: post-defcon impressions, ideas for next year (from winning team)

Go with MSRP on the parts then.

This method fails when the contestant fabricates his/her own parts.


It seems to me that the task of accurately determining the true cost of a contest entry (ie. robot) is extremely difficult. Too many factors are subjective.

-wsbpress

[YenTheFirst]

Re: post-defcon impressions, ideas for next year (from winning team)

Getting an exact price is pretty much impossible, but perhaps a price range could be determined. I.E, bot #3 probably costs somewhere between $300-$600, bot#4 is between $500-$1000, etc.

Instead of calculating score per dollar, you could have different price ranges, like weight classes in boxing. It shouldn't be too hard to determine which bots fall in the $0-$499,$500-$1000, or $1000+ categories.