Defcon 17 Ideas

[kallahar]

If you have an idea for next year, or if you would like to volunteer your time, please post in this thread! I've had a great time running the contest, but if anyone wants to try their hand at running the whole thing I'm totally willing to hand off the reigns (it means I can compete again, heh).

Otherwise, please post and discuss what you think would make a good contest next year. Whatever we do, we need to nail down the rules by October 1st or so in order to give people plenty of time to build.

Kallahar

[kallahar]

Re: Defcon 17 Ideas

Idea 1) Robot Guided Missiles. Shoot across the entire contest area, hit a target on the floor (big cross or something). Have to have on-board control, but can get commands remotely. I think it's doable with two micro-servos, a serial->servo board, and a wireless serial port to turn the fins. Then a wifi camera in the nose to relay video. The laptop on the ground does all the processing. Not sure if there's enough time to calculate and move the control surfaces though...

Idea 2) Same as 1, but dropped from the balcony

Idea 3) Take an R/C Blimp and either have the computer fly it to a target, or drop a bomb on a target, or just do auto-release with RC control, something like that. Probably too hard to do though.

Idea 4) Stealth bot. Have a bot have to deliver a package from point A to point B (probably R/C with cameras) without being either spotted or stolen/kicked/etc. Probably impractical due to the audience though.

Those are what I came up with last night. I like #1, but I think I should build a proof of concept to see if it's even possible...

[theamk]

Re: Defcon 17 Ideas

Here are my ideas:

1) I like the idea of autonomous blimp. I think it it practical to make one, provided that there are navigation assists. For example, there might be multiple IR beacons, one in each corner of the field, plus one more at a target. The blimp would to have IR sensors with various fields of view. A computer would be able to determine the directiion and rought location of blimp from the visible beacons. Then, it will zero on the destination and release the payload.

This seems like a reasonable task. The only possible problem is drafts from A/C system that might deflect the blimp.

2) A delivery robot (probably small-scale simulation). The playing field is a maze. It has markers (either barcodes or IR beacons) to help robot determine its location. The robot it dropped in the random location, and it has to find a shortest way to destination. The maze map with beacon locations could be provided.

3) An autonomous tank battle. There are two small-scale, autonomous tanks with tiny airsoft guns (such as http://www.raidentech.com/usm4bu1airct.html or http://www.bananahobby.com/1654.html ). They move under computer control and try to hit each other. To make contest more interesting, we can only count shots to the sides or rear of the tank. To help AI, it might be a good idea to have each tank flash a beacon in the direction its turrent is pointing.

4) Vehicle battle - simular to the tanks, but there are no flying bullets. Each vehicle has a flag mounted in rear, and when the flag is broken, the vehicle loses. Again, they are fully autonomous, and have beacons to help other vehicles identify their position.

5) Capture The Flag - a playing field with a mark in the center. Whichever autonomous vehicle can stand in the center for 20 continious seconds, wins. This could use standartized sumo robot bodies.

my most favorite ideas are (1) and (4).
idea (2) is most practical, as it could be used even after the contest is over (home beer delivery bot, anyone?)

Rules check:

* it's time to involve more people -- those projects are both simplier mechanically and more intersting programmatically. Assuming that the number of programmers is >> number of engineers, they will attract more people.

* Cost needs to be kept low -- all of those projects are based on standart R/C toy (less than $150) + micro controller + IR sensors + possibly remote camera. The whole setup could be under $500 new, much less for simplier robots.

* This isn't an engineering competition -- no special tools/skills required (one has to use soldering iron, dremel and hot glue -- a faily common skills)

* The event should be unique to defcon -- I don't know of any existing contests with same rules (except perhaps #2???). I did not search too hard, though.

My cooperation:

I am willing to provide some help with the contest. I think I can do few items from the following list. I can relase all h/w design and firmware.

(a) Get one of R/C toys above and convert it to microprocessor control.

(b) Create bi-directional radio channel to PC using bluetooth or XE1203-based modem and atmega168 microcontroller (think Arduino).

(c) Create a system of IR beacons and IR receivers.

(d) write some very simple control software for testing of (a), (b), (c) above

Mikhail

[Deviant Ollam]

Re: Defcon 17 Ideas

This suggestion is totally inspired by the folks at the I-Hacked skybox and their wicked-awesome laser pointer drawing contraption.

I wanna see a category of DEFCON Bots contest where people can lase their targets via IR beams. In fact, i'm fucking shocked as hell that no one has tried to do that before now in order to cheat in the contest. haha. Imagine someone with NVG on, up on a skybox balcony, surreptitiously spotting and pinpointing targets with an IR beam. heh, it'd be like calling in an air strike.

[kallahar]

Re: Defcon 17 Ideas

Heh, yeah, aerial bombers with their target designated by a person with a laser :)

I also thought about a lock-pick or safe cracking bot, but not sure if that's more lock picking and less robot.

[kallahar]

Re: Defcon 17 Ideas

More ideas:

Robot Duckhunt: play on an actual NES, control the gun. not really a new idea though, and I don't think you can play it on a projector.

Robot pong: robots move in one axis, play against each other with a golf ball or something.

Robot pacman/tetris/etc, not much "robot" part, but plenty of software that would be articulated to an original system?

Pool bot: driving bot that can play pool. We don't have a table though...

Lots of other sports could be adapted too... ping pong, bowling, etc.

More ideas?

[miked]

Re: Defcon 17 Ideas

I like the shooting robot theme so I thought of some shooting stuff that humans play.

1. Moving paper targets with rings for scores. Target moves across field. Bot has two shots (like a shotgun). Bot has five chances, Scores are added up form the five chances (ten total shots).

2. Skeet shooting. Target can start from one of two positions, moving across field. Bot has two shots to hit. Scoring based on number of hits.

3. Robot dual. Bots start back to back. Go out 10 paces, turn, shoot once.

4. Friend or Foe. Target pops up, bot has to determine friend or foe and take appropriate action. Score based on time to kill foes, penalties for killed friends.

[miked]

Re: Defcon 17 Ideas

Potential problems with other ideas:

Blimps Dropping Bombs - As soon as you release the payload, the blimp loses buoyancy and hits the ceiling. Alternative - what about shooting the blimp?

Other Sports - If this event changes too much, it may "infringe" on another event. It like this event needs as much support as it can get. If it resembles another too much, it may lose legitimacy.

IR/RF Beacons - Spoofing. This is a hacker con. I am surprised there has been no problems yet.

[YenTheFirst]

Re: Defcon 17 Ideas

This definitely steals another contest's idea, but:
Mystery Challenge.
Competitors are given a basic kit of parts, and have to construct and program a bot that will compete in a variety of events.

Either way, having some sort of predefined kit might be nice. It allows the hardware incompetent a chance to try it out, and still allows plenty of room for hacking. You could certainly make it not necessary to use the kit at all.

[YenTheFirst]

Re: Defcon 17 Ideas

Also, for the tank-related ideas (Which I think sound fun):
What about using nerf bullets as the ammunition, and just scoring manually? There could be a wireless AP if the bots want to check what their or their opponent's health is. It would be extremely difficult to spoof nerf bullets. Plus, they're something fun, and there's lots of existing nerf gun modification/creation to draw from.

[kallahar]

Re: Defcon 17 Ideas

Okay! Lots of great ideas, I think I've come up with a contest that is both easy to enter and yet hard to win.

We're still beta testing the idea, I have to make sure it's doable before I get everyone building on it. It really sucks when the contest turns out to be way harder or easier than you imagined :)

All I'll say now (definately don't start buying parts yet) is that it will be indoors and require a R/C style car.

Rules/etc should be finalized in a few weeks, until then, start thinking about how to control mobile robots :)

Kallahar

[miked]

Re: Defcon 17 Ideas

Rules/etc should be finalized in a few weeks, until then, start thinking about how to control mobile robots :)

Kallahar

You are a rule tease!

[kallahar]

Re: Defcon 17 Ideas

:) We've got a bunch of parts on order, so hopefully we'll be able to finalize how difficult the idea will be this weekend. Last weekend we tried to proof-of-concept another idea, and it proved to be too difficult.