Questions

[RoboHak]

This contest sounds very interesting. I haven't built a robot before, but I've always wanted to.

From reading the rules, it sounds like at least some of the targets would be moving. Will all of them be moving, or will there be some which are stationary? I was also wondering, will the robot need to move around or will it just need to aim and fire?

I'm looking forward to hearing more about this contest. I would also be interested to hear from others who might be participating.

[jotux]

Re: Questions

This contest sounds very interesting. I haven't built a robot before, but I've always wanted to.

From reading the rules, it sounds like at least some of the targets would be moving. Will all of them be moving, or will there be some which are stationary? I was also wondering, will the robot need to move around or will it just need to aim and fire?

I'm looking forward to hearing more about this contest. I would also be interested to hear from others who might be participating.

If you've never built a robot before you can use the reference design as a starting point: https://github.com/Defconbots/2014_Reference_Robot

All of the targets are moving (around 1/4 m/s). Your robot will be stationary on a table but will need to aim your laser to shoot the moving targets.

[RoboHak]

Re: Questions

If you've never built a robot before you can use the reference design as a starting point: https://github.com/Defconbots/2014_Reference_Robot
All of the targets are moving (around 1/4 m/s). Your robot will be stationary on a table but will need to aim your laser to shoot the moving targets.

I would enjoy the challenge of building it myself, even if I don't score well, but I will probably refer to the reference design. I already have some ideas about the aiming and targeting systems. I was looking into OpenCV for identifying the targets, but I have a few questions about the targets and the arena.

Will the arena be dark, dim, or normal/bright lighting?

I know you said the floor will be black, will the background behind the targets always be black?

Will anything else in the arena be blue?

Will the targets be full spheres and be able to be hit from any direction?

Any pictures of the targets, the target stands, and the "train" they will be on would be very helpful.

[jotux]

Re: Questions

>Will the arena be dark, dim, or normal/bright lighting?

We're at the mercy of defcon and where they place us in the contest hall. From past experience the lighting is a little low but consistent.

>I know you said the floor will be black, will the background behind the targets always be black?

That background around the targets will, in all likelyhood, be black. But keep in mind the targets are low to the ground so the background will probably be irrelevant.

>Will anything else in the arena be blue?

No.

>Will the targets be full spheres and be able to be hit from any direction?

The targets are 40mm speheres with the top ~80%, from any direction, able to detect laser hits. Because robots will sit on top of a table and look down at the targets the bottom area that cannot detect will probably not be visible. So basically all visible parts of the target, as viewed from the robot, can detect laser hits.

>Any pictures of the targets, the target stands, and the "train" they will be on would be very helpful.

I just finished up the targets which is why I'm a little slow to respond to your post. But here are some pictures:

The train with targets:
http://jotux.com/defconbots/target/train.jpg

A video of me shooting the targets with a laser and their behavior:
http://jotux.com/defconbots/target/laser_behavior.mp4

[n0tan4m3]

Re: Questions

Do you have any recommendations on the hardware to use for testing tracking and shooting, or is that the black box part of the challenge?

[jotux]

Re: Questions

Do you have any recommendations on the hardware to use for testing tracking and shooting, or is that the black box part of the challenge?

Do you mean recommendations on how to simulate targets? I think the trivial way to test a robot would be to put a blue LED inside of a ping-pong ball on the end of a stick and just manually move it in front of the robot. The train moves around 10 inches/second maximum, so you can wave it around and make sure your laser hits are landing on the ball.

[GurtDotCom]

Re: Questions

I appreciate all your answers!

Is it safe to assume that a standard webcam (like the one in the sample build) will be able to see the entire competition area without being moved? I understand that the robot will sit on a table and overlook the competition area, but I want to make sure it is far enough away from that area to see it properly without a special lens or moving.

Thanks

[jotux]

Re: Questions

Is it safe to assume that a standard webcam (like the one in the sample build) will be able to see the entire competition area without being moved?

If the webcam is stationary you'll probably be able to see most of the target area but not all of it. The targets are going to be on a stationary track in front of the table and we'll likely run the track close to the table (withing 3' or so).

Here's a concept drawing of the arena:

http://i.imgur.com/4RzQxc4.png

The track will be 20-30m total length and take 60-90 seconds for each wave. We can make the track more or less difficult depending on how we arrange the pieces and it *will* change between qualification(Friday) and the finals(Saturday). We'll probably try to make the track relatively "easy" Friday so we can judge how well competitors do then we'll adjust it to make for exciting finals Saturday.

[n0tan4m3]

Re: Questions

Here's a concept drawing of the arena:

http://i.imgur.com/4RzQxc4.png

The track will be 20-30m total length and take 60-90 seconds for each wave. We can make the track more or less difficult depending on how we arrange the pieces and it *will* change between qualification(Friday) and the finals(Saturday). We'll probably try to make the track relatively "easy" Friday so we can judge how well competitors do then we'll adjust it to make for exciting finals Saturday.

Yikes, that looks wicked. :)

Question - I see in the rules there will be 15 minutes to setup. Will the targets be illuminated in this time?

[jotux]

Re: Questions

Question - I see in the rules there will be 15 minutes to setup. Will the targets be illuminated in this time?

The targets will not be illuminated before the start but they'll always start in the same place (Friday and Saturday) so you'll know where the wave originated.

[mr_bandit]

Re: Questions

I am asking this to see if I understand the rules.

Assume the train has 5 blue balls (no snickering...)

Wave #1: each time you hit a single blue ball, it turns red, the rest of the balls go dark, then the hit ball goes dark and the rest of the unhit balls go from dark to blue. Thus, the only time a ball is a target, it is blue.

Wave #2 is like wave #1, except you need to hit each ball twice to make it go dark until the end of the wave.

Thus, for wave #n, you just keep hitting a blue ball N times until it goes red -> dark. Thus, order of hitting the balls does not matter, if you only aim/hit blue balls.

Given the way you are laying out the track, it is likely that a given (blue) ball may not be seen for a part of the time.

How does the robot know it is at the end of a wave? It seems it has to keep track of the hits and deduce the end of the wave. Or .. it doesn't care and just keeps hitting blue balls whenever it sees and can track/hit one. You (the game master) determines the robot cannot end the wave and stops the play.

[jotux]

Re: Questions

I am asking this to see if I understand the rules.

Assume the train has 5 blue balls (no snickering...)

Wave #1: each time you hit a single blue ball, it turns red, the rest of the balls go dark, then the hit ball goes dark and the rest of the unhit balls go from dark to blue. Thus, the only time a ball is a target, it is blue.

Wave #2 is like wave #1, except you need to hit each ball twice to make it go dark until the end of the wave.

Thus, for wave #n, you just keep hitting a blue ball N times until it goes red -> dark. Thus, order of hitting the balls does not matter, if you only aim/hit blue balls.

Given the way you are laying out the track, it is likely that a given (blue) ball may not be seen for a part of the time.

How does the robot know it is at the end of a wave? It seems it has to keep track of the hits and deduce the end of the wave. Or .. it doesn't care and just keeps hitting blue balls whenever it sees and can track/hit one. You (the game master) determines the robot cannot end the wave and stops the play.

That is all correct.

[mr_bandit]

Re: Questions

That is all correct.

Cool. Thanks.

[RoboHak]

Re: Questions

I was looking at the rule changes and noticed what seems to be a discrepancy.

The part of the rules talking about the waves says

A "hit" on a target is with laser for 450ms-1200ms (Specific laser above!) while illuminated

The part about the software says

Laser should be pulsed on for 500ms with a 1500ms delay between shots.

Does that mean many shots which hit the target for 500ms may not register?

[jotux]

Re: Questions

Does that mean many shots which hit the target for 500ms may not register?

A 500ms hit should always register and when I test the reference robot I built against the targets the targets always register hits that are exactly 500ms.

Honestly, I should have left the "450ms-1200ms" comment out of the rules as it's more a comment about target implementation and less an important characteristic competitors should worry about.