Re: Autonomous Impressions
 

We've done the same; most of our changes have been in the search routine. We were doing really well in autonomous until we had one unfortunate (and unlucky) match where we were spun around, so we added the PID control to prevent any more of that silliness.

Re: Autonomous Impressions
 

Autonomous this year, from what I have seen at Great Lakes and West Michigan, is broken, seriously. Vision was a component that was really meant to be in a game like last year's, where it took two colors to recognize and some darn good logic skills to get it right. This year, instead of look, find, approach, guess, check, and place a tetra, its more like aim, fire, and hope to goodness that the ball goes in the hole, not make one somewhere else. Quite frankly, any team could build a generic softball pitcher and win because they have the camera software all ready to go.

In other words, autonomous would be much better this year if there were two colors to aim for: red light on red alliance, blue on blue, or something of that sort.

However, I am NOT asking for pain like last year, where there was just camera and minimalist code. Somewhere between this year's and last year's available autonomous equipment in the KOP would be the sweet spot for next year's game.

Also, the dumping half of autonomous was an excellent back door for teams, like 322, who kept it simple. An easy and quick 10 points in what could take a shooter three times as much time to score just three balls. Sometimes I wish we would've done that, yet again, I'd probably hear it from the power cravers.

-Joe

Re: Autonomous Impressions
 

This year's autonomous was the best yet. They gave us a difficult challenge and i think we rose to the occasion.

This year was the year of effective control loops and cunning programming. I can only hope that subsequent games will be as much fun.

This year we saw corner dumpers, center shooters, defensive maneuvers, and even purely anti-defensive maneuvers. As the team's programmer, i had fun with the combination of a well made two axis turret with a vision system.

Re: Autonomous Impressions
 

CMUcam works excellently. I agree, the extra 5 seconds would be nice... I could do a heck of a lot more in 5 seconds.... have the robot dance, serve drinks, the possibilities are endless... :yikes:

The more sensors and autonomous hardware/targets we get, the better.

The thing I'd like best is if the 2007 robot controller would have 2 TTL serial pins so we could use 2 CMUcams for steroscopic vision / location triangulation... the processor already runs at 40mips or so, so it shouldnt have a problem eating data from two cameras...

Would be great, the robot could know exactly where it is on the field at all times. Then, the targets themselves wouldnt actually need a marker of their own.... would only need a lit target on the four or even two corners of the field...ok now i'm getting too techincal, probably can tell I eat this stuff up. :ahh:

I would tell you all what new commands I'm adding to our autonomous scripting language...

(its name is RALFF: Robot Autonomous Language For FIRST, would be happy to answer most any questions you have about it)

...but I'll just let you guys see what it can do when you see us at Atlanta.

-Q

Re: Autonomous Impressions
 

The camera is good, I wish the autonomous period was longer.

Having a lit target helped a lot, but at 2 regionals, we found many lights that would fool it. Someone from the crew should check out the area with a camera and laptop before things get started, with the final lighting. At Las Vegas, there were advertising signs that were off on Thurs, but on Fri. & Sat. These were backlit with fluorescent tubes, and our test camera saw them as matching the target.

Maybe next year they will design the field with a backdrop behind the target.

Of course designing an autonomous system to ignore spurious inputs, is an important thing to learn about. An ideal robot would use other means to determine its position and heading, then only use the camera for fine tuning the final aim with a narrow field of view.

Re: Autonomous Impressions
 

As you said, I don't think anyone was looking to repeat last year, which is why I think things are less complex this year. Teams have options, you CAN score in auto without the camera but the ability to use the camera gives you more robustness and flexibility in autonomous.

I think part of what makes this year so nice is that the range of "skill" levels is well distributed. Last year maybe 5 teams used the camera and it only sometimes worked, which wasn't very exciting. This year some have chosen to go for the side goal, some are "point and shoot" some are "aim and shoot" and then there are teams which navigate/aim via the camera and only shoot if they have "lock". This gives spectators a nice bell-curve type of situation, some do it really well, some don't do it at all but most teams are participating in autonomous at some level.

Re: Autonomous Impressions
 

Good point.

I think its the matchups that could, and that's a 'could,' make autonomous broken. Example: two triplets at GLR. They both, aimed and shot very well versus an alliance that could aim and shoot very well, just a bit slow (*sigh*).

In short, the championship will probably be the determinant.

Re: Autonomous Impressions
 

I wish auto was longer
I wish i had more weight
I wish i had more build time
First wants you to say these things. If you didnt then they gave to you much. :yikes:

shaun