REWARD OFFERED: A CAPPED VISION TETRA ON THE CENTER FIELD GOAL

[dradius]

Originally Posted by Chris Hibner

I agree with you, Dave. The time constraint WAS too short when all we had to worry about was operator control. Now it's WAAAAYYY too short. I was just thinking the other day about how much different things are now with autonomous control. We now have 3 times as many people working on controls than we used to have before there was an autonomous portion of the game.

One possible comprimise is to give teams the control systems and sensors one or two months in advance. If we had that extra time to work with the camera, I'm sure we would be capping vision tetras in autonomous. We decided that the time wouldn't allow it so we went with a more tried and true autonomous method.

not to be rude, but our most difficult thing to program was the camera. we literally spent 3 or so weeks, working every day out at oceaneering, to get the camera to track correctly. at that time, we hadnt finished the robot, and had more or so created an assembly line where the pieces of the robot came together close to the end (besides the tranny and drive frame). there was, therefore, enough time. it just had to be managed properly (not to be rude or haughty, like i said).

not to mention, think of it this way: this year WAS the "practice" chance to utilize the camera on the robot. you now have the software and hardware to accomplish the task during the summer or even sooner on a previous robot (or this one if you get it back and arent going to nationals). the first year of infrared was ridiculous, and, in a sense, so was last year. they canned it for that reason. THIS, however, is well designed and only takes the right programming to get it to work (what a statement). once the code is perfected, it works every time. the infrared wasnt reliable in that sense. considering that, you have the time now and later--you could even order another camera. we did.

[PURPLE!]

Originally Posted by dradius

not to be rude, but our most difficult thing to program was the camera. we literally spent 3 or so weeks, working every day out at oceaneering, to get the camera to track correctly. at that time, we hadnt finished the robot, and had more or so created an assembly line where the pieces of the robot came together close to the end (besides the tranny and drive frame). there was, therefore, enough time. it just had to be managed properly (not to be rude or haughty, like i said).

not to mention, think of it this way: this year WAS the "practice" chance to utilize the camera on the robot. you now have the software and hardware to accomplish the task during the summer or even sooner on a previous robot (or this one if you get it back and arent going to nationals). the first year of infrared was ridiculous, and, in a sense, so was last year. they canned it for that reason. THIS, however, is well designed and only takes the right programming to get it to work (what a statement). once the code is perfected, it works every time. the infrared wasnt reliable in that sense. considering that, you have the time now and later--you could even order another camera. we did.

of cource, he ALSO dosn't mention that, although when we praticed at ocenearing our ato. mode worked very well. Unfortunatly... we used the wrong shade of green on the tetras XD...

It seems most people are in agreement - the cost/benefit/risk of capping a vision tetra just wasn't there. If some team actually was consistent in doing this, it would just be too easy to counter-program a robot to drive out to the center goal and block. Also it is a lot easier to have your entire alliance get programmed to capture the entire home row in autonomous. - All that being said I am surprised that there weren't more teams using the cameras in a stealthy way, using it to position the robot near the loading zone or using it to score the non-vision tetra in autonomous. I am expecting (hoping) that some of you will be thinking up some unconventional tricks to use the cameras to your advantage. In such a competitive game starting out with a lead is a lot better than trying to come from behind.

Also - this was the first year of using the FIRST CMU Cam - I think in Botball, they have been using a camera for at least three years. I don't think anyone in botball was very successful using the camera the first year either. Their sophistication took some time to develop. I understand from a friend involved in botball that the first version of the software they had really didn't do the job, and it took a lot of work to get the sub-routines right in order to track objects of different colors. It seems the hue-saturation model works better than R-G-B - but I am not an expert in this area.

Good Luck all, see ya in Atlanta.

[iamnafets]

Thanks to Lynn (D) of 386 for hosting a Bittorrent tracker of our video. A link is off of his site or ours, I know I've got our server seeding it right now so if you need it fast, we got you covered . Thanks Lynn, much appreciated gracious professionalism at work at your team for sure.

[Dave Scheck]

Originally Posted by Dr.Bot

using it to position the robot near the loading zone

Just for the sake of argument, why bother if you can do it without the camera? This type of use can be compared to the IR beacons of 2004...it becomes an extra complexity for something that can be done using a sensor or time based system. The only way that I see to incorporate the camera, in this example, is to use it to double check where you are.

On a similar note, one problem that I'm sure people found, is that with the current color models, the camera sees blue the same as green. Granted, this can be compensated for (i.e. determining the size of the object), but once again it adds additional complexity.

I'm not saying that using the CMU Cam is necessarilly a good/bad thing, I just think that it, like many things, has its application. In most cases you wouldn't use a limit switch to measure angular rotation, you'd most likely use an encoder or pot. Its all about choosing the right and/or most efficient tool for the job.

[Wetzel]

Originally Posted by Dave Scheck

Just for the sake of argument, why bother if you can do it without the camera? This type of use can be compared to the IR beacons of 2004...it becomes an extra complexity for something that can be done using a sensor or time based system. The only way that I see to incorporate the camera, in this example, is to use it to double check where you are.

On a similar note, one problem that I'm sure people found, is that with the current color models, the camera sees blue the same as green. Granted, this can be compensated for (i.e. determining the size of the object), but once again it adds additional complexity.

I'm not saying that using the CMU Cam is necessarilly a good/bad thing, I just think that it, like many things, has its application. In most cases you wouldn't use a limit switch to measure angular rotation, you'd most likely use an encoder or pot. Its all about choosing the right and/or most efficient tool for the job.

One team at Annapolis used both mag induction sensors and the camera to tell when they were lined up with the goals. They had two of the mag sensors on the low front of the robot to "see" the metal of the goal base. They used the camera to see the yellow triangle in the middle of the goal.

I really wish I could remember the team number, their system was so nifty, and their tetra loader was so simple. They were at the DC scrimage too.(Maybe Dave can help me out)

Wetzel

[dradius]

Whoa, thats actually a pretty neat idea. see, we never really got the chance or the ability to test out and utilize other sensors on our robot. that would have been awesome, and possibly eliminated the "seeing" of the caution tape on the side of the field. hmm....must experiment!

[Dave Scheck]

Originally Posted by Wetzel

One team at Annapolis used both mag induction sensors and the camera to tell when they were lined up with the goals.

That's sounds like a pretty neat system. Was it used in autonomous, or did they use it in driver control with some type of feedback?

By no means was I trying imply that the use of the camera was a bad thing, I was just trying to get the idea out there that I think that it is good for certain applications and unnecessary for others.

[Wetzel]

I don't know about autonomous, but they did use it for driver feedback. Their loader just an arm that rotated upwards, and as it went past vertical the tetra fell onto the goal. That made it important for them to be lined up with the goal, and they had two different sensor systems to verify that.

I'll do some looking and see if I can't find the team number or a picture of the robot.

{Edit}
I found a picture, but no team number.

The mag sensors are the black and yellow boxes down low in front, and you can see the camera near the top in the middle. Their arm just pivots around the shoulder joint and drops a tetra on top of the goal. They were able to stack 6 on the center goal in two minutes at the DC Scrimmage .


Wetzel

[tbuo1]

I wonder if it would ever be possible for FIRST to add a rule allowing teams to keep their robots for an extra week but only allow software to work on it? Unfortunately, this is probably too unenforceable and would leave a large opportunity for abuse, and would extend the build season which a lot of people are (rightfully) opposed to. Unfortunately I just don't see how else autonomous modes can improve. Anyone have ideas?

Here's my two cents worth:
FIRST must support this (autonomous) part of the game better. Last year when we were given the IR challenge, we experienced many mishaps at regional competitions that made the IR sensors unreliable; the alumimum diamond plate around the playing field reflected the IR signal everywhere,the left and right IR beacons were sometimes crossed, and at one regional the beacons weren't even turned on! (they were eventually turned on)
This year, fifteen seconds is really tight! You really don't want to risk your robot getting damaged by making it go full speed during autonomous mode. There are a few things I would like to suggest to make this year autonomous (camera) mode more likely to be accomplished- First, allow more time (about 20 seconds). Second, the vision tetras should be placed randomly on the playing field with one on the left side and the other on the right side. Right now, some combinations are just impossible. Third, make sure the light is at a good level with no shadow on the playing field. Last, make sure no judges or FIRST volunteers place something like a Mountain Dew can around the playing field. Our robot actually locked on to a Mountain Dew can at Grand Rapids. If we are tracking green and yellow on the field, make sure green and yellow are only on the target!

[Beta Version]

I totally agree. This year autonomous mode is more of what it should be, something complex and challenging that wields a strong reward for completing. I think it's sad that at the sight of an actually challenging autonomous, most teams cowered and didn’t even try. While the vision thing is new and some things need to be ironed out, autonomous isn’t supposed to be some easy thing on the side. I love the challenge that programming a robot to see colors and do a complex task with this perceived information has put forth, but the time really is just a little to short. And our robot did the same thing, it kept staring at this strip of yellow caution tape on the field when looking for the goal. They should forbid yellow and green objects on the playing field unless its with the drivers behind the obscuring Plexiglas, or behind a covering.

[PURPLE!]

Originally Posted by Beta Version

I totally agree. This year autonomous mode is more of what it should be, something complex and challenging that wields a strong reward for completing. I think it's sad that at the sight of an actually challenging autonomous, most teams cowered and didn’t even try. While the vision thing is new and some things need to be ironed out, autonomous isn’t supposed to be some easy thing on the side. I love the challenge that programming a robot to see colors and do a complex task with this perceived information has put forth, but the time really is just a little to short. And our robot did the same thing, it kept staring at this strip of yellow caution tape on the field when looking for the goal. They should forbid yellow and green objects on the playing field unless its with the drivers behind the obscuring Plexiglas, or behind a covering.

Of course you can also argue that camo pants should be outlawed as well. A run away robot in autonomous mode chased one of our freshmen working in the pit. In the end, the cameras are just really sensitive and can be distracted by anything... Just like me.... LOOK THERES A PENNY *runs away*

[Beta Version]

But the only people wearing something like green camo pants (a team member) will be far away and elevated above the robot in the stands, or behind and out of site of the robot at the control station or human player pad.

[tkwetzel]

As we look forward to Atlanta, there are teams that have capped the side goals with vision tetras during autonomous and there are teams that have come close to capping the center with a vision tetra in autonomous. I think someone besides team 25 will own the tiki by the end of the Championship event.

Vision capping hopefuls for nationals (let me know if I miss any):
66
237
281
624

I want to see someone claim the tiki...don't let me down.

[Wayne C.]

Originally Posted by Beta Version

I totally agree. This year autonomous mode is more of what it should be, something complex and challenging that wields a strong reward for completing. I think it's sad that at the sight of an actually challenging autonomous, most teams cowered and didn’t even try. [/u]

Well I'm sorry you are sad about it but the fact is that there IS NO strong reward for capping the vision tetra. Even if you cap the center goal you only get 9 points. Any simple penalty will kill 10 pts. Maybe thats why soi many are so hot to get the TIKI I am offering.

Therefore any person who analyzes the game will see that you can pretty much offset capping the vision tetra by being a rapid capper. So the time spent developing an autonomous mode of complexity might be better spent developing something else.

Now everybody has all year to play games with the camera and develop all the modes they want outside the six week build period. I am sure if folks want to learn camera programming and if FIRST doesn't change it again, we will see a Kodak smorgasbord next year.

WC