Infared in Next Year's Game

[miketwalker]

Originally Posted by dlavery

Dead reckoning systems are wonderful - as long as you know where everything is located, and can predict with reasonable accuracy how well your robot will navigate. But what happens if your target location is imprecisely identified?

Line following systems are great - as long as the lines lead to where you want to go. But what happens when they don't?

Sooner or later, robots must make the leap from structured to unstructured environments. The only way to do that is to rely on sensor systems that do not require fiducial marks or calibrated hard points to guide the machine.

I'd personally like to see a GPS-like system. Just 2 or 3 beacons around the field that robots can use to assist in positioning. Then teams can use that in conjunction with other sensors to find the "imprecisely identified" targets, which if they're moving or something similar... it will make it a fun autonomous mode. I'd like to see a system where you must rely on more than one system like you said. Not just infrared sensors, or just dead reckoning, but sort of a mix (maybe you need to get to a certain point before your in a line of sight with the infrared sensor due to a wall/obstruction in the way at the beginning?) I'd love to see a really good challenge in autonomous... and have enough points devoted to it that teams will spend more focus on it... because I know that's why we haven't put too much emphasis into autonomous, there's just not enough value in it to do anything crazy... but if multipliers or something come back for doing something in autonomous... or even autonomous at the end of a match.

For example, I know one thing I really would love to see is where teams can choose when to go in autonomous. So, if say they want to be in autonomous the last 30 seconds of a match, they get a certain amount of points... however other teams don't necessarily go in autonomous as well, so it's like.. the robot literally has to be able to think and counteract what the other team drivers may do.. but if you can do it well... you'd be very well rewarded, making a very interesting twist to the game. So maybe something like... for every second you stay in auto (if you exit auto before the end of the match it voids though unless you restart it and if you restart it you only get counted for the last bit). Just an idea I've played around with in my head the past few months.

[Andrew]

Originally Posted by dlavery

Interesting ideas for summer research: Sensor fusion. Handyboard IR comm. Lego IR Towers. Pseudolytes.

OK, I'll bite. What's a pseudolyte?

[ChrisH]

Originally Posted by miketwalker

I'd personally like to see a GPS-like system. Just 2 or 3 beacons around the field that robots can use to assist in positioning.

You can do a GPS like system with what we have now. (x,y, and direction) What you need for location to be completely unabiguous is three sensors that can tell the direction to the the beacons relative to your robot's personal co-ordinate system. The sensors need to be arrainged so that they are never in a line with respect to both beacons. Though if you want to get fancy you can even accomodate that to some extent.

Only two sensors are required if you just want position and don't mind a little confusion when you cross the "equator" between the beacons.

ChrisH

[MikeDubreuil]

Originally Posted by MikeDubreuil

I might have to eat my Red Sox hat next year, but FIRST will never force us to use the IR sensors.

Originally Posted by Kevin Watson

Um, just to get it on the record, is that a clean or dirty red sock ?

Originally Posted by dlavery

Kevin, you supply the salt and pepper, and I will bring the ketchup. I am hoping that some condiments may be needed later this year....

Uh oh... The hat's clean right now, maybe I should put it away so it's clean for January I'm not sure what condiment would go best, A1.. A1 BOLD maybe...

[MikeDubreuil]

Originally Posted by Andrew

OK, I'll bite. What's a pseudolyte?

I believe the correst spelling is pseudolite. And it literally means pseudo-satellite, as in a land based GPS system. This link explians the pseudolite research being done at Stanford for such projects as NASA's Rovers.

[KenWittlief]

I hope they use IR next year - I was put off by it at first, didnt want to mess with interrupts while I was learning C at the same time

but when I realized you can use the IR beacons just by polling the interrupt flag bits, and you dont have to enable any interrupts, it was a walk in the park

they worked great for us - I just wish we had put the sensors on servos so we could track the beacons, instead of using fixed direction sensors.

Next year? Definately! bring it on!

[Ryan M.]

Originally Posted by Stephen Kowski

I believe the autonomous should be a little more important than this year, but I do not think limiting us to infared as the only sensor input would be a wise move. That would not be very open to creativity, and kind of takes the fun out of the autonomous.

Agreed.

When I heard the quote, I first took it to mean just that autonomous mode would be much more of a necessity next year.

[Kevin Watson]

Originally Posted by MikeDubreuil

Uh oh... The hat's clean right now, maybe I should put it away so it's clean for January I'm not sure what condiment would go best, A1.. A1 BOLD maybe...

Geez, I mis-read your original post. Eating a hat is more palatable than a skanky sock anyday (though the nylon portion can be a tad chewy). Suddenly, this isn't as much fun as I envisioned .

-Kevin

[EvanG]

I agree with most people. I have a feeling that it will be there, and if you use it, and it works right, your life will be easier, but I don't think it will be mandatory. Though at one point at the UTC regional, a judge came to our pit and made us promise we'd make one of our younger students learn IR for next year. So you never know.

[Ryan F.]

I've heard that there were problems with the infrared reflecting off the diamond plate. Might be an unseen problem by FIRST.

[Paradox1350]

Well, we know this much at least. Autonomous is here to stay. Which is fine. However, if the game is ever made entirely autonomous, with no drivers, I think that would take a LOT of the excitement out of the game. I hope that never happens. There has to be a human brain or four on the other side doing something.

Also, it seems that Dave WANTS IR to be mandatory. Which I don't think is a bad thing. Also, we can guess that next year, there will be some sort of non-stationary, or non-predictable element in autonomous, thus maing some sort of tracking system (like IR) necessary for all teams.

And if that's the case, IR won't be mandatory, but it will probably be the best solution.

[Specialagentjim]

Alrighty, here's my thought which seems like no one else has mentioned here, so this may hurt your brain a tad bit.

Imagine a "scoring object" that is wired up wtih an IR beacon emitting different codes. THere will be multiple "scoring objects" on the field at any given time, with different codes assigned to each one. Only certain codes will be worth points. SO let's say that "scoring object 1" has a 101 code while "scoring object 2" has a 110 code. Visually, you'd never be able to interpret which "scoring object" is which. BUT, using LED feedback on the driver's interface (wow, forcing drivers to use feedback, interesting...), and IR sensors on the robot, drivers could figure out which "scoring objects" were worth points.

Whaddyall think? I think it'd be really fun, and leave that extra bit of excitment in the crowd as they wouldnt know who won the match until the "Scoring object" codes were revealed.

[ngreen]

Originally Posted by Specialagentjim

Alrighty, here's my thought which seems like no one else has mentioned here, so this may hurt your brain a tad bit.

Imagine a "scoring object" that is wired up wtih an IR beacon emitting different codes. THere will be multiple "scoring objects" on the field at any given time, with different codes assigned to each one. Only certain codes will be worth points. SO let's say that "scoring object 1" has a 101 code while "scoring object 2" has a 110 code. Visually, you'd never be able to interpret which "scoring object" is which. BUT, using LED feedback on the driver's interface (wow, forcing drivers to use feedback, interesting...), and IR sensors on the robot, drivers could figure out which "scoring objects" were worth points.

Whaddyall think? I think it'd be really fun, and leave that extra bit of excitment in the crowd as they wouldnt know who won the match until the "Scoring object" codes were revealed.

That was the partial intent on my post about dave's other hint thread:

http://www.chiefdelphi.com/forums/sh...746#post232746

[danielkitchener]

IR is here to stay. Notice FIRST's pattern: bring in something new and make it easy, then make it harder. They did the same thing with alliances. THe first year it wasn't so bad. Then it got harder. Same thing with last year - no set requirements for autonomous. All you had to do was either go straight forward or do a 180-degree arc. This year, you had to find a 12" ball on the side of the field by going through a relatively narrow space and hit the ball off of a T surrounded on 3 sides (left, right, and top) by walls, as well as placing an opponent ball next to it so almost any inaccuracy would favor the opponent. I reckon (pardon the pun...heh) that they will do a similar thing with IR. THis year it was not necessary. There were other ways. Next year, I'm betting it will be a bigger part of the game, possibly even required. Like autonomous last year was not required. This year, you were penalized for not using it. Look for IR to become a bigger part of the game. I just hope they don't use jimmy's scenario (post#27)

[Max Lobovsky]

Originally Posted by dlavery

Interesting comment. What path of logic has led you to that conclusion?

I say that because from what i have seen so far, FIRST does what it can to mimic real life engineering situations, and in general, aribtirary rules are not placed on problems like "you must use this approach to solve it". The one plausible idea is something like the one suggested before that FIRST might put IR beacons on objects that are otherwise unidentifiable, but personally, i think this is unlikely for the same reason i stated before.