Hmmm auto at the end of matches. That would be interesting.

Personally I would not like to see a complete auto mode match setup. Part of the fun of FIRST is the human element on the field. Ya technologically it would be an awesome challenge but robots dont show the emotion of a human. I mean would a robot be able to jump up and down and cry and be so joyus after winning a match on its own...most likely not. But to see the emotion and joy(or pain) of the humans on the field is what adds that special intangible element to the games each year. I was lucky enough to drive for two years out of the four I was in it. Without the humans driving at all things just would not be the same..it would be missing one of those "intangibles".

have you seen AI? :p
(im sure wildstang could work this out too *think StangAI*)
i think it will be a combo of both as it was this year. drivers shouldnt be eliminated and wont. just as human players werent necesary this year, they still ended up finding a way for them to add to teh game.

Send that to DL or FIRST. I LOVE that idea. Mabe they should do that for 2005, and a skimmed down version for 2004.

As for general autonomy;
FLL: 2.5 Minutes of FULL autonomy
FIRST 2003: 0.25 Minutes of FULL autonomy (That's 10 times less then FLL)

I don't understand why you think this is the end of the world, if middle schoolers can do it, then high schoolers should be able to too.

Hehe, but when FLL goes isn't there only one 'robot' on the field at a time? With 4 robots, stuff happens :yikes:

If there were no drivers I don't think it would be good at all... You still need drivers to keep it interesting... Plus many teams were horrible with auto mode and some didn't even have one... Personally I wouldn't like it...

Full autonomous would scare me.. especially because i'm a driver.. if you recall at almost every regional for the first day atleast half the robots didn't have an autonmous mode

I kinda like the auto mode multiplier Idea

but I also would like to see it so that the game is more constructive than it was this year.. in years past you had to actually complete a task.. like get balls into a goal or something, or balance goals on a bridge

this year it was almost ( and I truely do mean almost ) who could push who around.. and knock over the stacks

just look at who won it all.. 111, a bot designed for pushing other bots around..

I hope its a lil more constructive next year

Full autonomous would scare me.. especially because i'm a driver.. if you recall at almost every regional for the first day atleast half the robots didn't have an autonmous mode

I kinda like the auto mode multiplier Idea

but I also would like to see it so that the game is more constructive than it was this year.. in years past you had to actually complete a task.. like get balls into a goal or something, or balance goals on a bridge

this year it was almost ( and I truely do mean almost ) who could push who around.. and knock over the stacks

just look at who won it all.. 111, a bot designed for pushing other bots around..

I hope its a lil more constructive next year

helpful info?
 

Based on my limited knowledge (keep any wisecracks to yourselves please :) ), it seems to me that ALL teams would be very wise to really work over the offseason to learn and improve their autonomous programming ability and use of sensors.

Consider this:
Dave Lavery has a degree in Computer Science, and did his graduate work in Comp Sci/Artificial Intelligence. He was the NASA Deputy Program Manager for Artificial Intelligence Research for six years and then spent 12 years running the NASA Autonomous Robotics Program. (turns out our resumes aren't as similar as I thought - ;) )
The very first year he is involved with the FIRST game design, autonomy appears as a critical addition to the game - added to the fact that Woodie and Dean continue to promote the use of sensors and programming A LOT. Anyone want to guess how big autonomy might be in next year's game? Anyone want to guess how valuable students who learn the sensors and programming are going to be to their teams next year?

And if that isn't some interesting info to help teams guess/anticipate some game aspects for 2004 - I've heard a rumor through the grapvine that "joysticks aren't expected to be in the kit of parts next year". Which I find kind of scary, but read into it what you will (basically, my thinking is refer back to the advice in paragraph one).

I hope everyone has found some much deserved rest and downtime,

Jason

what? no flightsticks? how will we SURVIVE????

:ahh:

::runs and hides in a bin::

I don't see that happening. Not next year, at least. Autonomous mode still needs time to develop for a fully-auto game to be viable. Think about it, given the lack of 100% reliable autonimous modes this year, would you want to watch a fully-auto game next year? There needs to be at least a few years of mixed auto/human games before FIRST should even consider a fully autonomous game, if they want any level of quality in the programming.

What Should Be Done With Autonomous?
 

I saw some posts earlier about making the hole game autonomous. I don't think it would work much mainly because one centimeter off in the program means distruction and robots woould hit each other a lot and also many teams did not even have an autonomous this year.
I think that if First wanted to do an autonomous that next year they should have the last 15 secounds be autonomous to make it so your drivers have to line it up...

Post your thoughts on the topic, Thank You!

-T-ton

I heard from a reliable source (also a very evil man) that making the autonomous period in the MIDDLE of the match was an idea with merit and was being considered.


Wetzel
~~~~~~~~~~~~~~~~~
*back to the depths I go*

I think a fully autonomous game would be great, however, I think at this time it is far too difficult for a high school competition, especially in a 6-week period. Here's why:

Unlike FLL, an autonomous FIRST game would likely have at least two robots competing against each other. The second one robot bumps into the other or disturbs the other in some way, both robots no longer have much chance of accomplishing their desired tasks... They have now both been disoriented with the field and have no easy way to reorient themself. Almost all robots this year in the 15-sec autonomous mode used this type of open-loop control... With preprogrammed paths to get on top of the ramp. This type of control would fail -- miserably -- very early on in a 2 minute autonomous game.

Now, the fairly obvious solution is to use sensors to keep oriented with the field (technical term: closed-loop control). This could be done with optical sensors, gyros, accelerometers, and maybe even GPS someday once it has good precision. Optical sensors would allow a robot to scan the perimiter of the field looking for something like a post with reflective tape. Once it finds it, it now needs to eather find another post, or use a second optical sensor from a different point on the robot to find the same post. Together, the robot can triangulate it's position on the field. I suspect that this process could be done in no less than a couple seconds even with the best implementations. Something like this or GPS is the only way I can think of right now for a robot to figure out where it is.

If that doesn't sound that bad, then now consider the robot turning after it has determined where it is. The robot now doesn't know which way it's facing, unless you use another sensor to detect how fast it turned. On comes the gyro. But in fact the gyro only tells you how fast you are turning. You'll need to integrate the readings from the gyro over time to get an approximated estimate of which way the robot is now facing. Now consider the robot accelerating. Now you don't know how far it is going and how fast. You'll need encoders on the motors (assuming your wheels don't lose traction at all) or accelerometers to estimate how far you've gone. These errors will compound over time, and probably every 10 seconds or so the robot will not have a very good idea where it actually is. So you'll need to recalibrate with the field very often, or even try to do it continuously if possible. Continuous calibration is very hard, because it will be difficult to locate the posts while the robot is in motion and calculate location, especially given the slow processors on board.

I have not yet mentioned programming strategy on board. How do you know where the other robots are? How do you know where any scoring objects such as balls are? You would either need a very simple game to avoid such questions, or you would need many more sensors and complex control algorithms.

Of course, you will need the processing power and storage space to be able to use all this information... which means giving the on-bot computers a major upgrade.

A suggestion: If there is enough interest and FIRST is to pursue this at some point in time, I think FIRST should provide the technology to tell each robot where it is located, what direction it is pointing, and the same information about all the other robots, and perhaps even about the scoring objects. This could be part of the playing field set up, and the information could be transmitted to the robot along in place of the human controller signals. Further, it might be a good idea to let the robots do their computing off the robot. In other words, have a laptop computer receiving feedback from robots and sending control commands to the robots. If these changes are made, I can see this working. If not, full autonomy is a long ways off for FIRST.

- Patrick

Our robot did almost everything you mentioned in this paragraph, and was surprisingly accurate. We could drive around for the full 15 seconds of autonomous mode and when we finished, our robot would be within 6 inches of it's expected location, and most of that was due to the robot coasting past it's last waypoint.

We were able to achieve this kind of accuracy this year with a very poor selection of sensors. If FIRST plans on having autonomy again next year, our team (and others I hope) will be encouraging them to provide more sensors in the kit or allow us to buy more sensors ($100 at Digikey didn't really buy a lot in terms of sensors).

With better types of sensors, it should be easier for other teams to make systems similar to ours with similar accuracy. Don't get me wrong, I don't think the idea of an entirely autonomous match is good. It would be amazingly difficult to build/program, and probably wouldn't be that exciting to watch.

Then again, I only expected like 1/4 to 1/2 of the teams this year to have an autonomous mode at all. I really thought that a lot of teams wouldn't be able to make it work or just wouldn't bother. I was clearly very wrong. Nearly every team had some type of autonomous mode, and the vast majority of those teams were at least reasonably effective at it.