Poll: Best Autonomous(read the 1st post before voting)

[Cory]

Originally Posted by GeeForce

There was a team that I saw on a practice field that actually used a sonar system to gauge where the spider legs were. Anyone have any more info on this? I can't remember the team number or anything else about it..

1425, maybe?

I believe they used similar sensors for their autonomous.

[Uberbots]

ours would have been good if there weren't major differences between our practice bot and our comp bot... it led to disaster (with the kPID gains). Also, our camera's smashing in SF1 at newton didnt help much either.

in any case, i found 111's auto to be very amusing... their method was awesome

[Karthik1]

I agree 111's autonomous was great.

[Tetraman]

when 340 was able to score after it got knocked...that was it.

[Squirrel Lord]

Originally Posted by Cory

1425, maybe?

I believe they used similar sensors for their autonomous.

It stopped working as soon as we got it to the Championship, but worked about 75% of the time in the finals at Vegas.

It seemed that 1902 had theirs working perfectly... Then in every match I watched at the Championships it failed, what happened 1902?

[Matt_Kaplan1902]

Originally Posted by Squirrel Lord

It stopped working as soon as we got it to the Championship, but worked about 75% of the time in the finals at Vegas.

It seemed that 1902 had theirs working perfectly... Then in every match I watched at the Championships it failed, what happened 1902?

The first time we missed in qualifying was due to a broken camera that we did not catch until right before heading out to the match, and the second time we missed was in our last match we decided to try out a brand new mode that went for the side goal and the rack was turned away from us just a little too far.


During eliminations we ende up having to replace one or both of our Banebots controlling our arm several times throughout the tournament and were not able to be as accurate as we had been all year.

Thanks for everyone that voted for us! The programming team spent many hours working on the robot an was a huge key to our sucess so I know they appreciate all the kind words.

[Lil' Lavery]

Originally Posted by Squirrel Lord

It stopped working as soon as we got it to the Championship, but worked about 75% of the time in the finals at Vegas.

It seemed that 1902 had theirs working perfectly... Then in every match I watched at the Championships it failed, what happened 1902?

Are you referring to Einstein? I think 1902, and others, had trouble with the additional lighting, although I may be wrong.

[DanDon]

Hey all, in case you didn't get a chance to see our autonomous mode in action, here it is, minus the scores from the New Jersey regional and the Championship event, those will be included later on.

http://www.chiefdelphi.com/forums/sh...d.php?p=621018

Thanks for watching,
/dan

[miketwalker]

Originally Posted by Lil' Lavery

Are you referring to Einstein? I think 1902, and others, had trouble with the additional lighting, although I may be wrong.

I know we had no issues tracking the light, the way our code is set up if it hadn't seen the light it would have stayed home (a safety measure we put in the code to avoid the autonomous doing something silly). From everything I saw I think that in the end the 1" tolerance rules probably was what hit us. I feel pretty safe in that for the fact that our camera alone has like an entire sheet of "pre-match" checks to make sure it's in specific alignment each time (since obviously a degree or two off can significantly effect calibrations) Something on Einstein was physically out of skew a little bit, but I couldn't tell you what or how to be honest. The light may have been an inch lower (which at those angles would effect the calibration the most) than it should have been or the chain a little taught on one of the spiders, I'm not sure... but something was different on Einstein from the calibration values we've used throughout the year at regionals/Galileo/practice fields/etc. and I think it caught us off guard.

At that point, when you are on Einstein, to risk an untested code change to hopefully fix that problem is too high of a risk for the possible benefit, hence why the same problem occurred in the same way each time. It's ashame, but that's part of the game!

[Doug Leppard]

Originally Posted by Squirrel Lord

It stopped working as soon as we got it to the Championship, but worked about 75% of the time in the finals at Vegas.

It seemed that 1902 had theirs working perfectly... Then in every match I watched at the Championships it failed, what happened 1902?

As engineer over 1902 and over auto mode, Einstein was frustrating. I don't think anyone made an auto mode in Einstein.

Reasons for us:
Round 1. We were having arm problems and needed to adjust. Fixed it.

Round 2 and 3. Arm ok. Didn't realize this until I reviewed the tapes. Our partner hit the rack just as we did and knocked the rack ringer over just enough to make us miss. both rounds 2 and 3, they hit just perfectly 1/2 second before us and since it was right next to us threw the ringer off. In hind sight we should have run our 5 second delay which is one of our auto options.

Going from 80-95% to zero in Einstein was and is frustrating, but all part of the game.

Now I got to decide between voting for 1902 or 386. 386 had an exceptional consistency in getting lower rack and would try again if missed and got hit and still made it. 1902 was fast, could do low and middle. Oh the big decisions.

[ldeffenb]

Originally Posted by Doug Leppard

Now I got to decide between voting for 1902 or 386. 386 had an exceptional consistency in getting lower rack and would try again if missed and got hit and still made it. 1902 was fast, could do low and middle. Oh the big decisions.

Thanks for the comments for 386! I can't vote yet as I haven't seen videos of all of the choices (the link offered in post 1 is down).

As for speed, I'd say 386 and 1902 were a tie provided that we (386) didn't need a retry. As for consistency, I'd have to give that one to 386 with a 13/14 hit rate (92.85%) on Archimedes regardless of what they did to the rack (and they seemed to twist and translate it EVERY time when we were on the field).

We only missed the first qualifier, but a simple distance tweak fixed that. We were drifting off to the left for a bit, but another small target tweak brought us back into the center. We ended the competition with 100% in the 7 elimination rounds on Archimedes.

As the alliance captain of the finalists on Archimedes, we were the backup 'bot for the division on Einstein. We did check the lighting and alignment on that field and found it to be within our tolerances (lighting was perfect, alignment was close enough).

I suspect that the "brand new field" on Einstein was the cause of 1902's failure to score. The other playfields had bungie's that were steadily stretching causing the spiders under the lights to skew steadily to the side for which we all adjusted our targeting. If you had checked Einstein closely, the spiders were very straight under the lights for the first time since all of our first regionals.

For YouTube views of our Archimedes autonomous along with our toughest and best runs at Palmetto, check out the following links. If you want to see ALL of 386's autonomous executions, check out the BitTorrented videos at http://ldeffenb.dnsalias.net:30049

http://www.youtube.com/watch?v=9iZxxcQADhM

http://www.youtube.com/watch?v=M_2BXB8aMU8

http://www.youtube.com/watch?v=McvLJym0lhg

Lynn (D) - Team Voltage 386 Software & Coach

2007 Winner of the Championship Innovation in Control Award

[Koko Ed]

Originally Posted by Cory

1425, maybe?

I believe they used similar sensors for their autonomous.

That's who I thought of but they turned it off for the championship.
Why isn't 1126 on this list?

[ericand]

Originally Posted by Koko Ed

That's who I thought of but they turned it off for the championship.
Why isn't 1126 on this list?

Actually, 1425 didn't turn it off. Our lower ultra sonic sensor wasn't working correctly (possibly due to a change we made to strengthen the front end). Looking at the video, I think it almost worked in one match after we made some adjustments, but that match a defensive auto mode pushed the spider legs towards us so we were too close to score.

We really didn't want to change much since we were doing so well in all other respects. In hind sight, the reward may have outweighed the risk.

Our system consisted of 4 ultra-sonic sensors. One low on the robot, aimed to detect the base of the rack. 3 right below our gripper to detect the foot. The basic program for auto mode went like this:

deploy the gripper/tube
start to move the tube to scoring height
start moving forward - timed -
this lets us ignore our lower sensor while it calibrated
and prevented false readings in case it saw the gripper deploy.
using the lower sensor - drive to within scoring distance of the rack
using the upper sensor pod, center the robot on the nearest spider foot
robot turns right/left to seek the foot
If foot found, lower gripper - release tube - wait for auto to end
else release tube regardless at 14.5 seconds

The upper sensors were read by a custom curcuit which fed back 2 bits of target data
to the RC. The data that came back represented the possible foot locations from the
sensor pod's point of view:
Center
Right
Left
Unknown

The original plan was to have more info go back to the RC but our first cut at the communications
did not work and we ran out of time.

[Sislith]

i vote 1126 as well, why are they not on the list they were consistent

[Jeff 801]

is it just me but is the web site down