MOE, if you are doubtful about auton bots getting to the stack in 2-3 seconds, then you didn't watch the Fla Regional. I think it was team 312 that did it consistently. I was totally amazed, but they did it several times (at least - I didn't see everything)

Regarding the possibility of slipping on the carpet, we had that difficulty in our testing before we shipped, so we created what we called "MotorSave" code. This is a special algorithm that runs in all modes. The last thing you do before you actually output a new motor command, is compare it to the previous motor command. Obviously, you must do each motor independently. You limit the maximum change in motor command to some value 10, 20, 30, 40, whatever you think is best.

The first thing we noticed when we did that is we didn't hurt our clutches nearly as much, so they last longer. The second thing is you can find the value that keeps your wheels from slipping on the carpet when you jam the joystick to full. Best of all, our driver says he has the best control he has ever had. (At first he was worried that the response was slow)

This goes a long way towards making auton work better. The other thing we do is change the battery between every match. Even in Elimination rounds we have the guys running to the pit to keep them charged.

Sparky takes ~4 seconds to get to the top and knock bins. I believe we could get there in 3 seconds given time to test things, but we do a couple things that slow us down.
First, we don't go full speed until we finish our turn and are heading up the ramp. Second, we slow down (yeah, that's right) just before we contact the bins. It yields a better bin-fall when you go slow. Third, we make a pretty wide turn. It took a lot of tweaking to get the turn radius the same in left and right-hand modes, so we won't be mucking around with that.

Final comment: Getting to the top of the ramp first does not guarantee a win. We have not been first in many matches we won. It is the remaining 1:45 that really makes the difference.

Final Final Comment: <rant> Why doesn't anyone really try to maximize the score by balancing the bin count on both sides?????
I mean, even if both teams have only a two stack left, you still really want to win with 20 bins on one side and 21 on the other - YOU GET WAY MORE QP THAT WAY. But every match I see the same thing, the team gets 95% of the bins on their side, and then make the ultimately unintelligent move of proceeding to clear the rest of the bins off the opponents side. Geez, those are 2-for-1 points you are losing. If you take out the 4x multiplier, you have divided you QP adder by 8! I really thought teams would understand that you absolutely WANT a close match. I am not talking about fixing matches - BOO! I just think before the drivers get caught up in the Battlebot mentality they should consider whether they want enough points to be able to pick their alliance at the end of the day. </rant>
Sorry if I drifted off topic a little *cheesy grin*.

[Disclaimer: the rant which preceded is not necessarily the view of Tucker HS or the rest of team Sparky, just the ravings of one slightly frustrated mentor.]

hehe yeah we hit the entire wall in three seconds or less consistently, but it isn't dead reckoning (definitely not line tracking either). ;)

Answer DR. Joe's Question
 

Joe,

To directly answer your question, the Thunderchickens give a preprogrammed steering angle (actually potentiometer value) based on where we want to hit the ramp (near, far, center) and full throttle until the gyro detects the ramp (threshold value). After detection, we can do various things based on what we select prior to the match. All the processing is done using a PIC and bit values are communicated to the RC and it uses a fairly simple case statement to determine what to do. After the ramp is detected we use a timer in the PIC to put wings down, up, change gears, and stop.

-Paul

At Mid-West the fastest were

16 - 2.75 seconds (my team woot)
217 - 3.1 seconds
111 - 3.4 seconds

We repeatedly timed these teams and they consistantly had these times.

Our Timers Correlate
 

We also timed teams, and got very similar results. Although we timed 16 at 2.6 seconds on a couple of occasions.

-Paul

2.5 Seconds = World Class Target...
 

So, with 4 of 5 Regional weekends behind us, it seems to me that 2.5 Seconds to the top of the ramp is the "World Class Target."

I know a lot of teams who have had less than stellar performances to date (Chief Delphi included) be aiming for better times in Texas. Time will tell if we've hit our marks.

From what I have seen on the web and with my own eyes, it will take a 2.5 second ramp run to win this thing (plus the ability to stay upright if an opponent meets you at the top plus the abiltiy to fight for the top at the end plus the ability to empty the opponent's carpet area plus the cool headedness to know what to do when plus, the most important ingredient, luck)

Luck and fast ramp runs to us all...


Joe J.

Fast Ramp a must ...
 

Joe,

We were discussing this with our team during the MidWest Regional. We decided to slow the robot down slightly to get better repeatability. We were asking ourselves, "How fast is fast enough?"

We came close to getting the answer at MidWest. Semifinal 1, match 2 we decided to go head to head with 16 (normally we go to the center of the ramp and put our wings down). 16 consistently went tight around the ramp getting the three columns closest to them (3,4,5 bins each). We decided to go to the inside to see what kind of stalemate we could create (did not put wings down).

Even though they beat us to the top by about 0.4 seconds, we got there fast enough to create a stalemate (2 or 3 bins fell into 16's scoring zone and most just went straight up ... pretty spectacular). What mattered more is what the drivers did after the stalemate. Anyone who was watching knows we got smoked when the human control started.

So, to make a long story short, I think anyone around 3 seconds will be very competitive and will put their drivers in a good position to win the match.

-Paul

Paul - I like your last post. You aptly point out that the match was won/lost during driver control. At the risk of being a little judgemental, it really seems that teams are obsessing about speed to the top. IMHO, although it is a factor, it does not presuppose the outcome of the match.
See you at Houston - we finally got school board approval for the trip, even though they have cancelled many other field trips due to the war.
Go Sparky!

we make it in and around 3.5 consistently, until we foolishly try to change our routine, in which case we dont make it, then lose..... o well.... we use dead reckoning, b/c w/ 2 wheel steering it is just easier than following a line, although it will look like we are following it.

Not true. We timed 2 Minute Warning at 2.75 seconds and when we posted it, Thunderchickens said they clocked us at 2.6 seconds. The key is a strong drive-train and dead reckoning. Many of our team-members have worked on Lego League teams and in that time we discovered that sensors usually, if not always, slow you down. Another way to increase speed is put more than one motor on a wheel. Lastly, curving, not forward, turn, forward drastically increases speed.

Re: 2.5 Seconds = World Class Target...
 

Check out:

http://www.soap108.com/2003/movies/il/
Q-matches 20, 33, 44, 56

Appears to me that 2.5+ will just earn a slap in the face; twice, in fact, when the second half of the "one-two" punch comes @ ~5 sec.

I like the idea of knowing what to do when => Go to plan "B" against #71

Jack J.
BTW - Joe, I put together a CD that includes the above - see M. Martus for your copy.

If you make it to the wall in less than 3 seconds my hats off to you!

If you draw out the line your bot follows from starting position, to the boxes, and calculate the max possible acceleration, including the turn, then you are pushing the laws of physics to get there in less than 3 seconds.

Im not saying it cant be done ( didnt you see the :c) in my post?)

if you are getting there really fast you must have sacrificed something, accuracy, weight (how heavy is your bot?), number of boxes you can hit...

Getting to the wall first is not the whole story - you want to take out 3 or 4 columns, push ALL the boxes into your scoring zone, do it consistantly

and still have a bot that is able to push other bots around, hold its own on top of the ramp...

I wouldnt be too sure about a 180 turn being faster than a V turn. either way your bot starts out heading south, and it has to accelerate to head north - whether its by a sweeping 180 turn, or a back and forth V turn - you are still converting all the momentum of your bot from one direction to the opposite direction.

Gettting their first makes a big difference:

1. you hit the boxes first, you send them flying in the direction you choose - sometimes over the top of the slower robot coming up the other side

2. you have more momentum when you hit the slower bot

3. you are on your way DOWN when you hit the slower bot - gravity is on your side

4. your bot is more stable hitting another on the way down, then being hit on the way up - less likely to be tipped over backwards.

Our general strat at the Canadian regional was to let our auton mode hit the wall fast (we got it down to about 4 seconds) - our flying squirrel spanned 5.5 feet and always took out at least 3 columns - the bot stopped in the scoring zone so the boxes were all 'keepers'

and then we pretty much stayed on our side of the ramp to get as many boxes in the zone, protect our stack, build a stack if we needed to

then hit the ramp at the end.

We were doing very well with this strat in Canada. We were ranked 8th in our second last seeding match, and were averaging 140 points (winning most of the time).

We lost our last match and dropped to 16th, but we were picked in the first draft by the 5th ranked team.

Kingman joined the "3 second club"
 

Kingman, Team 60, joined the "3 second to the top" club Thursday at the S. Calif. Regional.

They were clocked 3 times getting to the bins in 3.1 seconds.

The problem with your calculations is that you most probably did everything with linear movements which are very slow at turning. However, if you use radial movements (maybe a turret wheel or a deployable caster) your turning speed as considerably higher and thus it is very possible to hit the boxes in 2.5 seconds or maybe even less.

My philosophy on ramp speed is this: slow and steady wins the race, sometimes. You have to find out how well your opponents do on the ramp to accurately get on top and do some programing on the fly. I think if you go slower than usual and get more bins and are more accurate, then that's the best idea for slower opponents. But for a fast bot, a one time sprint would be worth it in the long run. I stand by our mentor Mr. Cook in saying that auton is a PART of the solution of the game. The rest is robot capabilities and driver control.

This game cannot be won by auton. Ever. I even have proof if you don't believe it.