Time to the top...

[Joe Johnson]

I know there was a thread on time to the top of the ramp in autonomous mode, but this is something different.

I want to know what times folks have actually seen robots make it to the top of the ramp (defined as touching the first bin).

Having seen 2 weeks of regionals, what were the best times and how many bins did they typically knock down?

We have all heard a lot of claims by teams. I am interested in how reality matches these claims. I suppose that many of the published times are 90% wishful thinking and 10% stop watch.

If this information is available in another thread, please point me to that thread.

If it is available on a scouting site, also, I would love to see the link.

Joe J.

[KenWittlief]

from reviewing the tapes of the Buckeye regional, we hit the wall consistantly in about 4.5 seconds.

we were only beaten to the wall once, by a bot that made it in 3.

From the calculations we did on energy and HP - 3.0 seconds should be the fastest - if anyone does better than that, have their bot re-inspected - they are cheating! :c)

[Andy Grady]

Times between NH and New England went down dramatically. In NH the fastest consistant robot (team 58) was making it to the top of the ramp in about 4 seconds. 157 was clocked at about 3.5 seconds in one round in NH. In CT, 571 was the fastest robot which clocked in at nearly 3 flat to the top of the ramp. Rage (173) was second at about 3 and a quarter, 157, 175, 126, and 716 each clocked in around 3.5 seconds. I don't have the actual times in front of me at the moment, but they were pretty fast, with all these robots taking down 4 to 5 stacks at a time. It was pretty impressive. If I remember correctly also, I believe that Paragon (571) even clocked in slightly under 3 seconds in a round. I can imagine with more tweaking time, these machines will only get faster.

Good Luck All,
Andy Grady

[sevisehda]

There is a video somewhere on CD that is a 12min video of the autonomous modes at BAE. After watching it and the webcasts. A few teams get there in 2-3 second and get 90% of the wall about 99% of the time. These teams are rare but very good. The next group is teams that get there in 5-10 seconds and take half the wall about half the time. Then there are teams that do something else or just sit there. I'd say there is very little middle ground, the teams who get there fast are also very consistent while the slower teams are less consistent. However as always there are exceptions.

[KenWittlief]

fast and consistant is the trick.

If your bot is fast you cant possibly follow the line

and the gyro limit is 180° in about 3 seconds (so if you turn faster than that, the gyro output is meaningless)

if these teams are hitting the wall in 3 seconds they most likely are running completely blind (dead recogning with no sensor inputs of any kind)

if they hit the wall in the center every time that is impressive

going that fast your wheels have to be slipping / spinning

can we get anyone to fess up? how do you find the wall that fast?

[Paul Copioli]

We scouted Cleveland and timed every round on Saturday. Team 27 was consistently at the bins in 3.5 seconds. The next times were around 4.2 - 5.0 seconds (about 10 teams).

In Pittsburgh, we were the fastest (team 217) at 3.2 seconds on Friday, but we got progressively slower due to our drivetrain starting to bind (using bevel gears) and we eventually levelled off at about 3.6 seconds consistently on Saturday.

Team 45 was at around 4.0 seconds and team 548 was at 4.1 seconds by the middle of Saturday competition. Many teams at Pittsburgh were from the 5.0 to 6.0 time range. We have all the times from Pittsburgh and I will try to have a list ready by Great Lakes.

-Paul

EDIT: I just saw Ken's post and I will try to respond to how we do it. We have 4 wheel drive with front wheel steering. We have a separate microcontroller (a PIC) which tells the robot what steering angle to be at. The wheels stay at this steering angle until the gyro indicates that the ramp has been hit (gyro used as a trigger), then we straighten the wheels and deploy the wings. After we hit the stack we shift gears into low and creep down the ramp. We use the Chips and Drills to drive with 4" wheels. Our gear ratio from the chips is around 6:1. Our tread material has a coefficient of friction with the carpet of 1.5.

-Paul

[Andy Grady]

If your bot is fast you cant possibly follow the line

I'm not completely sure, but I think Buzz (175) did, and they were lightning fast. Either that or it was a curved dead reconing program. As for the gyro, there is nothing saying that you have to use the gyro the whole 180, so if you mix it with some other means of sensor, you can get up the ramp in about 3-3.5.

Good Luck,
Andy Grady

[Solace]

You wanna know the secret? build a practice robot.

Yes, we run completely blind. We didn't choose to do it that way because we knew it would be fastest, but more because we were too lazy to do anything more complicated. But hey, it worked out.

[KenWittlief]

if you run completely blind / open loop

then how can your bot hit the center consistantly

with battery voltage changing between matches, esp in the finals?

if boxes are in your way?

if a wheel slips a bit?

Im not saying it cant be done, but getting an open loop system like that to be repeatable is not easy to do.

[Solace]

oh yeah, and one more thing. The rear two wheels on our robot pivot in order to conform to the ramp (we actually won the industrial design award at UTC for that). This helps our bot to stay on target even if it hits the ramp at an angle when going 12 ft/s. Its also pretty sweet to be able to spin in a circle even if we're straddling one of the angled drop offs at the top of the ramp

[edit] here's a picture http://www.chiefdelphi.com/forums/pi...&quiet=verbose
[/edit]

I think that in order to get there really quickly, you need to be running on four motors. I've seen bots that can go faster than us using only two motors, but the time required for acceleration negates their maximum possible speed because of the limited amount of space.

[seanwitte]

The program only ran through once, but we made it to the top in about 13 seconds. We integrated the gyro to make the turn and stay on course, and a line sensor array we built to detect the leading and trailing edge of the HDPE surface so we stopped on top. I think we could trim maybe 4-5 seconds off, the robot favors the left-hand side and we were not compensating for it. We will never be the fastest to the top, but we'll make it every time.

For the 3-second club, what happens when a robot hits the stack at that speed? Is there any feedback at all? At that speed (8-10 fps) you're nearly blind because the RC can't keep up.

[Andy Brockway]

We use the gyro (Team 716) only after we reach the top and go over the other side and only to tell the robot to stop. The rest is dead reckoning. We started to use the gyro to increase speed at the bottom of the ramp but found it too sensitive to bouncing. This caused the robot to think it had reached the top. It is not good to stop on your way up!

We were pretty consistent at taking the end three stacks using the two supplied batteries. We were lucky and had 1/2 to 1 hour between matches to ensure we had charged batteries.

The best matches are the ones when two robots reach the top at the same time and pass each other! We did this in a match against Buzz (175). Totes everywhere!

[Gope]

At St Louis we were never beaten to the top, i suspose we hit the bins in around 3 seconds.

While not incredibly fast, team 547 hit the boxes in around 5 seconds hand very consistantly got EVERY single box.

[Solace]

There are no sensors of any kind on our robot (besides limit switches) it can be done - our bot is living proof of that. (ok, so maybe not living...)

Ken is correct. changes in battery voltage do affect our autonomous mode. That's why we always put in a fresh battery before every match. That's easy to do in the qualification matches when there's time to charge up between matches, but in the finals we borrow competition batteries from other teams so that we can have enough.

boxes don't pose a problem. we've tested it succesfully with 4 boxes lined up behind it. more than that hasn't been tested, but i'd be surprised if another team could line them up in 10 seconds.

The front wheels are pneumatic and don't slip, but the back wheels are skyways and are supposed to slip.

Our bot starts backwards at an angle, goes straight in reverse, then makes a small correction to be lined up correctly and goes straight up the ramp and then makes a small correction before heading up the ramp at full speed. we have another program that lets us go in an arc, but it takes a good full second longer to get there. I'll see if i can dig up a movie of that or something.

sean, by the time we hit the wall we're going 12 ft/s. because we don't have any feedback, the RC doesn't have anything that it needs to keep up with.

For those out there who are still doubtful of dead reckoning, let me say that you were all correct - it is very difficult to do correctly. It took us three days of tweaking after the code was initially written before we could get it to work reliably. The tweaking was kind of random too - because it was difficult to predict what a programming alteration would do, we kinda just tweaked it at random until it tested out correctly.

We usually stick to our own side in order to not risk colliding with our alliance.

[Solace]

above all else, a practice robot and a practice field is a must. We didn't even start programming our code until at least a week after the competition robot shipped. Thank god for 173 and their practice field.

Personally I think that this years competition is a bit biased against rookie teams. Veterans who have the funds to build duplicate robots and have access to fully sized practice fields have an enourmous advantage over those who don't . I'm not complaining (being a member of one of those lucky teams), but I do think that its something FIRST might want to look into for the future.