Is Autonomous worth it?

[XenObliv]

Originally Posted by Caleb Sykes

There won't be many games that are "decided" at all this year. In fact, I would be quite happy with my team if we lose a match this year.

Right, I forgot that there is no winning matches. Well I am willing to bet many teams will be undefeated this year.

[Billfred]

Originally Posted by XenObliv

The time it takes to program the autonomous code, will all be worth it. One game can be decided by one point.

Actually, a maximum of three matches can be decided by one point: everything else is averages.

But since we're talking averages, let's discuss.

Last year's Palmetto Regional saw teams play nine qualification rounds. You get one more tote in teleop in one match (so, +2), and your average score increases by 2/9=.22. By the same math, that 20-point autonomous stack? 2.22 points to your average if you pull it off once.

In the quarterfinals, where you only get two matches? 2/2=1.00, 20/2=10.00.

If you're not scoring at least the robot set in autonomous, you'd better be lining up something pretty freakin' sweet for teleop.

[XenObliv]

Originally Posted by Billfred

Actually, a maximum of three matches can be decided by one point: everything else is averages.

But since we're talking averages, let's discuss.

Last year's Palmetto Regional saw teams play nine qualification rounds. You get one more tote in teleop in one match (so, +2), and your average score increases by 2/9=.22. By the same math, that 20-point autonomous stack? 2.22 points to your average if you pull it off once.

In the quarterfinals, where you only get two matches? 2/2=1.00, 20/2=10.00.

If you're not scoring at least the robot set in autonomous, you'd better be lining up something pretty freakin' sweet for teleop.

Your completely right, the teams that can do a auto consistently, will have a large advantage over others in terms of ranking. Autonomous is always worth the time during build.

[Oromus]

Autonomous is definitely worth it. Since qualifying is directly based off of points now, those 20 points mean a lot. Vision tracking, however, is a different story. From what I've seen, I don't see the point in tracking onto the yellow tote when you can just start in a pre-defined position (you can use measuring tools to position your robot now) and have a much fancier autonomous without having to spend a large amount of time writing vision processing code.

[ScottOliveira]

While the impact on points relative to a match may be smaller, because qualifications are based on average points instead of win/loss, those 20 points can have a much larger impact (there will likely be a number of regionals where one or more spots on the top 8 is decided by a margin of 20 points).

Also for teams with very specific subsystems, the hours are easier spent -if you have a specific group working on programming, then this gives them more things to work towards besides tele-op controls, which are frequently (but not always) relatively straightforward to develop.

[Kevin Sevcik]

At the very least, there's the standard disclaimer that scoring values may change at Champs.

Also, 20 points is 20 points. That's 10 totes or a RC on 3 totes or a RC+litter on 2 totes. I think that would be a winning margin in a lot of matches. If your non-autonomous programming efforts aren't going to added up to 20 points of improved efficiency, then your time really might be better spent on auton.

EDIT:And just realized I'm still falling into W-L-T thinking. It's apparently really difficult to shift into an average points mindset.

[Lij2015]

In my opinion the 20 points is worth doing the programming for, without vision tracking because I honestly think tracking is worthless this year.

Doing the three tote stack is the only reliable way to get any points in autonomous this year, at least at a regional. Also I think that scoring 20 points in auto will account for a good portion of your qualification average at a regional.

[one_each]

Depending on time and skill (we have all new programmers this year) we will look at implementing several auto modes that can be selected by the drive team.

We will start with the simple ones of robot set, tote set and container set where we move nothing or one game piece during auto. Once that is done (and if we have the time) we will look at doing an auto stack by ourselves (with the rest of the alliance getting out of the way).

I would love to see us do more, but we don't have the time / knowledge to work on vision this year.

[pfreivald]

I simply don't see an Autonomous Stack happening except by single robots picking up all three crates--and that's going to be a difficult task only accomplished by the best of the best.

The totes are about 21 feet apart, so you have fifteen seconds to acquire, move ~7', acquire-stack, move ~7', acquire-stack, move ~7' to get in the zone (Auto Zone!), let go, then move a very small amount.

All while avoiding your alliance partners and the Containers.

Possible? Sure.
Probable? Only at the highest levels of competition.

[jee7s]

Originally Posted by pfreivald

I simply don't see an Autonomous Stack happening except by single robots picking up all three crates--and that's going to be a difficult task only accomplished by the best of the best.

The totes are about 21 feet apart, so you have fifteen seconds to acquire, move ~7', acquire-stack, move ~7', acquire-stack, move ~7' to get in the zone (Auto Zone!), let go, then move a very small amount.

All while avoiding your alliance partners and the Containers.

Possible? Sure.
Probable? Only at the highest levels of competition.

Not to single out Patrick here, but this is indicative of a lot of "thinking before the change." There's an assumption here that you need to move to do each acquire-stack operation. Separately, there's an assumption that you need vision tracking to find the totes. But, I can just as easily picture a situation where a robot touches all of the totes with a mechanism at match start, pulls them toward each other in a known and predictable way, stacks them, and carries them into the Auto Zone and sets them down. Not that my team is planning on building that mechanism, but the stack can be made without moving the robot in (as a guesstimate) 5 seconds. Then drive forward and set it down. 20 points in 8 seconds. Boom done.

I know my team has struggled with the change. It's pretty dramatic, really opens up possibilities, but also takes effort to shift the frame of reference.

[pfreivald]

Originally Posted by jee7s

Not to single out Patrick here, but this is indicative of a lot of "thinking before the change." There's an assumption here that you need to move to do each acquire-stack operation. Separately, there's an assumption that you need vision tracking to find the totes. But, I can just as easily picture a situation where a robot touches all of the totes with a mechanism at match start, pulls them toward each other in a known and predictable way, stacks them, and carries them into the Auto Zone and sets them down. Not that my team is planning on building that mechanism, but the stack can be made without moving the robot in (as a guesstimate) 5 seconds. Then drive forward and set it down. 20 points in 8 seconds. Boom done.

I know my team has struggled with the change. It's pretty dramatic, really opens up possibilities, but also takes effort to shift the frame of reference.

Yes, that's true--you could. But "120 lbs" and "twenty-foot robot" are not things I expect to see, at any point, by any team. The teams that could pull it off won't, because they know the value of the containers on the step (which you won't be able to go get in any realistic manner if you're already 21' wide)...the teams that can't pull it off, well, can't pull it off.

[You also don't need vision tracking, if you know where the totes are (which you do) and you know where your robot starts (which you rather should.) Encoders and maybe a gyro should be sufficient.]

[jee7s]

Originally Posted by pfreivald

Yes, that's true--you could. But "120 lbs" and "twenty-foot robot" are not things I expect to see, at any point, by any team.

There are many reasons that a robot may weigh 120lbs, particularly if it's a forklift and needs a counterweight.

As for a 20 foot robot, one prototype 2789 is working on involves the most distant ends of our robot being 22 feet apart at match start. And, it has a valid strategic purpose. After we are done with that strategic purpose, the robot's envelope is reduced to 27"x42".

I'm going to botch the quote, but like Morpheus said in the matrix: "Free your mind."

[JamesCH95]

OP - auto is always worth it. It's 15s to get points, if you do nothing you are wasting 15s, a reasonably long portion of the total match time.

Originally Posted by jee7s

There are many reasons that a robot may weigh 120lbs, particularly if it's a forklift and needs a counterweight.

As for a 20 foot robot, one prototype 2789 is working on involves the most distant ends of our robot being 22 feet apart at match start. And, it has a valid strategic purpose. After we are done with that strategic purpose, the robot's envelope is reduced to 27"x42".

I'm going to botch the quote, but like Morpheus said in the matrix: "Free your mind."

So someone is trying to build the 'wings of death' robot that grabs all four cans off of the barrier first thing... interesting.

[Abhishek R]

Originally Posted by JamesCH95

So someone is trying to build the 'wings of death' robot that grabs all four cans off of the barrier first thing... interesting.

Team 71 makes its return?

[Kevin Ray]

Originally Posted by Abhishek R

Team 71 makes its return?

Wow! I was thinking the exact thing when I read about grabbing all three RC's at the same time.