Autonomous end game

[Donut]

Originally Posted by Noah Kleinberg

It could be interesting if any robots behind a certain line at the end of teleop went into autonomous mode, but robots outside of that area did not, and then any robot in autonomous mode would have some multiplier on points scored during that period (i.e. game pieces scored by manual robots are worth 1 point each in the last 15 seconds, game pieces scored by autonomous robots are worth 5). Could lead to interesting strategies.

This is an idea I had earlier this season I'm a huge fan of. Have an option at the end of the game where a team can hit a button of some sort that will put their robot into autonomous control. Teams can stay in manual control and try to finish out the game; however teams that can successfully do the end task autonomously get a 3X multiplier to what they just scored.

Even better, you could set this up to be possible anytime during a match, so that during it you could go into auto-control to get your huge score bonus, then go back to manual mode (obviously this would require some new additions to the field system).

[Vogel648]

Originally Posted by Bongle

I don't think this is the case. If you had an infrared emitter or vision target in the bonus zone, then all teams would have to do in autonomous is spin around until they find the target, then drive towards it. Already in autonomous teams have to acquire the vision target, so that isn't terribly difficult. As a programmer, I can tell you that when the camera is working and you've got KW's code, it is actually very easy to program with.

As a programmer I can tell you that doing much more than we did this year in autonomous would require more than 15 seconds and probably more time to program. Understand that there are ways of finding the light that don't involve spinning your robot in any way, but also don't allow for a full 360 of motion. Plus, again what if something is in the way, etc, etc etc. If we had another 2-4 weeks to do this, I would say, sure maybe we can get something worked out. But as a programmer, I know that when you say, "it'd be simple", really what you mean is that it seems like it'd be simple but it ends up being way more work that it was really worth.

Also, I don't really see much of a pay off of it being at the end.

[LordTalps]

I'd love to see this, it seems like a barrel of fun.

[Otaku]

Originally Posted by Tt321b

This is a very interesting topic. I would like to see more autonomous but there are a few problems that could come up when at the end of the game.
1) Doesn't the RC need to be reset to re-engage autonomous mode?
2) How would the RC differentiate between the beginning autonomous and ending autonomous?
3) You wouldn't see any last second efforts to race back to your side of the field and jump on a ramp like in the last two years

1) As far as I know... No. Autonomous is dictated by the IFI Field Control. When the Field is set to Auton, the bots are on Auton.

2) The code has 3 modes at the moment -- "Initiate", which is a sort of "pre-auton, start up the camera/etc" mode, then "Auton", which is activated by field control, then "Teleoperated", which is also activated by field control. To differentiate between another auton mode and the first one, add another mode and call it "auton2" or "autonend", etc.

3) Perhaps next year's game won't involve ramps. Dave's already mentioned Tangential Quadrilaterals, so I wouldn't be suprised if we get a major change in the game for '08.

[Bongle]

Originally Posted by Vogel648

As a programmer I can tell you that doing much more than we did this year in autonomous would require more than 15 seconds and probably more time to program. Understand that there are ways of finding the light that don't involve spinning your robot in any way, but also don't allow for a full 360 of motion. Plus, again what if something is in the way, etc, etc etc. If we had another 2-4 weeks to do this, I would say, sure maybe we can get something worked out. But as a programmer, I know that when you say, "it'd be simple", really what you mean is that it seems like it'd be simple but it ends up being way more work that it was really worth.

Also, I don't really see much of a pay off of it being at the end.

This isn't anything you don't already have to deal with though. In both 2006 and 2007, other robots would actively attempt to block you, block off your vision, shake the rack, as well as knock you off course. Opposing robots would be aimed to hit you. That would be a point of failure of many autonomous modes, and would be one of the higher-level challenges for teams looking to make theirs more robust.

And I'm serious when I say "it'd be simple". The easiest would be to simply have your robot drive forward slowly for 15 seconds and have your driver pre-aim at the end of the match. It might hit another robot, but that's a problem in autonomous as it is now as well. For a bit more difficulty, make a simplifying assumption that the light will always be in view (if it is mounted where it was in Aim High, this wouldn't be a problem), and then you can do the slow spin and acquire. Much more difficulty might be to detect impacts and attempt to drive around them.


Here are some assumptions a potential autonomous-writer can make to make it easier, and removal of each of these bumps it to a higher-level
-There are no robots in my way
-The robot is pointed the correct direction
-There are no game field elements in my way
-I can ram into the end wall at low speed without damaging my robot

You don't have to make a autonomous mode that detects other robots and field elements, avoids them, and pirouettes it's way to the end zone. You can make a simple little one, and if your robot is not pointed the right way at the end of the match, just hit the e-stop before final autonomous starts.

If you look at 2006, there were several levels of autonomous:
-Drive-forward defenders
-Drive-forward, then aim again defenders
-Stay-in-place shooters that were pre-aimed
-Shooters that drove a little, then shot
-Shooters that could drive across the field, re-acquire the camera, then shoot
-Shooters that, while shooting, could detect an impact, stop shooting, re-aim, then shoot

IMO, lack of participation in autonomous mode this year was because it was simultaneously perceived as 1) not worth it and 2) very difficult. If you make it worth going for, then teams will, at the very least, make a drive-forward mode. Aim High had plenty of participation in autonomous, at least so far as the drive-forward modes were concerned.

The pay-off would be to have a mode that is at its base level slightly more challenging than drive-forward that isn't JUST drive-forward. Plus, real engineers have to deal with variability, why don't our programmers?

[Taylor]

I would argue that the lack of autonomous this year was due to the risk-reward assessment. Just at one regional I saw several robots get hung up or overturned by the rack during autonomous mode.
Furthermore, the reliability of the cameras created an issue. In 2006 our dumping robot ran a fairly successful offensive autonomous mode without the camera guiding us; in 2007, without the camera, successful autonomous was nearly impossible.
If the GDC introduces a "safe area" on the field that perhaps only autonomously controlled robots could enter, I would certainly be in favor of that. Maybe when a robot crosses a certain threshold, drivers aren't allowed to touch the controls until the robot crosses back into the teleoperated area...? Some sort of door or gate that restricts use to that area...?
This could get very complicated very quickly.

[65_Xero_Huskie]

The game for this would have to be something that would prevent your robot from getting tangeled in the game element (This year, the rack would cause problems).
Also, If there was a problem on the field then there would be serious problems at the end. If theres a problem wiht auto in the beginning it wouldnt matter because there would be no points scored or anything. If it malfunctioned at the end then there would be a BIG problem.
But, i would love to see this happen. It would make auto mode more exciting (like it was in 05)

[Cartwright]

I think putting Autonomous at the end of the game would mix it up a bit. Challenge the teams a little bit further. When they added the lights a few years ago, it created a whole new programming level, but having to do autonomous from an unknown spot on the field with the robot in an unknown position...it could be good.

[Daniel_LaFleur]

I truely believe that if the reward was there (IE enough points) then you would see some wild automomous modes at the end of the matches.

[lenergyrlah]

If a robot tips the way the game is now there's always a bit of risk that you'll get hit, but in general the other drivers don't want to drive on your tipped robot. With autonomous at the end I think we'd see a lot more damage to tipped robots.

Also if the game is anything like this year's game it would be WAY harder to score (how is the robot supposed to know which spiders are full & which are available?).

But if the game were similar to Aim High I think endgame autonomous would be way too easy as long as the drivers collected balls & positioned themself in front of the goal prior to autonomous. Then their autonomous would just have to run their scoring mechanism.

Also doesn't FIRST usually have some sort of bonus at the end for ramping or hanging or something? (I honestly don't know as the only FRC games I've seen are Rack & Roll and Aim High). I would think the point of that would be increasing collaberation between alliance partners (definitely a point in 2007) and perhaps also giving "simpler" robots a chance to earn points even if they don't have a scoring mechanism. Do we really want to replace this with an autonomous period (which we have already)?

Sorry for the long post -- the more I typed the more stuff came to mind.

[Ken Loyd]

Autonomous at the end or beginning of the match...I like the idea of both ends. I would like to see more points or advantages for solving the autonomous problem. A great deal of time and sweat is involved in making the system work and I feel the effort should be better compensated.

Ken

[Brandon Holley]

I think FIRST would like to see more developed autonmous modes before they would consider moving the autonomous mode or adding another..

There would be way more variables to consider and way more obstacles to maneuver if it were at the end, along with the added danger of having robots possibly go where they are not supposed to.

I love the idea, but I think we need to see a consistent attempt at autonmous modes from a lot more teams before we'll see the auto mode at the end.

[robostangs548]

It sounds like a good idea maybe in a few years, but for now, I think that there needs to be some more development in the programming field, and the programming technology. Look at the difference from 2001 to 2007 and how much more advanced not only the robot controller is, but also the sensors and cameras. These things have become more affordable, and I think by 2009 the programming technology will be advanced enough to complete that task. But for now, I think that it is way to complicated to work with, and what if a robot flips over, or gets stuck in a situation..... You could seriously damage your robot, and other things.... I definitely think that this would need some development, but I can definitely see this happening in years to come, just not now.... (Please not now!)

[Bongle]

Originally Posted by Brandon Holley

I think FIRST would like to see more developed autonmous modes before they would consider moving the autonomous mode or adding another..

They'd see them if they were easier or worth more. As it is now, the payoff for spending another week building/practicing is more than the payoff for letting the programmers fiddle with the robot for a week, so teams won't do it. If you look at 2006, just about every team had at least a simple defensive autonomous mode, which is all an first-year endgame autonomous mode would require.

There would be way more variables to consider and way more obstacles to maneuver if it were at the end, along with the added danger of having robots possibly go where they are not supposed to.

But for now, I think that it is way to complicated to work with, and what if a robot flips over, or gets stuck in a situation..... You could seriously damage your robot, and other things....

I don't see how there are more out-of-control variables. You can control your robot's position at the end the match, so long as the driver knows where he/she should be. You can't control the positions of other robots, but you can't do that now. If your robot is entangled or will cause itself to become entangled, there is always the e-stop. Autonomous modes as they are now are capable of damaging the robot, I've done it a few times. The key is communication between the programming team and the drive team so they know what it will do and when to e-stop. Perhaps a change in rules so the drivers can carry the e-stop button behind the line with them would work, with a rule that any robots e-stopped DURING (not before) endgame autonomous score no points.

Here are the cliffs notes of my counterarguments. Note that I assume that the teleoperated period and the endgame autonomous would be seperated by 5-15 seconds of decision time for the drivers to decide whether or not to e-stop.

Someone: The robot might get damaged
Me: They can easily get damaged with the current autonomous mode. I have personally written modes that have damaged our robot.

Someone: The robot might get damaged in a noticeable way during the match, and running autonomous mode will hurt it further.
and Someone: The robot can get entangled or otherwise be in a nasty position
Me: Then e-stop it before the final autonomous section begins

Someone: The robot might get damaged during the match in an unnoticeable way that is inflamed by autonomous mode
Me: How could a robot get damaged in such a critical way that the driver will never inflame it, but the autonomous mode will? Wouldn't this inevitably either be inflamed by the driver? Sounds like a mean way to blame programmers

Someone: The robot can be ANYWHERE on the field when it starts
Me: If your autonomous mode can't handle that, then train your driver to place it where it needs to be, and if he can't, e-stop it before autonomous mode starts.

Someone: Other robots can be anywhere on the field when it starts
Me: With some limits, they can be anywhere on the field now, too. Fast robots could be on the other side of the rack in 5-10 seconds to abuse you. In 2006 when you only had to cross a field, you were getting purposely rammed by other robots almost immedietely.

Someone: But other robots might bump/ram us in their modes
Me: They do that now. In 2006 in particular, that was a widely-employed tactic in autonomous mode, with sometimes 2/3s of the robots playing a blocking or knocking role at high speed. Solution: don't drive quickly and build a robust robot with large bumpers.

Someone: The programming/sensing tools aren't there yet
Me: I feel confident saying that even now, the limiting factor in autonomous modes isn't the sensors, tools, and hardware, but rather the limited experience most HS students have programming robust code, which isn't going to improve given the 4 year churn inherent with a high school contest.

[Richard Wallace]

I'd like to see matches begin and end with autonomous operation. That would motivate teams to think. Thinking is good.