Auto mode woes.

[Rich Ross]

I've been reading several threads recently talking about autonomous, how rough play reminds them of stack attack, and other gripes. I have noticed a pattern though.

2003: game decided by autonomous for the most part. GDC made auto mode such that it was essential to winning.
2004: Auto mode was nice, but didnt do much with the bonus balls. Mobile goals and such were involved to a small degree, but the autonomous was not the deciding factor most of the time. Since there was no "winner", it didn't set as many people off, but it wasn't a huge success in autonomous robotics
2005: Much like 2004, no winner, no results. Few teams did much, and camera struggles made it even harder. Teams had little incentive-to-difficulty (too much work for too little a reward). Hanging tetras and such were involved, but not as much vision tetras as many had hoped.
2006: If you won auto mode, you most often won the match. Complaints abounded and the flames of frustration were hot. Many complained and said it was bad or what not, and it seemed to decide a lot of games, but at least it got people programming auto modes that were fun to watch.
2007: Auto mode is missing. Few teams try, and fewer make it. Those that do don't seem to feel the same type reward that was seen last year.

Now, people are griping so much about autonomous that it seems like we'll be on a cycle. We should decide as a community what sort of autonomous objectives we would like to see (in general terms) and contact FIRST, letting them know what we'd like. I feel like this would add to the competition beautifully.

What do you want to see? More auto mode? More bonus points? No auto mode?

Let me (and everyone else) know you thoughts. I'm seriously interested in what you guys think.

[Tom Bottiglieri]

I think this years auto mode would be better if keepers counted as two tubes in a row. For example a row with 5 tubes and a keeper would be worth 7 in a row, equaling 128 points. Also, 512 could be achieved.

[Joel J]

Originally Posted by Tom Bottiglieri

I think this years auto mode would be better if keepers counted as two tubes in a row. For example a row with 5 tubes and a keeper would be worth 7 in a row, equaling 128 points. Also, 512 could be achieved.

yea, or scoring one keeper, gets a ringer placed on the rack. I think by nats, the keepers will find their value (hopefully), as there will be two teams scoring them during the autonomous period.

[DanDon]

I, personally, am of the opinion that the only part of this competition that is really "robotics" is the first 15 seconds, and that that should be one of the main objectives of every game. The rest of the time, its RC car time, with us playing with overcomplicated remote control cars.

Maybe something like vex last year where there is a separate autonomous game that is weighted as much as the operator control game. I think that the best bonus that could've been given to autonomous this year, is that if you put a keeper on, you get a tube to the left, right, above, and underneath the one you placed, so if you place one on the middle row, you get 4 tubes, if you place on the top or bottom, you get 3 tubes.

Because it seems like this year's autonomous challenge is relatively hard, there should be a bigger bonus.

Now, sum people are saying that the bonus gets bigger if you put more tubes on, however that is a big conditional, and if you are stuck with no alliance partners that are capable of scoring, and you have 2 robots playing defense on you the whole match, keeping you to 2-4 tubes, the current bonus does not really give you anything too worthwhile.

I loved the fact that last year's autonomous was so decisive because it emphasized what I think is one of the most important parts of this year's competition, a part that seems depressingly absent this year.

/steps off soapbox

[Richard Wallace]

Thanks for the summary history of automode. I agree with the way you have characterized its importance through the years.

Actually I do like the auto mode challenge this year and I think auto mode success will be highly correlated with winning on Einstein. Three robots with strong drivetrains, effective claws and arms, and reliable auto modes can win on Einstein without a ramp/lifter. For such an alliance, the endgame will be about spoilers. But I don't think we will see that kind of winning alliance before Atlanta.

I also think that endgame (ramp/lifter) bonuses are too high this year, and that is the factor motivating over-aggressive defense. A lower endgame bonus would mean a relatively greater reward for efforts devoted to auto mode scoring.

[SgtMillhouse648]

I like the idea of the two in a row rule. For about the past month, we have been working like no other trying to get multiple autonomous modes working on a test robot, thinking many teams would atleast try autonomous. To be honest, i was very dissappointed with the autonomous modes this year. I personally thought teams like wildstang would have used the camera, but at the midwest regional we were the only team that could halfway get the autonomous mode working consistantly. There needs to be some sort of a bonus to add to autonomous. whether a keeper is counted as double, or a stock bonus like last year, something more needs to be done. I think FIRST thought the fact that a keeper cannot be spoiled would be enough, but it isnt. Hardly any teams have used the spoilers successfully.
Malhon

[The Lucas]

Here are my rankings for autonomous based purely on spectator value:

1. 2006
2. 2003
3. 2005
4. 2004
5. 2007

2006 was the pinnacle of autonomous. It had the right mix of excitement, scoring and difficulty. It was a full contact chess match. All the bots started in the center of the field so if you could score in auto you could aways try to hit a scorer. It was so dynamic (not just the same old routine) that it was memorable. My favorite memories were:

NJ - After being stopped in the previous Finals match, team 25 used a different auto routine (they had always used the same one) to avoid the defense and score. The whole crowd audibly noticed the new routine . How many other games did the crowd (not scouts) notice or care about changes in an auto routines?

PIT - After we (365) took the camera off for weight we played mostly auto D. we played against this one bot (which shot blindly), several times and we always sent it firing into the crowd. The next time they were on the other side of the field so I walked over to where the field reseters were sitting and said "Heads up! As soon the match starts there will be 10 balls coming right for you." My prediction was right on and they caught the balls with the laundry cart.

CMP - We got our camera back and working. In one match 462 fit us before we could get set in our shooting spot and nearly flipped us . Then our 1 or 2 wheels on the floor manged to turn us back on target. We started shooting and hit 5 for 5


2003: Who didn't love watching bots race to the middle and destroy that box wall?

2005: Introduced starting game pieces so there was scoring at least of the hanging tetra type almost every match.

2004: If you could/wanted to, knocking the bonus ball off triggered the ball dump which was pretty cool, especially if your partner caught all of them

2007: It would be kinda neat if 90% of the robot actually did anything. Most of the time everyone is just staring at blink orange light on 6 stationary robots. It is so boring it is comical. It makes the start of the match anticlimactic. And in those rare matches where a robot actually moves and manages to place a keeper, everyone is exited till they look at the scoreboard see 2-0, worthless.

I don't mind a challenge but this game took so many measures to make it random and difficult, it caused teams to not try, which is not inspiring at all. To those that tried very hard and succeeded, the points are exactly a BIG reward. It is almost like they knew that most teams wouldn't be up to the challenge so they didn't give out an auto bonus to make the game more "fair" for the autoless teams.

[PhilBot]

I'm with dhoizner!!!!!

I was totally sucked in by the little "production" that occured just before the finals at Nationals last year. Consequently I was REALLLLLLY looking forward to dealing with three different colored target lights this year. How cool would that have been?

OK, I guess you don't have to look at my bio to see I'm a programming mentor

I think the actual robot building part of FIRST Robotics should try to excite the budding mechanical, electrical and software engineers. At the basic level, anyone who can assemble the KOP can play and have fun. But once you get past that level, all three of the engineering deciplines should be tested. This means better structure, beter power and smarter controls.

Since I'm also a First LEGO League coach, I have to switch from FLL, where the entire competition is "autonomous", to FRC, where we only get 10 or 15 seconds.... High School is meant to be harder than Middle School... right

Seriously, seeing your robot perfom it's assigned task in autonomous is the ultimate "we did it" experience. That's where robots are going... right? space exploration, under-sea exploration... military robots.... it's all about autonomous operation....

I personally thought Aim High was great. A good Autonomous score gave you a good edge, and then being able to climb the ramps at the end iced the cake. Structure, power and control.

-------------------

So I'd like to see a challenging Autonomous mode with a couple of levels of bonus (like with the low and high goals). Then have a different bonus for a mechanically capable (powerfull) robot.

Finally, make these bonuses self contained (without relying on other robots). Cooperation is great, but since alliances selections are random, there is no real opportunity for "actual" design collaboration. It's even worse when you don't have an alliance partner. (FLL had the same problem with the Space Elevator this year)

[Priyadarshy]

autonomous mode doesn't have to be about scoring necessary, at VCU 2068 played stellar defense in the 15 seconds also my team 612 uses the autonomous mode to position ourselves for the rest of the game and we are usually the first team to score in the match, autonomous can be used for a lot more than just scoring a keeper

[Phalanx]

I've said this before and I'll say it again..

I believe the autonomous period should be lengthened slightly with more options and opportunities than a single task or two. I suggest allowing for multiple different challenges some harder than others.

Some can be done with simple dead reckoning; others will require the use of different sensors or combinations of different sensors to meet the challenge. The greater the challenge the greater the bonus.

This way teams that don't yet posse the technical knowledge and skills for the more complex tasks have a chance to learn those skills and succeed through the simpler ones. Using the success of the simpler ones as building blocks for the future.

The teams that do posse the technical knowledge and skills are now presented with a challenge that will push the limitis and test those skills.

By doing it in this way, everyone is given a challenge and a chance to succeed, as well as being given motivation to improve thier knowledge, skills and performances to meet greater challenges over time.

So make it a three tier autonomous challenge... a 1, 3, 5 or a 2, 4, 8 scale or similar.

This program in my opinion is designed to enhance and challenge the students. Not everything should be made easy, simple or plug n play. It's the learning, the challenge, the knowledge gain and the creativity that comes from it that is important.

[Doug Leppard]

I agree with the various years assessment. On the team I was on we had a good auto mode every year except 2005.

I went back and analyzed how our auto mode and ramp use affected the games at Florida regional. We came in 2nd qualifying and 2nd in finals. We made 13 out 16 tries at auto mode. Only once did another bot make auto mode when we did.

Looking at the videos and points of the qualifying matches 1 match of 8 the auto mode made the difference, if we hadn't made auto mode we would have lost. One other match it could have gone either way on winning or loosing. So without auto we would not have been in 2nd place in qualifying, it would have dropped us to 4th or maybe 12th if we would have lost the maybe one.

Turns out the effect on ramps on us was the same as auto, helped win one and one was a maybe. So auto mode was as helpful as having a good ramp partner.

So autonomous gave us the edge.

In finals, we would have lost a game in Semi-final and would have lost in semi-final if we had not made auto mode that game. We would not have moved on.

We went 2 out 3 in finals and lost the last match. auto mode failed, if we would have made auto or made it on the ramp we would have won at Florida regional.

So does autonomous matter, YES when playing tight matches it made the difference.

[pheadxdll]

It is quite a disappointment to see the state of autonomous mode this year. Most teams stayed away from it and for the teams that did, the reward of successfully hanging a tube, through all these obstacles, just isn't that great.

For our team, we spent hours perfecting autonomous mode. We were a little disappointed when we watched some of the regionals and saw how little it meant and how few teams tried it.

I think a successfully hanged tube during autonomous should have its score weighted. Maybe have the tube worth 10-20 points. Besides the points, its a real reward seeing your robot go by itself and hang a tube with 5 other motionless bots.

Another problem I have with this year's game is the ramp scoring. Some of you mentioned it above. The scoring for the ramp is just crazy. Bascially this game comes down to robots getting on the ramp at the end. If both teams do get robots on their team's ramp, the rack helps push one team into winning.

Due to some of the problems in this year's game, watching it is a bit anticlimactic. Its often easy to predict who is going to win by seeing who has the better ramp-bot. I'm not saying the game is boring, because I can't wait to get out there and compete. Hopefully next year's game won't have the flaws in this year's game.

[Dave Scheck]

I think that the problem with autonomous this year is that there are too many variables involved and not enough development time to deal with them. Yes, there are teams that can and will be able to accomplish the challenge, but the majority of teams just don't have the time nor resources to do so.

The variables that I'm referring to are 1) A target that can be rotated 2) A target that can be translated in any direction 3) The chance that multiple green lights may be seen at the same time 4) A relatively small target to hit (especially if more than one robot make an attempt at scoring).

In comparison, in past games, only the defense played on a team or other robots on the alliance were considered variables.
2003 - Wall of boxes was fixed
2004 - Ball tee was fixed. The moving goals were always close, and if you were 190, the bar was always in the same place.
2005 - The starting tetras and goal locations were known. However, the vision tetras started in one of a few random places.
2006 - The goal was fixed and even gave you something to aim at with the green light.

Notice that all of the years where we saw good autonomous action had tasks that were relatively easy. Once multiple variables get put in place, less experienced teams (in terms of programming) start to struggle.

The other thing to notice is that when the task allows for true dead reckoning, more things happen. That just isn't true in 2007 or 2005 (vision tetras).

Another other key that has been mentioned above, is the reward for having a successful autonomous. If the reward's not there, there really isn't much of an incentive other than the "cool factor"

This year would have been much more exciting in autonomous if a) The reward was greater or b) Variables were taken out of the equation - maybe something along the lines of fixing the location of the goal (i.e. no rotation/transliation).

So my advice to FIRST would be this. Keep it simple enough for less experienced teams to do something worthwhile yet keep the door open for more experienced teams to do awe inspiring things.

[Tom Line]

I don't think the motion of the goal has that much to do with it. With much of the tracking code already written thanks to Kevin Watson, you could really just plug and play the camera this year without a huge amount of heartburn.

Because you actually have to drive the robot this year - I suspect that many teams that didn't build a practice bot or didn't have an old bot to use initially were up the creek. I couldn't have imagined waiting till the week before to start working on Auton - we started working on literally the first day.

I think it simply comes down to risk vs. reward. Many teams saw that 1 extra ringer and decided it simply wasn't worth the work. I think that 80% of the time they've been proven right. A good set of ramps is far more beneficial than a working auton mode.

They should simply call this "year of the ramp and drive train".

[Doug Leppard]

Again looking at actual matches in the qualifying rounds, auto mode was worth as much as a good ramp bot.

This was a hard year and I wish it could have been worth more.

Reason we did so well in autonomous was hardware finished the bot a week before shipping and we put 60 hours into software.

Last year was our rookie year for 1902, last year had a hard auto shooting a ball into the goal and easy goal take them to the corner. We did that the easy one in about 8 hours and it worked perfect for over 30 times in a row. It was a reason we were 9-0 at Houston.

I don't think we could have done this year’s auto mode if we would have been a rookie.

I agree, I do wish auto mode for next year would be worth more but I am happy that this is worth something.