Auto Strategies

Seeing how many different variables we need to take in during autonomous, I felt as though a thread specifically for Auto strategies would be necessary.

What strikes me about this year is the sheer variety of options, as compared to last year where most teams stuck to three for gear-loading. We can expect to see robots only crossing the auto line alongside bots that can claim 4 points/sec on their own.

This year, teams can have dozens of different (mostly sub-optimal) auto programs to run, based on:
-Initial starting position
-Auto line requirements
-Position of switch
-Position of scale
-Loading multiple cubes in auto
-Choosing to load switch, scale, or exchange
-Preparing position prior to teleop
-Coordinating with teammates to ensure optimal points

While the clear priority goal is to have a robot that can load into either side of the switch from the center(most controlled variables and best for auto RP), I expect most teams with autos to try for left/right to same-side switch, and slightly fewer to aim for the scale. An alliance doesn’t need two bots to load into the switch, and runs the risk of messing up each other’s autonomous.

What is your team planning for this year’s autonomous? What are you prioritizing, and how are you accommodating for other teams/strategies?

We’ll set up on one side or the other and cross the line for sure.
Now that we’ve decided to focus on scale, we’ll try next for a “load a cube this side scale if it’s our side”, and later for a “cross to the other side” and if that works out, a “score on the other side”.
We may also do a “load a cube in the switch” if it’s on the same side of the field as us, but the scale is opposite.
In any case, each of the versions we have will be selectable during setup.

As I see it, there are 36 possible combinations for an autonomous program. This is based on three starting locations, four target locations (L/R scale and L/R switch) and three delays (whether your robot moves first, second or third).

I intend to advocate boiling it down to these eight:

From position one or three (for a total of 6):
Place cube on scale if only scale or both scale and switch are on your side
Place cube on switch if only switch on your side
Cross line if neither are on your side

From position two:
Place cube on switch

These choices should avoid crossing paths and the resulting collisions that could occur.

If anyone sees any other viable options, chime in!

The key to this year is consistency. Though many people are very worried about extremely fancy autos, the key is to 100% nail down simpler autos. As you get this, continuing on to hard auto will be the next goal.

However, something that many people wouldn’t consider is what their alliance partner will do. Though optimally, every robot can run any position, that isn’t the case. This year, scouting robots for their autonomous abilities will be very important. Using pit scouting to learn what others can say they do and match scouting to verify their abilities to do so. Knowing which autos actually work with yours will be very important. My team did a similar tactic in 2015, Stronghold, using heat maps to determine where other alliance members shot from. This allowed us to pick robots that worked best with ours, but also know what the opponent’s potential was, developing strong counter strategies.

Based of the way our practice chassis was behaving in auto on Saturday we feel comfortable we can already secure a ranking point in auto for most switch positions .Now will have to see if the regular robot behave this well or make slight changes .Once the competetion bot is assembly an subsystem go on our programming team is looking to add the scale and maybe a 2 cub if time permitt

Based on our team’s past performance, I’m encouraging our programming team to work real hard on the “score the scale (or switch) if it’s on our side”, approach. So we don’t have to try to navigate much to get a cube in. I have a feelint that if most teams did this, then most of the ranking points would be got.

We seem to find a serious programming ability barrier to autonomous modes that require making turns. Like it’s an order of magnitude more difficult.

Well with this, you can make it better AND simpler by changing the ‘who goes first’ option, to a time delay. On the SmartDashboard, have a number input box for ‘Start Delay’ that is immediately fed into a time delay command or block. This allows you to accurately coordinate with your partners, and adapt to them. It worked great in FTC, and should work in FRC just as well. An adjustable time delay can help prevent autonomous collisions better than three set delays.

I agree on the time delay part, glad to hear that worked in FTC. Its going to be interesting this year with the variance in auto capabilities between partners.

I think all teams should consider the delay option in each of their autos, to avoid those pesky collisions. Also double check LL RR LR RL signal to avoid scoring on wrong plate.

4513 loves using point turns, so much so the middle wheels are 2in smaller so theres only 4 wheels touching on flat ground as to keep the CoG in the same spot (and not 2 different) when turning, thats how far theyre willing to get their autos right. The amount of planning they have in general already for auto is mind boggling to see.

I was in fact meaning a time delay of some length when I spoke of who goes first, second or third. I am not involved with our programmers and was not aware of the Start Delay command. Thanks for the tip. I will make sure our programmers are aware of it!!!

Now this…is the discussion I was attempting to get started the day of game release. Use time wisely, collaboration very early and often with all match partners (practice, quals., playoffs), use your added Technician or even programmers in the pits early on after the releases of match schedules. Do not wait until you are in queing lines. Use the queing line option for last minute changes and corrections only.

Avoid robot to robot interactions or collisions in auto like they (the robots), are each infected with the plague. Avoid the infection and have a great season in 2018. Or don’t, and allow the infection(s) to take hold, and collisions will be the norm rather than the exception.

It all boils down simply to this…1. FMS plate color randomizations at auto start, equals 4 possible targets (will always equal 2 Red, 2 Blue targets) (A04 takes the opponents switch out of play completely at auto start, so never ever program a robot to cross the centerline as penalty to a red card can be the result if you cross either far null zone lines, equals a subjective decision by a ref, which is never going to be good to force - as you may not like the result only you can avoid in advance). 2. 2 actual right targets -get in these and win, 2 actual wrong targets - avoid these always or lose, minimum play from far behind. 3. 3 basic robot start positions (R C L). 4. 2 limited longest/shortest paths to opposite side scale - multiple paths choices, end or field centers to reach the scale, 2 longest/shortest paths to switch. 5. then a million variations of robot setting spots, angles, starting delays or not.

And…all but the very first choice by the FMS you fully control. Choose to respond, but collaboration with all alliance partners early is the key. (Just a prediction).

LL, RR, LR, RL (Always disregard the 3rd character delivered by the FMS as it is for the opposing switch - see A04). There may be a penalty in your future if you don’t!

Getting there first won’t always score more points as points are accrued by seconds of established ownership, until ownership is interrupted . Auto is 15 seconds long. POINTS will continue to be accrued long into teleop period through to match end, if ownership is not interrupted.

Power Cube deposit delays could also be another high points scoring key. (Now, figure out how to accomplish that task!) Lol.

When has 2nd to accomplish an auto task ever been better possibly pointswise to accomplish? It could certainly be here if followed up by another. Or not.

This added new element is going to be so fun to watch! Yeah to FMS plate color randomization! Looking forward to it here.

My suggestion to teams is to try and limit their autonomous to as few different settings as possible with maybe a way to delay the start of your auto. Having dozens of different auto modes for different positions is neat, but quite often, tuning and getting a few or even one auto mode down will lead to more consistency and greater success. Even small things like the direction of the carpet can cause very large variance when starting from left, center, or right.

Completely agree with this. We are in the process of finalizing 2 to 3 different autos we will be able to run consistently.

  1. Cross Start Line —> stop
  2. Cross Start Line —> Switch---->stop
  3. Cross Start Lin ----> Scale ---->stop.
  4. Cross Start Line—> Scale----> Pick up Cube----> Switch ---->stop (this is a last play IF the other ones are consist.)

I would personally go with Switch THEN Scale on the last one during Quals. This is due to the fact that switch is a RP in Quals, and scale isn’t. In Elims though, I would go with Scale then Switch so that you hopefully prevent the other alliance. This would change with the other alliance’s and you alliance’s abilities though, with your partner getting the switch or scale, and consistency.

Much will depend on your initial starting position and the ability to determine the switch field state from that position. A further complication is how fast your robot can move and score and not collide with other robots. For a sophisticated autonomous, you need both optical detection of field state, localization and obstacle avoidance. (Hint, our team is working with a Nvidia tegra with Open CV, and an XV 11 LIDAR. We hope to have a 2 cube auto, but I would be happy if the robot just reliably would move forward in auto.)

So from position 2, the choices are just cross over the line or if you can determine the scale state attempting to score slowly on the scale. If fast you may attempt to load another cube from field

If you are in position 1 or 3 and you can determine the position is favorable attempt scale score. If super capable, after scale, you might attempt second cube pickup and positioning. If your initial position is unfavorable, cross the line and head for scale. My perfect Auto would be from a favorable 1 or 3 position, score on scale, pick up a cube, score on switch. While a 3 cube auto is theoretically possible, I think it would be a rare achievement, because of all the variables.

…all while avoiding your alliance partners.

Which is why I want us to drive forward, and place a cube in an appropriately lit platform. Or just drive forward, if there is none in front of us.