What would you do differently?

With the majority of teams seasons over, I was wondering what would your teams have done differently to perfect your robot.

My team wished we decided to go with an active gear mechanism because with our passive gear mechanism the gear would not always line up perfectly on the peg resulting in dropped gears. We would also improve our climb speed. The drivetrain worked perfectly and we are happy with that.

I wish we had gone for a smaller-footprint active gear intake than the one we had, allowing us more space for ball storage. Also, I really wanted to tune our shooter, but we never did get it during the season.

I personally wish we had decided to go all gears, which is ironic given that I was the one who was pushing for fuel shooting day 1 recognizing its importance. It was just too hard to execute. If we had gone all gears, we could’ve spent more time and weight going fast, like 15 fps fast, so that we could gear better. We also would’ve been able to do a ground intake, maybe even a human intake as well.

We actually modified our passive gear mechanism during the out of bag time before our first event to be an active mechanism after watching the other events in the first few weeks of competition. In retrospect I wish we had done it sooner because there were a number of bugs we didn’t work out until our second event that substantially reduced it’s effectiveness (worked great once we were done with it though).

I’ll also add that I wish we had built a floor pickup for gears, and looking back now what’s frustrating is it wouldn’t have even been particularly difficult to modify our ball intake to be able to “toggle” to a gear intake mode if we had designed it that way in the first place.

I wish I wasn’t as shell-shocked by 2013. When the game came out, we looked at the idea of tossing novelty flying discs and thought it was ridiculously hard and didn’t even attempt it.
Whoops.
Since then, I’ve been struggling to find the balance between aiming for specialized simplicity and leaving behind credible gameplay abilities. Initial prototyping of fuel launching seemed pretty simple; we overreached and tried to do too much rather than focus our energies on something more attainable.

tl;dr: I wish we had designed our Week 4 comp bot from the beginning and placed a priority on driving/refining that design.

One of our mentors put it well: “we built the best, most capable robot that we were capable of building” given our time, resources, space and experience. We focused on gears and climbing, no fuel, and we ended up with a capable gear cycler (typically 4-6 gears in a match, with a consistent auto delivery to any of the 3 lifts). We climbed in 9 seconds and we only ever failed to climb 3 times all season - twice because we ran out of time (and one of those because another robot was camped under our rope trying to deliver a gear).

We ranked 27th out of 60 at Ontario District Champs, so I think our strategy worked.

However, we overestimated the importance of not dropping gears. We assumed that if you dropped a gear at the feeder or at the airship, that was a Very Bad Thing that would block up the station until someone swept it away, if that was even possible in the middle lift. So we assumed that we’d better not ever drop gears, and neither would anybody else. Therefore, we underestimated the importance of a ground pickup.

We designed our active gear handler with an aluminum backplate, but a lot of our failures, particularly in autonomous, was when the spring would go through the gear, but fail to clear the hole in our backplate. We had a sensor back there to detect when the spring was through, and in autonomous, we would not let go of the gear until the spring was detected. (Part of the “we will NOT drop a gear” philosophy"). We would have liked more time to adjust the backplate configuration.

Decisions I think we got right:

  • AM14U3 drivetrain worked fine and was up and running very quickly
  • low ground clearance and Lexan sweep plate to push stuff out of the way and ensure we never got hung up on balls or gears
  • gear chute designed to work best from 5-10" away from the wall, so we were resilient to gears and balls littering the feeder station area
  • simple, strong climbing winch with extremely low gear ratio, ratchet for anti-backdriving. Maybe not the fastest, but very reliable

I don’t mean to sound braggy or over-competent, but I don’t think we would have done anything different. This season has been, by far, the best season ever for our team. For we built a simple robot that plays the game well. I think I can speak for all of our team when I say we are incredibly proud of this robot.

Do the things that I always say we need to do every year… doh.

  1. Build a realistic representation of the field for if you are depending on sensing the field. Plywood does not look the same as clear lexan to a photoeye, but it does to an ultrasonic. I should have known better, I have been through this before.

  2. The mantra is touch the game piece, own the game piece. We did get the floor pickup requirement right, and the clamp and lift works well, and great if you can pick up on the fly, but the feed station is always blocked, and you can’t just drive though it as we thought. Now we are working for roller intake. Touch the game piece, own the game piece.

  3. Weight… Always weight. We plan for 10 lb cushion in design, and go into the bag 18lbs overweight on bag day. Again this is not the first time, and we played week 1, and poorly the first day.

  4. Climber gear ratios, designed at 12 volts, wouldn’t lift at end of match at 8 volts. Simple fix, but poor performance that first day week 1, dogged us the rest of the year.

Things we did right, low CG, fast 2 speed gear box, 4 CIM drive, ground pickup, accurate shooter, and scraping the turret requirement week 4 of build, as it was not going to be complete mechanically or software, we definitely did not have the weight for it. (but we didn’t know that then)

I’ll second rethinking a clamping gear pickup. We wasted a lot of time scooping around or dragging it back to pick up.

Also, prototype before committing. We had a mostly-working climber, wide acquisition, velcro with a ratchet wrench, where the biggest problem was losing the chain halfway up.

Someone brought in a boat winch with a tiny acquisition and decided that’s what we were using. Ratchet could switch both ways, so we did it the wrong way one match. Velcro almost never attached during a match. We ended up completely rebuilding it for our second competition.

My team would have done a couple things differently, from my own opinion, these are what I would have done differently.

  1. I wish our climber was on the front of our ball intake rather then inside our hopper to make it easier to climb. Which also as part of climbing, I wish we would have gone with 2 775s originally instead of 1.

  2. I wish we would have gone with a shifting drivetrain to allow us to spring across the field but still stand up to defense when they came around.

  3. I wish we would have created our gear ground intake using a roller rather than a claw to make it quicker to grab gears.

  4. I wish we would have used a 1986 style ball indexing system, as currently, it’s similar to 118 with belts feeding to the shooter on the side but on the bottom of the hopper, we have a spinning blade essentially instead of belts.

Although this is just to modify our robot, if i was to 100% change the design, I would create a robot similar to RoboWranglers with their sliding hopper and intake for balls and gear, but instead of a turret, I would want to add a dual shooter similar to Cheesy Poofs (if I could dream).

If we had bit the bullet and decided not to even touch fuel at all, we would have had a much more successful season. Everything else was about right, other than some implementation details or resources drained due to trying to shoot.

Scrap the ball intake (just load directly from the hoppers), devote more space on the robot to our floor gear intake, and have the option to load from the gear feeder as well as floor intake right off the bat.

Lets see… I’ve got a bit to say here.

Gear manipulator: We should have stuck with our priorities and made a floor gear pickup, or if we didn’t do this, at least make a much bigger active gear manipulator. We are currently switching to active pickup, but we should have had this done before our first regional.

Climber: our climber is scary fast (2 sec) but one change would be to make it wider, so it easier to line up. (Think 118 before they changed it) Other than that I think we nailed the climber.

Shooter: We needed to go with a large open hopper instead of whatever we decided to do. We also should have optimized for weight so we wouldn’t have to take off our floor ball pickup. The shooter itself is fine though.

Overall changes to design, schedule: Take more time to prototype EVERYTHING and analyze strategy. Make sure mentors don’t overtake design teams.

less turrets

For my team, I’d like to see the prototypes get used by the software and controls (electrical) team in addition to the design and fabrication teams. The design/fab teams built a full scale (but static) plywood prototype of the gear handler mechanism, and we should have bolted that on top of last year’s robot so we could drive around with it and simulate shuttling the gears.

There were certain idiosyncrasies with our design that we all figured could be resolved with trained driving and software/sensor assistance - ultimately this didn’t work out as well as we’d hoped. This could have been de-risked sooner if we said “we should really just try it and see” rather than “don’t worry, that won’t be a problem when ___”.

QFT. There were a number of times this season where we didn’t fix a problem because we assumed conditions would change such that it wasn’t a problem anymore. That turned out not to be the case, and every little problem with our design came to a head at Finger Lakes all at once.

Built 340s Robot

I like many people underestimated how hard it was to shoot. Our prototype made it seem easy and it wasnt until week 0 when I really processed how hard it is to shoot 120 balls into the boiler. We very much did consider ignoring fuel all together in the beginning of the season as following the golden rule was something that seemed like a good idea. Sometimes your first decision when seeing the game rules is your best one.

Shooting is easy. Aiming is hard. :stuck_out_tongue:

Honestly despite our relatively low performance on paper, I was pretty happy with our design this year. The only things I think we should’ve done differently were more testing, making our practice bot truly the same as our comp bot, and maybe floor gear pick up as a stretch goal.

I think we could have devised a better way to index balls into our dual flywheels, a la CheesyPoofs with the rollers on the bottom. I also would have attempted to implement an active gear mechanism, since we didn’t realize the time investment needed to wait for a pilot to pull the gear out of our robot.