2019 Retroactive Build Thread - Galaxia 5987

I recently found an old email thread from 2019 with mentors from my high school team explaining Galaxia 5987’s progress throughout the season. Open Alliance was not a thing at that point, and we were not publicly sharing our progress, so the details of what we were working on were kept secret. I thought it might be interesting to share some excerpts in a sort-of retroactive build thread. If people like this idea, I have email threads from other seasons I can share as well.

For the record, here are our robot reveal video, technical binder, and behind the bumpers from that year. We ended up winning the Israel District 4 competition, and being pre-qualified for the Championship in Detroit that year.

Jan 9:

Happy build season! Now that you’ve had a few days to digest the game and come up with solutions, I’m interested to hear what your guys are thinking.

My team here is trying to do basically everything. I’m hoping they didn’t bite off more than they can chew.

Jan 10:

For the manipulators, we’re still prototyping so I don’t have anything good to suggest yet. I will say that I’m a bit afraid of using the velcro to pick up the hatch covers because I think the velcro will degrade over time and not be replaced as often as needed (like the springs in 2017). It’s probably fine if you just get them from the loading station though. If you haven’t already seen, the Ri3D teams have been posting videos and explanations of their manipulators so that might be a good jumping off place.

Jan 27:

I thought with week 3 wrapping up I’d check in with you guys again. We just about finished CAD tonight, and hopefully will place our large COTS orders and start manufacturing tomorrow. Theoretically we can manufacture everything in a week, get the COTS parts shipped here in about that time, and have the robots assembled by the middle of week 5. Realistically, I’d be happy with early week 6.

We bought a CNC router this summer, and have been practicing using it. I attached some videos to show it in action, and the kind of cool stuff we can make with it. It really helped us quickly iterate on prototypes because we can now make 3 or 4 iterations in the time it would normally take to make one. We can also now manufacture a lot of stuff (like the gearbox plates attached) in-house, which is quicker than waiting for them to be made externally.

We decided to go for a very risky strategy this year: basically try to do everything. We know that means that we won’t do everything as well as some specialists, but we think that getting one RP by ourselves and doing a good fraction of another is worth the risk. Hopefully we have the resources to pull it off. I attached some pictures of the CAD (it’s not the most updated version so it’s missing the climber, some axles, some bumpers, etc.)

Feb 8:

Two weeks to go; I thought I’d check in again. We have most of the parts for our practice robot manufactured, and it’s about halfway assembled. The students are working 9am-11pm today and tomorrow to try to get it finished. We just got the practice field elements from our woodshop sponsor and set them up yesterday. The robot has a lot of small parts with some serious weight lightening (via material choice and holes) at the expense of strength. There’s probably more than a thousand small 3D printed parts, tiny bearings, 3mm bolts, nuts, and washers, etc. Hopefully everything works together like it’s supposed to and it can stand up to the stresses of competition. Here are some pictures to show a render of what the robot is supposed to look like:

how it looks right now:


our CNC manufacturing:


and the practice field we just assembled:

We were planning on using leadscrews to lift the robot on four legs to climb to HAB level 3. Unfortunately, the supplier we were going to buy the leadscrews from is having some problems and probably won’t be able to get them to us by stop build day. They seem to be the only company in Israel that sells the leadscrews that we need. We’re hoping to be able to get them in time to assemble the legs after stop build day and attach them to the robot during the 6 hour unbag time before our first competition. If that’s not an option, we may need to rethink how we’re climbing. It’s an important part of our competition strategy, so hopefully we are able to get it done. Editor’s note: they didn’t

The programmers have been working on some cool new features. They’re using a special 3D camera from Intel that gives a distance measurement for every pixel. They’ve also been working on neural networks to detect the game pieces, and vision systems to detect the retro-reflective targets. With this they’ve gotten a test-bed robot to identify targets, plot a path to them, and accurately drive to the target with the press of a button on the joystick. I attached a video from their testing. They’re working on getting the robot to automatically update the paths as it’s driving to adjust for errors in measurement and inaccurate path following. The goal is to use this to ensure good alignment with the target in first approach to save time with realignment before scoring.

Feb 10:

My team here just about finished assembling the practice robot last night. There’s a number of small things that didn’t work as planned. Most of them got fixed some way or another; some will need to be fixed more permanently for the competition robot. Unfortunately the biggest problem is that our elevator winch drum doesn’t grip the rigging correctly, so the elevator doesn’t go up and down. We’re still working on fixing that problem; hopefully we’ll have an answer soon before we need to make those parts for the competition robot.

Feb 14:

We finished building our practice robot a few days ago and now programming is doing their magic with it. We’re still waiting for our pneumatic wheels to come in even though we ordered them almost a month ago. In the meantime we borrowed some from 1690 from their 2016 robot – the one that was a finalist on Einstein. We’re hoping some of this good luck rubs off onto our robot. Here’s a few pictures:

Feb 20:

Happy last bag day to you too!

We stayed up all night till 7am to finalize the robot. Unfortunately, it’s not as final as I would have liked. Our hatch panel holder broke during driver practice so we’ll need to 3D print a replacement. It didn’t work as well as I would have liked to begin with, so maybe we’ll re-design it a bit in the meanwhile. Because of delays with getting the ball screws, our robot can’t climb for the endgame. We now have them, and we’re working on getting the practice robot to climb so we can check the mechanics and test programming before putting it on the competition robot. I attached some pictures of the robot as it looked before bag last night.

We’ve got a lot of programming bugs to fix before the robot will be ready for competition. The programmers are doing amazing work, but they need more time to test all their code. They are using special 3D cameras and a whole host of sensors to completely automate scoring. The driver drives up to the rocket or cargo ship and presses a single button. The robot automatically finds all of the targets in view, decides which one to go to based on the game piece it’s holding and which game pieces are already scored in the visible targets, aligns itself with the target,and scores in an open target. All of the individual parts have been tested and work well, but they haven’t had a chance to put them all together. Hopefully they can work on that with the practice robot before we need it for competition. I attached a picture of the driver station view with annotations as the robot sees.

Feb 26:

Thought I’d share our robot reveal video. It doesn’t really show all of the stuff the robot can do because it was shot in about 10 minutes in the hour before bag, after our hatch gripper broke. You can see it here.

I’ll also share these gifs that show what the programmers have been working on.


Mar 6:

District 1 is over, so I’ll give an update. We finished ranked 14th, moved up to be 8th alliance captains, and got knocked out in quarterfinals by the 1st seed alliance. We also won the creativity award. Our vision tracking code works very well, but we’re running into some trouble getting the robot to follow the path needed to actually get to the target. Hopefully they can improve on that before our next competition.

As I feared (and expected) our climber was not working for competition. Climbing is extremely important for ranking this year, and teams that can climb every match rank relatively high even if they don’t win every match. The 9 points you get from a level 3 climb is also hard to beat doing cycles in the time it takes. In my opinion, the lead screws were a bad idea to begin with and we should use our withholding allowance and unbag time to switch to rack-and-pinion style climbing legs like many other teams are using successfully. We have a meeting to decide how we’re proceeding tomorrow, so the team will decide whether or not to follow my advice then. Editor’s note: they didn’t

In our last playoff match, we broke a critical 3D printed piece on our elevator, which caused a chain reaction breaking even more pieces. I did warn the team about using 3D printed parts in a high shock load environment like elevator bearing supports, but they didn’t take my advice. Now I think they will make the replacements out of milled aluminum or delrin. Unfortunately, the time it’ll take to fix the broken parts takes away from our ability to fix our climbing mechanism before next competition.

Mar 15:

I wanted to give an update after our second district competition. After a rather disappointing qualifications fraught with bad luck and reliability issues, we were ranked 22/36 and were right on the edge of not advancing to DCMP. However, due to some good scouting and smart strategy, we were chosen as the first pick of alliance #4. They were good at controlling cargo and climbing, but had significant issues with the hatch panels. Our robot was one of the best at hatch panels at the competition, but couldn’t climb and sometimes struggled with cargo. It was a match made in heaven. Together with a second robot that played good defense, we went on to win the competition. This is the team’s first win (one of our goals for the season), so the students were very excited.

Unfortunately throughout the competition, we broke a number of parts and had some significant mechanical failures. Our climbing is still very shaky and not very efficient. We only successfully climbed to HAB 3 once in competition, because the lead screw “legs” just aren’t stable enough at those lengths. Over the course of the event, we fell on the legs a number of time, significantly bending them, and the profile connecting two legs took extensive damage while they were “stowed” under the robot during the match (picture attached). Hopefully they will be able to come up with a better solution for climbing than standing on lead screws for DCMP, otherwise I don’t see us ranking well. Editors note: they didn’t

23 Likes

A poll, with absolutely no selection bias:

Would you want me to post similar threads from other seasons?

  • Yes, I would like to see more
  • No, get rid of this trash
0 voters
1 Like