Your robot looks amazing! If you don’t mind me asking, do you have any fasteners/brackets holding down the top of the first telescope stage to prevent twisting? Thanks!
Great job! Looking forward to seeing it in action in a few weeks.
End of Week 5 Update
The team continues to make progress, but now it tends to be in less obvious ways, and sometimes is partnered with setbacks. We have been affected, like many of you, by the news about Falcon motor quality control issues, and we spent a large portion of this week trying to organize disassembly and repair of Falcons while still using the robot in other ways. Sometimes this meant a loss of work time for programmers or drive team, and that effort and expertise among our mechanical crew that could have been used better was instead redirected to Falcons. I estimate we’ve put about 40 worker-hours into this, which is… not making me happy. At least we know we are in good company. In the end we found that the use of Loctite was rare (only four had any, and that was not consistent). Two were also missing the shims, one had an M3 screw that was held in place by… air, I guess, since it didn’t have teeth that would engage in the housing threads, and we stripped the screws on one motor, which we have set aside and will deal with later.
We are not having great luck with the color sensors as a replacement for beam breaks in lining up the balls pre-shooter. We will likely go back to using the same sensors we had on our 2020 robot to resolve this issue; I will post an edit here later when I verify what those sensors were, and will let you know if this swap fixes the problem.
We finished the bumpers this week. With most of our frame perimeter taken up with component attachment points, we decided to use rivnuts on each corner, with a 1/4-20 bolt and washer through a hole in a corner mounting plate. We have used similar solutions in past years and we think this is the right choice for this year. It is simple, secure, and I think it looks cool. It also makes it very simple for two people to swap bumpers in less than a minute.
We are planning on spending tomorrow with a limited team going to the practice field. The goal is to get real work done on the climb, and to test autonomous and drive as much as possible. I currently feel somewhat behind on this, because we’ve been chasing too many gremlins this week. I hope very much that tomorrow is productive!
That’s it for now, we will keep you posted about tomorrow’s progress. Hope you are all doing well.
Looks great! How much does it weigh?
Here are some videos of our testing on our practice field yesterday. I’ll try to make these instructive.
We spent a chunk of the day testing climbing using the “Lego” method. This video demonstrates the amount of tilt we experience. While cadding we were careful to locate the CoM near the physical center of the robot so as to avoid tipping and swinging during the climb.
Climb test 1339 https://imgur.com/gallery/7j7bLUP
Here is one of our climb tests. As you can see there are still some issues to work out, including an imbalance between spools, and a need to slightly redesign the passive hooks to release earlier. There is still very little swing which is good. We are using the Thrifty Bot climbers. Shortly after we took this we broke one of the 3D printed secondary stage internal stops. I strongly recommend getting the upgraded (solid print) versions of these, and of getting extras so that you can replace yours if they break. Thrifty is a great company to work with and will help get you fixed up. Other than the fragility of the early versions of the internals they are an excellent product.
Driving over a bump 1339 https://imgur.com/gallery/MtXHWVJ
We put 4" wheels on this robot, with no belly pan and no center drop. As you can see it does not struggle to get over a 7/8" obstacle, in case any of you worry about your ground clearance.
Ball pickup and shooting 1339 https://imgur.com/gallery/0RQavJH
We are not really finding it difficult to shoot accurately into either high or low goal. The lack of backspin due to top and bottom rollers means, as well as the relatively low energy shot from the fender, that these balls are unlikely to bounce out.
Testing acceleration control 1339 https://imgur.com/gallery/UeSxcNJ
One thing we are finding challenging is fixing the instantaneous acceleration from having the most powerful drive base we’ve ever run. Even with a pretty low COG (9"), it wants to get a bit tippy when accelerating or reversing. Trying different ways to fix this in the code.
Hope this is all helpful info! Have a great weekend.
Have you considered putting something like tread on either of the hooks to increase friction? That way you’re passively dampening the motion and reducing swing.
Have you considered using the slew rate limiter?
Indeed yes, that is one of the things the programmers are implementing.
Regarding using damping material on the hooks, yes this is also one of the things we will likely do. We used gum rubber surgical tubing on the 2020 robot climber for the same purpose and it worked well. We will try different products soon.
Weight wise, CAD shows it as sub-115. Our veterinary scale went missing over covid so I just got a new one delivered today and will weigh it this evening. It doesn’t feel over 100. I’ll update you when I know more.
Well, at least now I know how much weight I can curl. Also this is with five Cargo balls in it, so all told we have about five pounds to play around with.
Thanks for posting this we are doing a very similar climb and it is nice to see some success.
I do have to ask can you pinpoint what happened when you broke the stops on your secondary stage of your climber? Was your bot hanging on them, or was it under any unusual loading? Just curious if there are movements we can design around or be cautious during.
HI, About your shooter, you have the wheels on the top of your hood working as a Backspin? And if the answer is yes, how does it works? Thank you so much, it looks so cool!
Climb looks great, definitely the best one I’ve seen in the wild so far… And plenty of room to pull time out as you optimize it.
Haha no. I am old and out of shape, young person.
I forgot that I had a video of this. We unfortunately skipped from step one to step ten in speeding it up, resulting in an early release, before the 3rd stage was further inserted into the 2nd stage tube. This caused a wild swing, which snapped the insert at its inside corner.
The top wheels are spinning reverse of the main shooter wheel, which allows us to control backspin (less is better) and also gives us some control over the shooting angle.
After speaking with the programmers I see that I was incorrect, partially. The error was that they took the climb motors out of brake mode. I apologize for the error
Thanks for posting the video. We haven’t had a chance to build or test our climber yet and this test illuminated that we need to consider the severe swing if we don’t tune the PID correctly. Was your previous climb (the more successful one posted above) just in brake mode with no feedback/feedforward? If so, it seems like brake mode might be all you need to get the system to be critically damped? (or close to it)
Yes, see my above correction:
Thanks for the clarification. Are you finding that only brake mode is needed to damp the swing? Or do you have a tuned feedback loop, controlling the position (angle) to get your results in the first video?