Hello! Welcome back to the build thread. We will be posting shorter more regular updates in season, as well as updates on merch…
Our Github is always open to view and CAD will be posted regularly
7 Likes
Today was the first day of our kickoff, with an emphasis on strategy and game analysis. Read all about it and a brief update about our offseason below. Tomorrow, look for updates on general robot architecture, layout sketches, and prototyping for shooting, intaking, and indexing notes.
This year we decided not to consider archetypes limited by build complexity first. Instead, we decided to start by analyzing what an ideal robot is capable of and consider different architectures that meet those goals.
Below is the list of all robot actions. They are categorized into must have, quality of life, and ignore, the last of which are not listed. Quality of life is a gray area, but it generally means something we would like to have but are not willing to sacrifice simplicity and reliability for. Most QOL items will only be incorporated if we can find a clean solution to perform them with the core features of the robot.
Must have:
- Drive – 16 feet per second minimum free speed, ideally more
- Travel underneath chain – see chair game below for a justification. This requires our robot be no more than about two and a half feet tall in idle state
- Play effective defense – requires power and high weight
- Hang on chain
- Hang on chain with a robot already on it
- Touch-it-own-it wide ground intake of notes
- Score in hatch from chain
- Shoot into amp from variable distance
- Shoot into speaker from variable distance
- Shoot into speaker from stage
- Shoot game piece toward a location, i.e. closer to protected scoring positions so alliance partners can finish a cycle
QOL:
- Ground intake from both sides – very small improvement to cycle speed?
- Directly intake note from human player station – we see no issue with picking it up from the ground but if it packages so we can pick it up from the human player station directly then why not
- Automatic robot avoidance – our programming lead suggested autos that use vision to detect opposite alliance robots and grab center game pieces while avoiding contact penalties in auto
- Get safely off chain – in case of bad coordination or if we need to reposition in endgame
- Hang on a chain with two robots already on it – just in case of bad in match coordination
- Human player scores on mic – in an ideal world
This lists sets the standard for deciding robot architecture and starting design. We have yet to determine lots of information we need, including how far and how accurately we can shoot, how best to pickup notes and index them, and more. Reach out if you are interested in our kickoff plans and our new kickoff worksheets. Stay tuned for updates!
1 Like
We have created a list of three main auto paths we plan to work on in the next few weeks. These are ambitious, but we believe that these are the auto paths a competitive robot at worlds would be capable of. These set an end goal for our design and programming, but hopefully we can get a start on these in the next few days.
Path 1
This auto path shoots its preload and grabs center pieces immediately. Each center piece left in the middle is an easy cycle for the opponent, so the goal is to get to them first.
Path 2
This follows the same concept as path 1 on the other side. The problem is that it’s not easy to shoot into the speaker directly from the far side. Instead, we would pass game pieces close to the stage to be shot later. Notes scored in auto earn the same number of points as those scored while amplified, so being able to ensure three easy cycles has more value than finishing two cycles in tele-op.
Path 3
It makes more sense to us at higher levels of competition to have two robots focus on clearing out the middle in autos as they can do it more effectively when they don’t have to go back near the speaker. In this case, one robot is left to shoot the three notes into the speaker.
We anticipate needing a lot more auto paths than this, but this sets a good starting point and lets us make goals for the speed of our pickup and shooting to make these paths possible. Hopefully we can also add features like using vision to avoid contact with other robots while picking up center pieces in the auto.
2 Likes
For the gameplay in tele-op we have developed a rough concept of our cycles, although we plan to drive the drivetrains around to test these concepts in depth tomorrow. Aside from on-paper analysis of points and speeds, we used six members in office chairs on our field to simulate gameplay and make observations. We were tentative to try this as it seemed like a waste of time, but we found that we got a lot of applicable information from it, and it improved our intuition of the game. We will likely make the chair game a yearly occurrence.
Our big takeaways are:
- We must keep the amp on as much as possible. Drive coaches will have to coordinate cycles very effectively
- Triple offense is clumsy and inefficient, there is significant value in having a full time defense bot
- It will be incredibly difficult to effectively coordinate cycles to reduce traffic while still scoring in the speaker while the amp is going
- We have decided going into this that we feel highly competitive teams would focus on prioritizing match points over ranking points, especially in a game where earning ranking points is aligned with scoring highly. However on this note, the coopertition button does not seem particularly consequential
The fast cycle path is indicated in tan, the secondary cycle path is indicated in white, and the “third robot path” is indicated in gray. We found opportunistic “bump and go” defense to be very effective, so the gray route would be fighting traffic upstream for cycles and hopefully achieving some sort of effective interference. Defense seems to play a very big role compared to Charged Up and Rapid React because of the amount of traffic.
We anticipate high traffic in the lane close to the wall with the human player stations. There is not a lot of space for robots to come and go between the stage and the well at high speeds. Defense will be very effective when blocking the opposing human player station. In the chair game, we also found it very easy to bait penalty points through contact with the stage as teams try to get to and from the human player station.
7 Likes
A Brief Offseason Recap
Because we did not post much during the offseason, here is a quick recap:
- We focused on training and replacing the skills leaving the team, as well as creating more permanent training resources for programming, mechanical, and design
- We built an angled elevator robot, with our team’s first ever elevator
- We won a local off-season competition, Bordie Recharged, along with 2910 and 2046
- We worked on recruiting younger students and training since we had far more upperclassmen than freshmen and sophomores
- We welcomed new mentors to the team
We also expanded our technical capabilities quite a bit. Most of these things will be used in season, and we will expand them with upcoming posts as they come up:
- Custom swerve kinematics and odometry
- Process improvements with our CNC Router for fast cuts and an affordable setup
- Generative design in Fusion 360 (shout out to 1294 for showing us how) for lightweight high strength parts
- Elevator design and construction
- Orange Pi vision setup with multiple cameras
- State space control for simple mechanisms
3 Likes
I am curious about this stance, from my point of view scoring a single note and hitting a button potentially lowering the difficulty of your rp by 16% seems consequential.
4 Likes
Very nice thread and analysis. I do see some assumptions that might not be correct.
we also found it very easy to bait penalty points through contact with the stage as teams try to get to and from the human player station.
The stage protection is only valid during the endgame. Playing defense within the stage during most of teleop is fair game.
- Shoot into amp from variable distance
- Shoot into speaker from variable distance
- Shoot into amp from stage
The opening on the AMP is only about 4" deep. It seems designed to accept only upright notes. When you talk about shooting into the AMP, if your mechanism is shooting the notes oriented horizontally, it is very unlikely to work. Basically, shooting notes into the AMP from any considerable distance is probably very hard.
Not sure whether you had considered those 2 things or not, but hopefully this is helpful.
Stage protection can still apply during Teleop, as defined by G424, the rule you posted.
The 2 cases for a penalty:
the opponent robot isn’t in contact with the carpet
OR
the opponent is in the stage zone with less then 20 seconds left
Only the 2nd case actually requires that the match is in endgame. Since the first case doesn’t have a time stipulation, that means it applies for the entire length of the match, so while the Stage will not be a safe zone before Endgame, a team still has climb protection if they wish to climb early (or bait fouls)
And, perhaps more importantly especially early in the year, the cooperation bonus is the first tiebreaker.
1 Like
The original context was baiting penalties, I don’t think it’s very viable to climb the chain just to bait penalties.
Yeah, just to clarify: by shoot from angle and distance we meant from enough of a range to not have to spend time aligning. We don’t want to be limited to driving up to the fender and lining up to score. We would like to have a couple feet or range so we can approach score and leave quickly.
For the penalties, even in the last 20 seconds we anticipate robots staying out to cycle and then doing a last minute climb because an amped speaker scores more than a climb. In this situation it seems like an easy way to grab penalty points with some defense while still leaving time to climb.
1 Like
Most of our prototyping is complete! Here are some of the major prototypes we did in the last week
We went through a couple iterations of the shooter. On the 9th we tested this design. One important thing to note with shooters has been that compressing the whole torus when shooting is not as effective as compressing the foam on a portion of it, leading us to believe a vertical wheel design may be worth exploring. Predictably, bigger wheels lead to a higher contact speed and a better shot. In slow motion, we are discovering some slippage of the wheel on the note, but hopefully tuning the compression alleviates the issue. The slow-motion video is below
We are working on a vertical roller shooter like the one 95 posted and a higher fidelity motor driven shooter with this design. We will post those videos once we get a successful test.
This was an attempt at a shooter/manipulator we designed on day 2 of kickoff, but the compression and the wheels seemed to be fairly ineffective at shooting.
This was a test of an under the bumper intake. We added a piece of 1x1 tubing for the note to slide under to simulate the robot frame. Some discoveries from this iteration are that the flip-down should ideally have two sets of rollers to get two points of contact, as occasionally we would get the note flipping up rather than sliding under the “frame.” We also found ourselves ripping into the notes quite a bit, we may experiment with different rollers. As a whole, under the bumper intaking seemed reasonable from our testing, even if the specific implementation of it needs some work.
2 Likes
Prototyping and alpha bot update!
Vertical shooter was a big success. We decided to run three motors on this shooter, one for the kicker and two for each set of vertical rollers. Hopefully by using two motors in this manner we can have some control over the spin on our shot. More experimentation is pending.
Our indexer is also fairly successful. We are finding that the ring is compressed too much (7 inch width) for an effective handoff. Between solid rollers and squish wheels, we found squish wheels to be more effective at the moment because of the amount of compression generated. We are reworking the geometry to try to widen the indexer a few inches.
The intake slap-down initial prototype was not as successful. Though the rollers would move the note to the frame, it was not enough to squeeze the note through the rollers on the frame and squish it into the indexer. A redesign has been completed, videos to be posted soon.
3 Likes
It took a little bit of goofy but we managed to get this handoff off successfully. We will still widen the indexer but this gives us enough confirmation of our overall archetype to move forward with building the first full fidelity robot. To widen the indexer, we need more space from swerve module to swerve module. Unfortunately our business lead said no to dropping $1500 on mk4s so we’re flipping some stuff around on the mk4i to make space in the front:
It’s a little more jank than we’d like but it gives us enough space to widen the indexer out to 11 inches while keeping the under the bumper.
For the time being, alpha is off to programming for more testing and dry running auto paths. Meanwhile, we begin Beta assembly
1 Like
Under the bumper intake turned out to be a success. Adding the omni wheels on the sides and changing the arrangement of the wheels in the front helped a lot. Again, though its not ideal the wheels stick out over the frame, so while the swerve configuration will remain square the robot itself will be made a little longer. However, we are very pleased with this intake particularly because the omni wheels mean the corner of the intake can make contact with a note and it gets grabbed fairly reliably.
From the bottom the beta robot design looks like this based on our findings on alpha intake.
11 Likes
We are looking to make our first robot reveal since 2019 and love how your robot reveals are formated. I remember that you made a white paper on it, but I couldn’t find the link. Do you know where I can find it?
Design Team Updates
Our design (CAD) team has been working hard to use what we learned from our alpha robot to design an improved competition bot.
Here are the noteworthy changes from alpha:
1. Elevator:
- Continuous
- Belt run
- 17” inside clearance
- 25” total width
- 2 Neos
- 24 : 1 Gearing
- Clamping Bearing Blocks
- We were planning on using the inline bearing blocks but they were on back order so we pivoted to the clamping earring blocks
Our elevator is designed with two major requirements, those being strength for climbing and width to house our manipulator. Achieved using a kevlar belt run and 1 : 24 gear ratio. Additionally the 17” gap in between the elevator helps give ample space for our manipulator.
2. Manipulator:
- 3” stealth wheels
- 1.25” versa rollers w/ grip tape
- ¼” polycarb side plates
- ⅛” polycarb sliding surface
- 2 shooter Neos
- 1 kicker Neo.
- 1 Neo w/ Max Planetary for the pivot
- 75 : 1 pivot gearing
Our main takeaways from our alpha manipulator was the need for a low friction sliding surface (achieved with the ⅛ poly) and two handoff rollers instead of one to minimize the dead zone without sacrificing consistency. New additions include the pivot which consists of a short max spline mounted to a dead axle using bushings. The sprockets and pulley for the encoder are mounted to the max spline and bolted on to our manipulator to reduce the torque on the max spline. We have one of these setups on each side to allow the game piece to go through the middle.
3. Slap-down Intake:
- Under the bumper
- ¼” Aluminum
- 2” omnis outside
- 2” stealth inside
- 110 : 1 gearing
- 1 Neo w/ Max planetary for the pivot
- 1 Neo for the wheels
- Hex shaft pivot
For the intake we tried to make it as light as possible not only to make it easy to control for programming but also so that we could get away with the generally low load bearing hex bearings + ½”hex shaft for the pivot because we wanted to keep our motors off the intake. To do so we use sprockets on bearings bolted to the intake so the shaft can still run freely and then we use a belt run and a motor on the indexer plates to run the intake. Additionally the omni wheels plus the roller box allow us to intake game pieces even when we only have half of it under our intake. That along with the 10 in extension is really convenient for stealing center game pieces in auto and makes pickup easier in general for our driver.
4. Roller box:
- ⅛” box tube
- 2” squish wheels
- 1 Neo
- 1 : 1 gearing
While the indexer is mostly unchanged from Alpha (except for the versa rollers instead of 2” squish wheels) the roller box which routes game pieces into the indexer is changed. For starters, edits to our swerves allowed us to move our drive base cross bars out more, reducing the necessary compression. Additionally you might have seen that our previous roller box consisted of plates sticking out of the drivetrain. To make our Beta bot legal we instead use a box tube with holes, which will also hold our bumpers effectively increasing our frame perimeter to 28 by 30.
- Climb: We will make a separate post about this once it is completed.
Finally, our rookie members have been working on their own projects such as designing cases for our Orange Pi’s and a mounting system for our cameras. The mounting system that we designed minimizes the blind spots between each of our 4 cameras while also allowing the cameras to adhere securely to our robot’s frame.
Shout out to design team members Karthik and Adri for writing this post. Feel free to reach out to us if you have any questions about the design.
7 Likes
I’m not exactly sure where the white paper is (or if we made one). I have asked our imagery and media team for more information.
However, we did publish some of our imagery documentation on chief a while ago if that is what you are looking for. (Sushi Squad 7461 documentation release)
As previously promised, here are some merch updates. Read to learn more about entering our special merch giveaway!
1 Like
Mechanical update
This year’s machining went smoothly - but not as smoothly as we liked. Early in January, we faced issues with our new parts tracker system which caused minor delays and miscommunications. However, we were able to quickly resolve these issues and educate our team members on using the tracker to avoid human error. Let us know if you want a post covering the specifics of our part tracker.
Our tolerances regarding tubing have gotten better, thankfully, falling within a 1/8th inch. The Omio workflow has also improved, with aluminum and polycarbonate parts going from CAD to fully processed in just a day. We’re extremely grateful for this improvement as it has majorly increased our in-house production capabilities and reduced our reliance on outside sources. In parallel with an ahead order on Fabworks, we got all our plates in-house almost on time. (Use code 7461 for 5% off on your own fabworks order).
Assembly for Beta went smoothly. By putting all parts necessary in a “kit” beforehand, we made the grueling part of the assembly (digging through boxes and bins searching for parts) a lot less arduous. Assembly started as soon as all painted plates landed in a kit’s box, and every subsystem took less than two days to finish assembling. Integration was also light work - we put every subsystem on the bot in just a few hours.
Thanks to our mechanical lead Ed and mechanical member Ameya for writing this update. As always, feel free to reach out if you have any questions.
Robot Test
Below is a video of our robot test. We are still fine-tuning our control systems (which is why the intake is so wobbly). We will send you guys a programming and electrical update soon.
9 Likes