Build Update 2/4/23
It works. (No more cheese fries too)
62 Likes
112 Build Progress 2/4/23
Our robot moves!
We got our swerve working on Tuesday, so that’s great! We have the gearing set to fast, and it seems like we can keep it at that.
No new CAD updates, our feed isn’t done yet. We also want to see how the intake interacts with cubes first.
We got the majority of our claw done over the weekend, the spinning of the wheels is done but we still need to put on the pistons to pneumatically actuate the claw.
Our arm is also done! We have a single neo powering the arm at a 240:1 ratio, enough for the arm to swing 180 degrees in a second at max speed. Our arm is very short and pretty light, so that allows us to be able to have such a high gearing.
Our robot is currently 54.3 pounds without pneumatics, our feed or intake. However, we expect those subsytems to not weigh more than 30 pounds, putting us at ~75 pounds w/o ballast. We’re going to have a very light robot!
I don’t know why the youtube links don’t automatically play, you’ll have to click the youtube button in the bottom left to watch them 
14 Likes
Love the intake! How’s your experience picking up cones from different orientations?
2 Likes
Today was our first day with the intake mounted on a functional bot so we haven’t extensively tested the orientations it can pick up, but it seems to work fine with the tip pointing towards us with ±15° tolerance, as well as upright cones with a different arm and wrist position. More in depth explanation of our intake design process coming soon.
1 Like
Love it. The arm has so much energy it makes the whole chassis jump. Slick intake, too.
1 Like
Yeah, that jumping is actually something we need to work on. I think we can address it with our chassis ballast, as well as backing off our arm rotational speed a touch. This is just first pass bring up, so it’s all optimization passes until midwest.
33 days!
3 Likes
Intake Development:
Our intake mechanism has undergone many iterations to reach its current state. Our early iterations were heavily inspired by this mechanism used on actual highways:
This style of reorientation was pursued for about the first week of build season, with lots of testing of specific geometry to figure out what worked and what didn’t. Our earliest tests were posted early in the thread through this video:
Despite showing lots of promise initially, this concept was eventually dropped once early CAD was done for it and the OTB linkages for it looked like this:
While testing the “Cone Logic Gate” we were also developing several end effectors intended to be used with a hand off system. The most promising prototype was this:
This prototype taught us how well grabbing the cone by its tip using rollers worked, which proved very important for our current design.
After we dropped the passive reorienting design and had settled on robot architecture which favored a system which acted as both an intake and end effector, we had to come up with something completely different than our early prototypes. Eventually we came up with the idea to use horizontal rollers to act as a top roller for the cube and a pincher for the cone. We initially tested with just rollers, but found that the cone wouldn’t make it all the way to the pinching rollers. Our solution to this problem was adding polybelt between the wheels to push the cone past the dead zone. The first iteration of this concept looked like this:
The prototype worked even better than expected and met the constraint of being able to work off of both sides of the robot, so we decided to model a more finalized version. The first iteration of CAD looked like this:
It did everything we wanted it to do, but weighed 11lbs, well over our 5lb goal, so it underwent substantial weight loss programs to get down to just over 6lbs.
This was the first high fidelity version of the intake we designed. After testing we were pretty happy with how it was working, but found that it needed a way to ensure that the wheels didn’t contact the ground, and that the wrist mounting point needed to be raised to allow the cone to pass through. The iteration currently on the robot looks like this:
Videos of some of our prototypes can be found here:
23 Likes
I have to ask- on your frame you mounted the aluminum deck to the extrusion pieces and attached with many rivets.
How did you make the holes in the deck so they would line up with the hole pattern in the extrusion?
Thanks.
We first modeled the drivetrain in CAD, making sure the holes matched up, then used our CNC router to cut the belly pan.
3 Likes
Harrison talked a little about the CNC router here:
1 Like
Is there a public CAD for this manipulator by any chance? Looks amazing!
It is public on our robot CAD here: Onshape
8 Likes
Forgot to ask above… is the belt skipping fixed?
We’re using a similarly large reduction on our arm and currently have chain. Lash and eventual stretching is a concern. Considered belts but was worried about excessive skipping.
It’s 90% of the way there. The belts for the wrist haven’t given us any problems. However, occasionally when the arm goes from vertical to horizontal, the arm belt will skip a single tooth while decelerating. Since this motion sometimes lifts the rear of the robot off of the ground, it can be hard to tell if you a hearing the belt skip, or the wheels smack into the floor. Oddly enough, the belt doesn’t seem to skip when the arm goes from horizontal to vertical. The encoder for the arm is after the main drive belt, so this isn’t a show stopper, but it may cause damage to the pulley long term.
I don’t know if you looked at our CAD, but we went with the over built jacking block belt tensioner. It has a 10-32 dog point set screw pushing against a hardened steel stop so we don’t have to worry about the tip of the set screw digging into the aluminum. Once we tension the belt with the set screw, we lock down the assembly with four bolts and hard washers. The belt tension is enough to bow the carriage plates. That is why there are support tubes surrounding the arm gearbox.
To fully eliminate the belt skip, I think we need slightly higher tension, slightly lower deceleration torques, and slightly lower arm speed. I’m worried too much belt tension is going to shear the 3D printed pulleys or damage the bearings in the max planetary, but we’ll burn that bridge when we get there.
14 Likes
Thank you for the info!
3 Likes
Looks like y’all are printing the rollers. If so, do you find you need to coat the pulley in anything to keep the belting in place/powered? Usually the crown pulleys I have used are keyed with rubber but thinking that is overkill in this application.
1 Like
The polybelt rollers are used right off of the printer, nothing additional has to be done for them to work.
2 Likes
How much did you have to stretch the belts for them to transmit motion without slipping on the rollers?
1 Like
an option would be go the nascar route and fill the lowest frame rails with lead or tungsten. helps you make minimum weight (or max in our case) and alter the balance of the robot
1 Like
The length in CAD is roughly 17", if I remember correctly the length we actually cut them to was 16.5". There isn’t an exact science to it, mostly just picking a length a little below the desired size and testing.
1 Like