@Jack Chapman has joined the channel
Get robot to level3 climb
@Andrew Peterson has joined the channel
@Jack Chapman set the channel purpose: Get robot to level3 climb
@Chris Rininger has joined the channel
@Cruz Strom has joined the channel
@Enrique Chee has joined the channel
@Ethan Rininger has joined the channel
@Harrison has joined the channel
@John Sachs has joined the channel
@Julian Bonifaci has joined the channel
@Kenneth Wiersema has joined the channel
@Paul Vibrans has joined the channel
@Violet Advani has joined the channel
@Justice James has joined the channel
@Ryan Olney has joined the channel
@Riyadth Al-Kazily has joined the channel
Question for the climbing team: Is the plan to climb from the L1 platform to L3, which is 19 inches? Or is the plan to climb from L1 to L2 (6 inches), then L2 to L3 (13 inches)? I believe the plan was to go straight up to L3, but I'm looking at it in the simulator, and it is a long way up, so double checking.
Looking at the CAD design that Paul made, it looks like the idea is to go all the way from L1 to L3 if we can. The piece that gets pushed down below the robot is a little over 24.5 in.
The robot climbs from L1 to L3 directly. Anything else takes too long.
Sharing... If there is some vision for how the cargo intake will assist with the climbing process, please let us know so we can factor into our requirements and design. Thanks.
@Micah C. Glasby has joined the channel
@Jack Scheiderman has joined the channel
@Sunny Gregson has joined the channel
@Jacob Preston has joined the channel
Here are the 2 1/2 inch solid pvc rods we need https://www.usplastic.com/catalog/item.aspx?itemid=32414&catid=733
Please check what we have in the R8ZZ bearings in inventory .
the proper gray pvc rod
we actually have the correct size bearings in inventory so we don't need those then
We have some of the bearings, but not enough so we will need to order more
we are going to need 12 more right?
Who is putting this order together ? Quantity , etc ?
I have already made the order and sent it to Kenneth
Please send me a copy of the email . @Andrew Peterson
I will not make it to the meeting today until 11, work on figuring out exactly what we need for the programmers to automate the climbing process until I get there
Once you figure out what we need, start figuring out where these things will need to go on the robot
Programmers may come to either of you, Ian and Jacob, so be ready to answer questions about the mechanism
If you finish everything I have given you to do before I get there, ask Cruze if you can help him with working on the robot chassi
@Binnur Alkazily has joined the channel
@Austin Smith has joined the channel
@Randy Groves has joined the channel
@Harper Nalley has joined the channel
@Lane Johnson has joined the channel
@Violet Advani this is the link to the metal that we need to order, we will need 3 feet of it
I added it to the online metals cart-however I would like to see if there's anything else that we need to order
We need 12 inches of 1" Schedule 40 aluminum pipe for spacers on the intake arms.
OHHH! We can do this! (only on the practice field of course;-) https://tenor.com/view/low-rider-hydraulics-bounce-cars-garage-gif-5103931
@Austin Smith @Ryan Olney ^^^ :wink:
@Paul Vibrans @Peter Hall and I would like to test the climbing cylinders to see if they work so that we can hopefully order the five additional cylinders by the end of the meeting on wednesday. Do you have any ideas for how to test them? Peter and I were thinking about putting something heavy on one of them and seeing if it could lift it. The problem with this would be building a mount for the heavy thing as well as reinforcement for the cylinder.
@Peter Hall has joined the channel
They are easy to test in the pull direction. Drill a 1" diameter hole in a wood 2x4 and mount it in a vise with the hole vertical. Stick the nose threads through the hole so the piston rod points down. Thread one of the clevises that Ulysses was making onto the rod. Put a pin through the cross hole in the clevis and tie 130 pounds of weights to it using a suitably strong rope. Add 60 psi air and time the lift.
I went ahead and orderd the additional 5 cylinders for climbing.
@Enrique Chee okay, thank you!
@Paul Vibrans so, do that with all four cylinders at once I assume, correct?
Just test one with a single air tank charged to 120 psi feeding a 60 psi reducing station and one air tank on the 60 psi side. Be prepared to reduce the test weight in 5 pound increments to 100 pounds.
I would suggest adding a flow restrictor to the exhaust port of the solenoid so that the weight will be let down slowly, or maybe just put some foam padding underneath to minimize the shock on the cylinder and mounting block when you re-extend the cylinder after timing the retraction.
@Paul Vibrans @Peter Hall another idea I had was using elastics to simulate the weight.
Also, doesn’t the telescoping of the climbing mechanism effectively double the weight of the robot from the perspective of the cylinders?
Our climbing is a constant force process. Elastics are linear springs where force increases with distance. The only accurate test is lifting a fixed weight.
The telescoping does double the force required from the cylinders but the test weight of 130 pounds compensates for this.
A maximum weight robot that has a perfectly centered center of gravity only has 39 pounds of reaction force from each jack. That corresponds to 78 pounds of cylinder force. Add a factor of 25% for 1/4 G vertical acceleration and you get a cylinder force of 98 pounds required with a lift time of 0.64 seconds. Friction will slow this.
@Declan Freeman-Gleason has joined the channel
@Violet Advani What you will need to figure out at the meeting is how we are going to mount the system to the chassis. By this I mean something like the L brackets at the bottom. As for the upper support across the front set of the climbing mechanism, I do not think we can get them to go across without interfering with the shoot mechanism. I would still check this with Paul just in case I am missing something on this. Other things that needs to be done is to finish the rectangular brackets, filing down the brackets that are already made, start figuring out cable lengths and start cutting the cable to these lengths, sanding down two of the wooden legs that still need to be sanded, and start figuring out how we can attach the drawer slides to the wooden legs. If you have any CAD questions go to Jack on the CAD team, if he is unable to answer your question go to Kenneth. I am sorry that I will not be able to make it to the meeting tomorrow.
@Darwin Clark has joined the channel
@Violet Advani After checking over my work I realized that the upper support across the front set of the climbing mechanism will, in fact, work, with 2 inches of room between the top of the shoot and the top edge of the climbing tube.. I will give you a drawing for the 2x2 in hollow tubes as soon as I finish up with them in CAD
@Andrew Peterson Will you be at the meeting tomorrow?
@Chris Mentzer has joined the channel
Check out post 14 to see a cousin to our bot: https://www.chiefdelphi.com/t/post-video-of-your-first-hab3-climb-attempt/343744
@Mark Tarlton has joined the channel
Nine tower legs are finished except for deburring.
Is that four per robot and an extra?
Yes, Thanks Paul
Something more like this might work as well. The advantage is that there's no play once it's latched. The real challenge is to add a latch and not require reworking of the arms and joints to integrate it.
something like this arranged in parallel with the pneumatics is an option. The idea is you have a telescoping assembly that doesn't latch until it reaches full extension and when it does latch, has strength enough to support the weight of the robot.
@Paul Vibrans Is it possible to rotate the front legs 90 degrees to move the cables away from the center of the chassis? The hoops are running into the cables and pulleys.
To answer your question Mark, no we can not. The idea of the cables is to link to two rows together so that when we go to climb the front and back will at least be parallel to the platform.
The locking button is more like what will work
I think the front cables are going away. There's an interference problem with the chute. My understanding is that the rear legs will be responsible for keeping the robot level.
Which cables and pulleys are in the way?
the upper pulleys and the cables running down to the bottom of the legs.
It looks like the hoops will be pressing against them
Do you mean the pulleys attached to the piston rods?
The answer is absolutely not.
Ok. what happens if the front legs are rotated?
Since the crossing cables for the front legs are going away, it looked like the leg assemblies could be rotated one way or another to make more room.
The wheels at the bottom of the stalks can't rotate 90 degrees because of the direction of travel. The stalks and guides are designed around the orientation of the wheels and the towers would need to be changed. The towers are made.
do you really need wheels on the front? I thought the front legs would retract as soon as the arm goes down
If we don't change the legs, is there going to be a problem with the hoops rubbing against the upper pulley and the cables?
OK, well, here are nearly 1cm locking tube pins: https://www.amazon.com/Paddle-Button-Spring-Locking-single/dp/B00N01CLW6/ref=sr13?ie=UTF8&qid=1549339973&sr=8-3&keywords=locking+tube And here's a source I found for telescoping metal tube: https://alcobrametals.com/products/telescoping-tube Are you guys thinking the tube goes along side the cylinders or the cylinders go inside the telescoping tube?
I don't know without seeing the extent of the interference. The piston rod and its pulley are relatively stiff so I would expect both the pulley and the hoop to damage each other. The cylinder foundation could become distorted as well.
my thought was the tube is parallel to the pneumatics (on the inside) so that the pivot points for the pneumatics and the tube are identical.
Ok. I'll try to skinny up the hoops, but they will flex more as a result.
Can you trim the hoops and add steel flat bar with a much greater modulus of elasticity to make up for the removed material?
possibly. Especially since I have to cut and glue the hoops at that point, i might be able to insert a stiffener at that point. I'll keep working on it.
Can you send me a .stp file of the hoop and a datum for locating it in the robot? The datum I use is the mid point of the line of contact of the center wheels with the floor. The towers are symmetric about longitudinal and transverse lines through my datum.
Not seeing it. Seems like there are timing belts & pulleys in the way on the inside and probably not enough room on the outside for parallel mounting like that. I think it would need to be mounted below the pneumatics to the outside of the drivetrain wheels. Or we could make the intake a bit narrower to give us room on the outside of it.
it would be just inside the stacked pulley. there's a clear shot from there back to where the pneumatics mount. not in the center of the arm but the space between the pneumatics and the legs
so a tongue of metal linkage connects there and extends back to the telescoping tube with the button lock that will lock when the arm is in the full down position?
Ok. just got it integrated into kenneth's robot and it is the BACK cables causing problem. I can work around that, I think. I'll send you models anyway. I'd like to slim down the hoops but I don't know how much I can slim it down.
I understand the suggestion now, I think. It seems a bit much- the telescoping tube moving back and forth every time the arm moves back and forth, but if it's our best idea then should we order a few different button locking tubes (e.g. https://www.amazon.com/Stansport-254-Telescoping-Tent-Pole/dp/B004Z10D16/ref=sr18?ie=UTF8&qid=1549342688&sr=8-8&keywords=telescoping+pole+button+lock) and start experimenting? Climbing won't work without a lock-down mechanism, correct? We're sure the pneumatics will not be strong enough?
or you could have the nesting square tube the entire length for strength and stability.
here's a different direction that you might be able to do on top of what's there now
(I meant to send this earlier but forgot to copy the link)
you need something stiff enough that it won't buckle under load. the simple button at the hinge might have worked but I didn't see room for it
Another idea is to put a notch in the arm right at the pivot point, so that a latch would press straight up into the notch when the arm is fully down.
That's more what I had in mind, except higher up.
Something like this tied into the arm pivot mount
that latches onto the arm here
The clip was what I was thinking for it, but just a piece of flat bar with a spring or elastics attached to pop it into position and then have a hook to connect into a cylinder that comes out of the arm around where you've indicated. However I think we're going to need something pretty strong to support the weight of the robot
that's the problem with putting the latch at the pivot point. You've got a long lever arm putting a lot of force on the latch right there. If you put something there, it needs to be tight fitting and really strong. any slop in the latch and the robot will sag if the air goes out
for the sliding bars, you would need to check how much travel it has and also double check the forces on it. The force depends on where you mount it and what the leverage is on it.
I guess I go back to your earlier suggestion.... Strong telescoping square tube like they sell here: https://alcobrametals.com/products/telescoping-tube/telescoping-aluminum-square-tube/6005a-t6-telescoping-square-tube Plus custom installed beefy locking tube pins like these: https://www.amazon.com/Paddle-Button-Spring-Locking-single/dp/B00N01CLW6/ref=sr13?ie=UTF8&qid=1549339973&sr=8-3&keywords=locking+tube Mounted along side the pneumatics. We might have to try a few things to find something that works right - this seems to be a nontrivial problem. @Paul Vibrans when you're back on here... of all the ideas thrown out here, for which ones should we start ordering stuff so we can try them out do you think?
There is no way to delete the back cables.
Replace the part of the hoops at the side joints with steel flat bar having joints above and below the equator. Use steel for stiffness. Attach with countersunk screws so the outside face is flush. Put the longitudinal guide above or below the equator by enough distance so the curvature of the ball provides space. Attaching the longitudinal guide may require countersunk screws and tapped holes for a flush finish inside and out.
Ok.. I'm working on an alternative .. I think if I tilt the hoops so that they're parallel to the legs when the chute is at 30 degrees, then I can place them anywhere. The hoop shape is weird but it should be much easier to work with.
@Mark Tarlton Will this be printed today?
I've got other stuff ready to print now on the MarkForged. Let me know when the bed is cleared and ready to print, and I'll launch the job
@Nora Wilson has joined the channel
@Nora Wilson has left the channel
@Jeffrey Tappen has joined the channel
I just rewatched Ethan’s Climbing Oof video several times on repeat and I think locking the wheels might help a good bit, but it won’t prevent it 100% of the time. I think we may want to put needle flow regulators in the system for the front pistons since they are extending more quickly.
Those two things done together should fix the problem.
Paul is looking to add "spurs" to the front stalks to provide leverage against the front legs tipping over. These spurs would attach where the wheels and act like shoes pointing to the rear of the robot. If the robot starts to tip, the extended length of the shoe will move the tipping point further to the rear and should stabilize the robot. That may also provide a similar effect to locking wheels. We discussed adding flow regulators. The conclusion was that since the climb system behavior is inconsistent, it wouldn't be likely that we could tune the regulators to give good results across various possible tank pressures. As a result, the focus has been on possible mechanical solutions to the tipping.
We bought smooth hard plastic adhesive backed floor glides (1"x1" squares I think) last year that will slide with low friction on the HDPE - they might be good for the spurs if I'm understanding them. I know where they are if interested.
@Paul Vibrans Regarding the width of the "toe" -- since the grippy stuff behaves more like an adhesive than a friction surface, making the toe wider may provide more grip than a regular width toe.
After pondering the performance of a "clown foot" on the back stalks when trying to climb to the lower level of the Hab, I no longer support the clown foot as a climbing solution.
The way to keep the stalks from moving backward is to fit the wheels with one-way clutches, which I have ordered from Zoro Tools. They should be here in about one week. In the mean time, we need to replace the 3/8" aluminum tube axles with 3/8" steel tube axles and fit them with anti-rotation features. The wheels should be, but do not have to be, aluminum with rubber tires made from O-rings or stretched bicycle tube material glued on with Barge Cement. PVC wheels can be used but their grip on the roller clutches, which depends on a press fit, is not too reliable.
Let me see if I understand the reasoning. When the robot starts, the front and rear legs are "x" apart both in the chassis and where they hit the floor. If the robot starts to tip, the legs are still "x" inches apart in the chassis but on the floor they start to move away from each other as a function of the tilt angle. Therefore, if we can prevent the wheels from moving apart on the floor, then the tilt angle will be 0.
Also, the hab wall prevents motion toward the top of the hab until the bumpers clear the top of the hab. At the lower level, the robot will be tipped radically toward the hab once the front stalks retract, which would firmly engage the toes of clown shoes with the floor, preventing any subsequent movement toward the top of the hab's mid level. Wheels with one-way clutches could roll forward still.
I like that solution — watching our robot during bag day testing highlights the challenges of high level of dependence on air tanks for cycle time execution
@Paul Vibrans Since it would fit in the same or smaller footprint and we should be able to reuse much of what we have (just a different way to actuate) could we possibly trade our pneumatics on the climber for a rack gear solution like this? Advantages: 1) Motor control reliability, 2) L2 climb should be as easy as L3 because we can use motor control for that as well, 3) weight savings (possibly major). The climb is near the end of the video. Since motors don’t count in holdback, we should have holdback weight to spend on it.
Start working on the detailed design. I will be happy to review your calculations and drawings before manufacture. You will learn a lot. Please try to keep it within the envelope of the existing climber.
I might just take you up on that, because the solution advantages over our current one would help the team. You're saying you will collaborate if I take lead? I'll start by reaching out to that team to find out where they got their gears.
I don't have time to collaborate, just time to review.
Then I'll pass - it's your system & collaboration would be required to upgrade to motor controlled actuation. I believe you know this. Please consider the advantages in my earlier post & let me know if you change your mind. Thanks.
Here's an "endless" 3D printable rack gear set. Could probably design one like this that is herringbone. https://www.thingiverse.com/thing:3444515 Clearly, these don't integrate with our solution as is, but worth considering for a future need. There's a different one I saw designed to integrate with drawer slide rails.
What is to save if the climber is to change to electric motor driven rack and pinion? Besides, it would be faster to remove the existing stuff in whole assemblies and replace them with whole assemblies that are pre-tested. As I consider potential details of a rack and pinion climber it begins to take on many of the undesirable fabrication features that made me abandon motor driven screw jacks. I could make one robot set in about 40 hours. How long would it take the students? Rack and pinion jacks also would not be as fast as what we have now, based on timing videos of our climber and motor driven climbers on You Tube. I would not put more than one 775 Pro on each leg because you start to chase battery voltage drop caused by adding motors such that the available climb power doesn't rise and the climb doesn't speed up.
The 3339 climb took 8.5 seconds, with the end of the climb demarcated by the instant the back legs begin to rise (climb successful once contact with lower floor is broken). Our climb in the video on our marketing channel took 8.9 seconds by the same standard. The raising and lowering of legs IS definitely faster with our pneumatic approach; it is the other steps in the cycle that took us longer. I do think we could cut that time by up to a couple seconds with automation, so let's say we can get to a 7-second climb (which I think we will think is great). The questions are concerning reliability and the opportunity of the level 2 climb. 1. Reliability: The speed of the pneumatics seems to create a risk of tipping with potential major consequences because one set of legs gets ahead of the other unpredictably. Will the proposed fix truly solve this? We are all hoping so. 2. L2 Climb Opportunity: This may be the more significant one. If our L3 climb isn't the fastest and most reliable, and alliance captains also have very fast, reliable L3 climbs, then the coach has pointed out it will be a disadvantage to not be able to climb to L2. With rack & pinion + motor control, it's mechanically free - just some programming work (I think). With our current solution, I'm not sure how we do it without tipping risk again.
I agree with what you say about just switching out full assemblies rather than retrofitting, BTW. Before, I was thinking the 2x2 and even rails/legs might be adaptable, but your point about testing on a separate robot and just switching out makes sense.
I’ll believe the numbers, but I think the primary issue would end up being weight. The 775 pro is 0.8 pounds and the pneumatic cylinder ended up at 3 pounds I think, so if we’d just look at replacing the pneumatics rather than the entire system as what seems to be implied, we have a hard time getting a rack and pinion that light.
Starting a new thread, exploring a rack-and-pinion + motor/encoder climbing solution like the one FRC team 3339 did in the shared video (see conversation above). I used one of the common engineering spreadsheets (named AMB Design Spreadsheet) created by a mech engineering prof in Israel as I understand. The spreadsheet indicates that given...
- 10v left on the battery
- 154 lb robot
- 22 inches (19" climb + 3")
- single 775pro per leg
- 0.5" pinion gear
- a 20:1 gear ratio (each leg motor)
- total efficiency across gear reductions: 70%
a climbing leg under load should rise in around 0.7 seconds
Here's a screen shot. This and the similar JVN calculator are designed for use by students, and I would be happy to help any students learn to use them. It is extremely common for students to execute this aspect of analysis on FRC teams, using these tools. (Ask around at competitions if you doubt it :slightlysmilingface:.)
The 0.5" diameter pinion driving on the rack implies small teeth. Check rack tooth strength and final drive pinion tooth strength before settling on a gear ratio.
Thanks, I was wondering about that.
From their email, they used a rack gear like the dr12.5/10/.5 on this page: https://www.beltingonline.com/1-25-mod-x-0-5-metre-plastic-acetal-cut-teeth-rack-6937
Climbing with robots #2 and modifications.
It did climb? More than once?
It fell over. More than once.
Every attempt (2 or 3 tries) resulted in the robot falling over backwards (away from the step).
No climbs were successful.
Wondering if the back legs are slower to extend due to being cabled together; more friction in that system?
Probably. Loosening the cables might help.
What if we sequenced the intake arm to go down first to act as a brace against falling backwards?
I just watched the climb video a bunch of times and there are many things that could be the source of our troubles. It is clear that the stalks nearest the wall are first to go and the back ones go later. With no manifold, the pressure drop in the air supply to the back stalks may be greater so they start later and develop less force than the front ones. Reverting to a manifold would help with this. The tubing runs to the back jacks could be longer than those to the front ones, adding to more pressure drop. The cross wiring on Robot 2 is much tighter than on Robot 1 and has more drag; it should be loosened slightly at the dead ends and this looseness distributed throughout the system if possible. The stalks should drop freely when the robot is on its cart and the safety pins are removed.
Robot 2 has a different weight from Robot 1 and the center of gravity could be much farther aft. If Robot 2 can be tested with ballast on the end that goes up first, it would help our understanding of the problem.
The intake arm can't be put down first because wall is in the way.
Can a delay be built into the programming so that the front stalks go down reliably later than the back stalks so that if friction were equal front to back, the robot would always tip toward the wall?
Yes, we did this when first testing the climbing system.
Today or some time ago? When I saw the robot attempt to climb in the past, it was tipping toward the wall and the timing was adjusted to make it tip away from the wall. The wheels were modified to work with the robot tipping toward the wall and having a center of gravity on the wall side of the middle axles.
The first time the programmers had the robot, so some time ago. It was working relatively (? half the time, about) well without any timer. When it was tipping, it tipped away from the wall. We played around with the timer, but it didn't help much at the time.
@Paul Vibrans Justice is correct that we had a timer, but he missed the nuance that the back wheels now lock. We tried a timer at CK once today (I put it into the code), but we’re testing driving now to take advantage of the field. We will test more back home (maybe Monday?)
I will be prepared for Monday.
Could we use needle valves like on the intake arms?
One anecdote from another team: Rock could not get a pneumatic climbing solution to work because of a similar problem to ours of legs rising at different speeds unpredictably, and they spent much of this weekend switching to a motor-driven solution involving chain inside 80-20 extrusion. They demo'd it, and it was still not Rock solid (:)), but it seems like they'll be able to get it dialed in.
Of the things I've seen us try, the clown feet seemed to have promise given how they seemed to stabilize the robot against tipping while the climbing legs caught up to each other. Not sure why the roller clutched wheel didn't help more.
We tested the system at CK with a delay and it fell the other way. If we fine tune the delay on the first robot, we will be good. The clutched wheels worked perfectly but the weight of the second robot is different than the first robot. I think we should test the first robot with clown shoe and clutched wheel without the delay, and then try adding the delay if needed.
Lets get Robot 2 to work and truly understand the climbing problems before doing anything to Robot 1.
I watched the video again and saw that the roller clutches are oriented properly. We can add clown shoes and fix the center of gravity enough for testing. Loosening the cross cables should be the last thing we try.
I will bring clown shoes for both robots.
Something Mark has mentioned a couple times that we could fairly readily test is different, higher traction center wheels on the intake for pulling more forcefully during climbing. The idea of the omnis was that they would continue to allow the ball to center while being intaked, and I still like it, but it might work just as well with grippier wheels. Mark suggested the compliant wheels, and I think andymark traction wheels (maybe one click softer than our drive wheels - blue ones?). We could also make some slightly larger-than-4" diameter wheels out of plywood or something, and fasten the super grippy diamond-shaped tread. The slightly larger diameter would help prevent breaking the mecanums as much, and it would put the robot's weight on the grippier wheels.
Another thing: I feel once the intake wheels are spinning there isn't much pull, and with the intake motors at 100% they start spinning almost right away. What if we fast pulsed the motors instead, go to 0% (or even brake mode deliberate stop) for a moment to restablish static friction, then 100%, then 0%/brake, etc.? It seems to me, this may be more effective than just full speed spinning.
@Paul Vibrans has left the channel
@Max Morse has joined the channel