@jack_chapman set the channel purpose: do hatch intake and ejection
@andrew_peterson has joined the channel
@Ethan Rininger has joined the channel
@Harrison Gilmore has joined the channel
@julian_bonifaci has joined the channel
Effective hatch panel handler: https://www.chiefdelphi.com/t/team-2655-simple-touch-it-and-own-it-hatch-panel-manipulator/342926 If we can get this fast, we should be in pretty good shape
Another hatch panel grabber: https://www.chiefdelphi.com/t/frc-2363-triple-helix-2019-gripper-prototypes/342403
Very insightful discussion about hatch panel handling
@chrisrin it might be a good thing to try and make a polycarbonate mechanism because of the flexing and “give” that will follow, allowing the driver to be a little less accurate when placing the hatch panels but still get it on almost every time.
*Thread Reply:* Agreed! Or at least consider how the design can be done to similarly enable some leeway for the driver. I heard the hatch panel handling design direction is Velcro and pistons again... is there a sketch/CAD of that anywhere?
@declan_freemangleason has joined the channel
@Merrill_Keating has joined the channel
What is the status of the hatch panel handler? It seems like a redesign was planned - is the design done? Work was done on other systems over the weekend and yesterday - could some work (CAD maybe) be done today on this system?
I think this is worth a try as an enhancement or replacement to Velcro. Can we please buy some to try it? @violet_advani Sewell Direct AirStick Microsuction Tape by Sewell, 0.8mm, 250mmx300mm Sheet - As seen on MOS products - Mount - Retail Packaging - Black https://www.amazon.com/dp/B00ITX9OU2/ref=cm_sw_r_cp_api_i_nz6ACbCMYK0A4
Also let’s try these - please purchase: Bluecell 10 PCS Rubber Strong Suction Cup Replacements for Glass Table Tops with M6 Screw https://www.amazon.com/dp/B07CZVMRF2/ref=cm_sw_r_cp_api_i_PP6ACb1ES87EK
@channel Here are some thoughts on the hatch panel handler
I think center hole grab is most consistent however due to our exit cargo chute design on the same side limits our options .
Regarding the 'inconsistent eject' problem: The pneumatic 'kickers' need to hit the hatch panel very close to the points where the velcro is holding the hatch. Otherwise, as soon as one point releases, the other velcro simply acts as a hinge and the hatch panel pivots instead of fully releasing.
*Thread Reply:* Agreed! I’ve also been thinking that. We should extend the kicker bar up to the high Velcro grab points on both sides. It’s shape will align with the grabber .
Big thing we also missed is the impact of the bumper kickers on the game piece pickup and drop points being deeper than the thickness of the bumper. And the height of the kicker opening is a few inches above the chassis level. We must deal with this and the hatch panel handler design together. It’s no wonder the hatch panel handler took a beating last night; the force of the entire robot ran into it like a freight train. Three options I see: 1) Build a 2nd bumper - strong barrier at frame perimeter that extends 4 to 5 inches up from chassis. I put a piece of 4” high thick L on the 2nd robot as reminder 2) Set the hatch panel handler a couple inches back & don’t have anything on the first couple inches of the robot at all (lightest weight option), 3) Use sensors/programming to never go that far forward (this one seems unlikely).
*Thread Reply:* @Mark Tarlton how far from the frame perimeter is the chute edge? Hoping it isn’t also at risk. Bumpers are around 3.5 inches thick uncompressed, and I would guess as thin as 2 inches when compressed by a full speed hit. The kickers on the loading station and Cargo Ship are 5 3/16” deep. So I think, conservatively, anything within the first 3 1/4 inches of the frame perimeter is at risk of damage, and anything within the first 2 1/2 inches is at HIGH risk. @paulvibrans any ideas for what would be strong enough to act as a barrier to a full speed robot hit at 4” above the chassis level? I’m feeling like option 2 (see original post on this thread) will be the way to go if we can. @CruzStrom this is what I was trying to explain last night @Ethan Rininger @jack_chapman @Jack Scheiderman FYI
*Thread Reply:* The "shooting" end of the chute is right at frame perimeter. the first things to get hit are the conformance wheels and the hatch kick bar. The expectation was that the structure for the hatch panel handler (polycarb, etc) would protect stuff behind it -- until it all went away, that is.
<begin mild rant>
The whole idea of a "system" is that it's more than a collection of parts. In a system, the parts work together and often serve more than one purpose. It often isn't obvious at first glance all the different roles that a particular component is serving. For example, I expected the polycarb panel on the back to wrap around the rear of the robot and protect everything behind it. The wrapping was important because the wrapped part would anchor to the frame and provide stiffness from rear impacts. It didn't wrap and so it was left to the hatch panel bars to provide the structural support. They weren't up to it by themselves, so someone decided to anchor them to the climbing structure. When it became apparent that the climbing structure was getting moved by the impacts, the answer was to remove the polycarb, air bottles, etc. So now, the chute is exposed and expected to take the hits. Another example is that the combination of moving parts and removing structure changed the CG of robot (which admittedly was something we thought we needed to do) but it had the undesirable side effect of changing how climbing was working -- the fix we put in for the front going up to fast was no longer relevant.
It seems to me that we're now in a cycle of adding band-aids and patches. We need to be more thoughtful about how we work. I understand the time, weight and functionality pressures that we're all under but I think this is the time to step back and look at the robot as a system again before we make more modifications. More importantly, I want to think about the robot as a system and see what we can learn from our experience. And to be clear, this rant is aimed at me as much as anyone else.
<end mild rant>
*Thread Reply:* Valid points, and it aligns with our desire all the way back to the summer to stay focused on the "whole robot" design. It seems like we need a design session. @Kenneth Wiersema thoughts?
For this specific problem of anything in the first 2 to 3 inches of the frame perimeter on that side being at risk of taking full speed robot hits, even those original 1x1 vertical pieces were not really working. We need something quite strong that takes the force low, and it needs to have some surface area in order to distribute force and avoid damaging the field elements. Our chassis dimensions are 28x31 I believe. Could we attach a few 0.25" thick bars that extend 4.5" up to the outside of our current chassis and declare the outer edge of those bars the new edge of our frame perimeter (effectively they become part of the chassis)? We have ~4 inches of total frame perimeter to spend, and if we did that we'd only be adding 0.5". Could also use polycarb to distribute load. This is just a prelim question - not suggesting a solution as much as trying to understand what is possible/allowed vs. not.
*Thread Reply:* Gets back to that systems thing...changing frame perimeter would mess up the current bumper mounting....
*Thread Reply:* Actually, software is something that should be considered. A distance sensor that's enabled when we're close to either loading station or the cargo ship could be used to prevent ramming into the walls would be ideal. It also means discarding the strategy of "ramming into the walls to square the robot up" -- another good thing in my opinion.
*Thread Reply:* The other option is no not place anything in the danger zone on top of the chassis. The recess at the floor is 5.25" deep. The danger then is to the chute running into hatch panels. There's some cushion inherent to the hatch panels and the compliance wheels on the chute.
*Thread Reply:* Based on initial calculations (see engineering channel), I think not placing vulnerable things in the danger zone will be wise. Even those original 1x1 uprights were not going to be sufficient given the force required to decelerate that quickly. I'm now checking with the bumper team to see what kind of pool noodles we used in our bumpers (hollow vs. solid)... hollow will compress more, so the danger zone will be a tad larger.
Is there an opportunity to shorten the chute by a couple inches by chance?
*Thread Reply:* Shortening the chute will be difficult due to available belt lengths, motor placement, etc. @chrisrin Are you concerned about collisions on the chute or do you have a different issue in mind?
I have started working through an alternative hatch handler that stays clear of the danger zone. I'll share images of what I'm thinking about ASAP (probably tomorrow).
*Thread Reply:* I'm concerned about the chute. If you believe the chute can repeatedly absorb the large impact forces, then I guess we can be less concerned, because I agree we can come up with a hatch panel handler that is back far enough to avoid impact.
*Thread Reply:* With the chute, I'm hoping that the combination of conformance wheels hitting the polycarb panels will offer enough give that it won't be as destructive as solid, braced aluminum hitting wood structures. 🤞
*Thread Reply:* OK, just measure 2 - 3 inches back from frame perimeter. If my calculations are correct, whatever is there will absorb over a thousand lbs of force to stop the robot at full speed. Bumpers will start helping absorb force when the field element corner is around 1 3/4 inches inside the frame perimeter, and by 2 1/4 inches the bumpers should be taking a lot of the load. I haven't been able to get information about how fast or how much bumpers compress under force, so I can't say for sure.
*Thread Reply:* #programming -- please read @chrisrin message above. full speed collisions with walls generates a half-ton of force. Help here will pay off.
*Thread Reply:* I think we'll want to put something down low to help. I'm thinking lots of carbon fiber golf shaft remnants. Or maybe "build a wall" of 5mm titanium or bamboo rods sticking up out of the 0.21" holes in the top of the chassis plate, surfaced with polycarb to distribute load. Or something with rubber or springs. Or maybe bolt on this titanium shovel head (only 5.6 oz) down there: https://www.amazon.com/gp/product/B00U464ZQU/ref=ox_sc_act_title_4?smid=A2FFPAD14D23KE&psc=1
*Thread Reply:* But seriously 🙂, we'll need to think of something. I was just pondering using polycord or some other rope somehow, maybe laced sideways through the lower holes in the chassis plate on that side to distribute force that way.
*Thread Reply:* Here's a sketch of a lightweight rope suspension bumper idea
*Thread Reply:* My opinion is to leave it clear and make sure to not place structure where it's going to get hit. Let the bumpers do their job and keep everything else out of the way.
Right now, the chute is the only thing sticking out and it should only run into hatch panels with minimal support behind them. The way I see it, we should be able to withstand impacts. First, the hatch panel will flex a bit -- perhaps 3/4" or more. It's being struck by the conformance wheels. They stick out 1 inch and provide initial cushion. They're mounted on aluminum axles but are supported directly by bearings so the bending risk is minimized. Next, the chute pivots around a mounting pin at the very front of the chute which means that horizontal impact will get translated into a torque around that mount. The forces resisting that torque come mostly from the pneumatic cylinders that lift the chute. They will act as a bit of a spring. The remaining forces will try to move the top axle which is mounted by versablocks. they may slide a bit but not enough to really loosen the belts because they are braced by the pneumatic supports. Finally, forces will go into the chute rails which are made of polycarb. they will flex but rebound. If anything does fail, it will most likely be the mounting block at the very bottom of the chute. These may break because (on robot 2) they are made of printed PLA which is brittle. On Robot 1, they are Onyx (Nylon with carbon fiber) and will flex more. So, rear impacts to the chute get dissipated through a variety of flexible structures with no one structure having to take the full load.
Still, the best solution is to not run into field elements having recessed bottoms at full speed. Right Drivers?? 😉
@Jack Scheiderman has joined the channel
@Mark Tarlton If you're working on a center grab, the SERT robot shared on engineering channel is an interesting one that both pulls the panel inward and kicks it outward.
I worked on hatch handling over the break. Most of the ideas didn't work out due to space conflicts with the chute. Since the outlet of the chute is right where the center of the hatch panel is, it's difficult to place any devices there without interfering with the the chute itself or with cargo as it get ejected. There are a couple of promising directions, though.
The first is a refinement of the proposed 'use velcro to grab and pneumatics to eject" direction. One thought is to attach velcro mounts to the rear climb legs that are soft and flexible so that impact with the loading station will be firm enough to lock the velcro but not so rigid as to damage the climb system. A similar mount would be used at the bottom center so that hatch panels are grabbed at three points around the perimeter.
The eject system needs to make sure that all attach points release fully and at roughly the same time. The first suggestion is to make the ejecting arms longer and stiffer and then place them so they touch the hatch panel close to the points where it's being held. This will minimize the effect of hatching flexing and ensure the velcro releases completely rather than act as a hinge. We should also use longer pneumatic cylinders so that the hatch is pushed it all the way to the rocket body. The extra length is needed because they have to be mounted at an angle to avoid blocking the cargo as it moves through the chute.
If we need a more secure grab of the hatch, the first option is to place a hook in the middle of the chute so that we can lift the hatch by lifting the chute. Frame perimeter is an issue with this approach so we will need to use a pneumatic cylinder to extend the hook when it's needed. I think there's room under the chute to do this.
Another option for securing the hatch panel is to grab it at the bottom edge. This would be a small gripper that stays behind the frame perimeter and pinches the edge when we dock at the loading station and then release before ejecting. This would require a new actuator as well.
In all cases, the eject system described above would be used to place the hatch panels at the destination.
*Thread Reply:* We should test the Sewell microsuction tape on the hatch panels. It may work much like Velcro, but provide an ability to grab the hatch panel in a wider range of locations, including between the center hole and the bottom where we avoid chute interference easiest. Grab and eject contact could be right next to each other as well. Hopefully the tape sucks 🙂.
*Thread Reply:* do we have this on order? if so, what size and shapes?
*Thread Reply:* Yep, also some screw mount normal suction cups. Scroll up & you’ll see what I requested
*Thread Reply:* With suction cups and suction tape, you need to apply force to engage them, and then you have to apply even more force to release them. Velcro is similar in that regard. The idea of using a roller to release them is a good idea if combined with a kicker so they don't engage further. A little bit could be helpful in preventing swinging and sliding as the robot moves. Too much, and our ejection system might not work. At best, our pneumatic ejection system produces about 48 lbs of force -- about the same as a suction cup with 1" radius.
Suction cups that can be easily released, like the ones Paul suggested, make placement much easier.
I found that solutions like the grabber thing got complicated quickly. Frame perimeter issues combined with the chute, meant actuators were needed to extend it out past the frame and another to grasp and release the panels. Placing panels on the rocket also required another surface that's extended out past the bumpers to press the panel into place before the panel can be released. Once again, the locations for those actuators and shafts were all lined up with the chute exit. With time and effort, I'm sure we can solve the problems, but it isn't simple or easy.
*Thread Reply:* Thanks for all the analysis. If I think about fast cycles, the velcro/suction drive-in attach and pneumatic eject placement seems fastest IF it can be refined to be reliably repeatable. That's a tough problem though, since the velcro/suction "grab force' will vary, and the eject force needs to be greater than the max grab force. Along with that, to your point, the points where eject force are applied need to be near the grab points.
A reliably repeatable solution will be more achievable with an actual center or under-panel grab. But it'll be slower because of the need to dock / undock. A solution that is forgiving when it comes to alignment will help.
What direction is your recommendation for tomorrow? Or maybe a couple possible solutions in parallel?
*Thread Reply:* I've got a rough design for the velcro/eject system. I talked to Paul about a grab-from-below thing that has promise. I'll work on that tonight and we'll see how that turns out. I've completely given up on 'grab-from-center' ideas as the chute conflicts are so severe.
*Thread Reply:* Possible impact dampener if needed: Sorbothane. Available on Amazon
I’m wondering if a more resilient extension off the climbing 2x2’s to catch the top of the hatch panels would be a Velcro covered pillow sticking out. Here’s one. We have a similar one - should not break, and the styro beads are very, very lightweight. (Just trying to think outside the box here 🤓) Cushie Pillows 3.5” x 8” Microbead Bolster Squishy/Flexible/Hypoallergenic/Extremely Comfortable Roll Pillow – Blue https://www.amazon.com/dp/B00UB4S3MC/ref=cm_sw_r_cp_api_i_PklFCbVYQXE0T
What is the risk of tearing one of these things on a field element and creating our own sandstorm of electrostatically charged styrofoam beads? Looks like a yellow card waiting to happen. How embarrassing!
*Thread Reply:* our 6-month puppy has destroyed numerous things at our house & has gotten hold of our similar pillow several times & chewed on it & never ripped it open - just too much give to it... it actually might not work for that reason... I do think bent polycarbonate might be better than the 3D printed stuff that broke today
*Thread Reply:* I have some ethafoam shapes that might work.
Maybe something like a knee pad NoCry Home & Gardening Knee Pads - with Lightweight Waterproof EVA Foam Cushion, Soft Inner Liner, and Easy Fit with Adjustable Velcro Straps https://www.amazon.com/dp/B06XJ2BFW7/ref=cm_sw_r_cp_api_i_KcvFCb6FS7J8W
I still like what we used yesterday. I had a close up view of them working and saw why they failed. They were stiff enough to not hang onto the hatch during eject and they were compliment when hitting field elements. It wasn't the impact that killed them. The ends of the bumpers were getting caught inside the field elements when the robot came into the target off line. If the robot turned the wrong way when backing out, the caught end couldn't release. They survived the first three or four times they were caught, but it was clear they wouldn't last long.
The bumper design needs to be fixed so that they can't get caught inside the field elements. It's a small, easy change. .
*Thread Reply:* Where and how are the bumpers catching? I’ll be late tomorrow because of track. Maybe you could show me after dinner?
*Thread Reply:* Sorry! A bad choice of words on my part. I have small "bumpers" that hold the hatch to the climbing assembly. My bumpers broke, not yours!
Hey @Mark Tarlton I'm not sure this will be of any help at all with the non-Velcro hatch panel handler actuation, but in addition to pneumatics, servo motors that cost <$75 are allowed. Here's an inexpensive high-torque one on Amazon. https://www.amazon.com/LewanSoul-LD-3015MG-Standard-Digital-Control/dp/B073F4TRSK/ref=sr_1_3?ie=UTF8&qid=1547534999&sr=8-3&keywords=linear+servo+PWM
I actually ordered 2 in Jan when @chrisrin mentioned them.
I would definitely consult @peter_hall, @Terry, and @riyadth on servos. I don't know much about them other than they are allowed and a good alternative for some purposes. The link above was an example that seemed potentially workable, but I now see the working voltage is 6 to 7.4V, and not having worked with servos i don't know what implementing one entails. The overall point to Mark was maybe a servo might integrate into the design in a way that a pneumatic would not. thanks
To use a servo, it is simply plugged in to the RoboRIO PWM port. It gets power from the RoboRIO directly. They do not draw much current, and achieve high torque by adding gears that lower the output rotational speed. If you merely want to activate a latch or raise a grabbing arm, they are very effective and easy to add to the system. The programmers can easily set their position with relatively fine control and software adjustability.
Lucas suggested servos earlier. Paul estimated we needed something like 16 inch pounds of torque to pull the hatch panel in. Lucus didn't know what our servos were rated for so we didn't follow up on the suggestion. I can check the inventory for model numbers if I can't get the pneumatics to work
*Thread Reply:* I like the look of this dual shaft servo with the mount and bracket. I think we could incorporate this pretty readily into an under panel gripper.https://www.amazon.com/dp/B072N26Q3T?psc=1&pf_rd_p=c35a81db-c1c9-4121-b93f-97ef6cbb19f8&pf_rd_r=1CBC0AWVG1QFHEDZBEX8&pd_rd_wg=uZCIx&pd_rd_i=B072N26Q3T&pd_rd_w=6uFh1&pd_rd_r=0e5b87aa-08ff-4372-aea9-0d7d97db02ab&ref_=pd_luc_rh_crh_rh_sbs_03_02_t_img_lh
*Thread Reply:* @Mark Tarlton could have a back stop that makes contact with disc and then rotate this up around the front. ~35lbs of force
The 17kg ones coach said he ordered should be plenty
Hey, one other thought: Go back to Paul's pneumatic-powered suction cup idea. Mount them on the kicker. Cycle: 1) Push kicker out, 2) Push kicker against hatch panel at loading station, 3) turn on the suction, 4) go to hatch panel placement location with kicker still out, 5) place hatch panel on velcro at placement location, 6) turn off suction. We could probably order those type of suction cups and develop the solution by Glacier Peak competition. Not sure about an expeditious source. Paul, would an array of this type available on Amazon work? --> https://www.amazon.com/gp/product/B07MHGDT5D/ref=ox_sc_act_title_1?smid=A1THAZDOWP300U&psc=1. Hey @Cruz_Strom, pneumatic-powered suction cups are often used for repairing dents - would your dad by chance have any?
The suction cups are a small fraction of the system. What about the pneumatic system that breaks the vacuum?
http://www.vacmotion.com/Products/Main_SuctionCups.aspx?PpcRef=MSN Should be able to find our system here or a similar specialized outlet. Wow, there are a lot of suction cup options in this world 🙂. I bet we could actually get an experienced person from there on the phone, explain the problem we're trying to solve, and then receive recommendations... might be a good way to go.
#programming,. Another solution to the problem of inadvertently removing hatch panels when shooting cargo is to extend the hatch kicker after shooting and while driving away. That way, the hatch is pinned in place until after the robot is clear of the cargo ship.
*Thread Reply:* @CruzStrom @chrismentzer See above. Also, this approach may work best for delivering hatch panels, at least to the cargo ship. With hatch panel on robot, align and drive all the way up to the hatch so velcro on both robot and hatch are gripping the panel. Then trigger the kicker. Then back away, and the panel should stay with the cargo ship as the kicker extends.
I think we're just going to add a manual hatch toggle for Cruz and he'll be able to find the best strategy.
The best velcro hatch panel handler I've found is 6329's. Here's their reveal - go to around 27 seconds in: https://www.youtube.com/watch?v=u96V1Hwdyd4. And here's a screen shot: https://1drv.ms/u/s!AikCDwtdoW5Lx141S70CXESQHX-x They make Velcro contact on the top and the bottom, and they eject with pistons RIGHT at the points of contact (actually THROUGH the points of contact it appears), actuating in parallel with the floor to push the panels onto the hatches.
@paul_vibrans @Harper Nalley @Ethan Rininger Here are some shots of a first pass at 'fingers' on the hatch grab. The way they work is that they are connected at the bottom to the shaft that links the pneumatic piston to the flipper linkage. As the piston goes up in the cylinder, the fingers move up as well and pivot forward around the shaft. The inner edge of the fingers ride along the top edge of the support frame. The slope of the finger assembly is designed to act as a cam. As the fingers move up and down, the edge causes the finger tips to move in and out.
The attached screen shots show it in action from the front and back. Don't worry about the flipper penetrating through the hatch panel, that's just a tight grip 🙂 Also, the fingers, by design, don't touch the inner ring of the hatch panel. I don't want anything to impede the flipper grabbing at the bottom. I also expect a bit of elastic will be used to force the finger out and to prevent them from rotating backwards
This is a first pass so comments and questions are welcome.
Looks great Mark! Excited to see it in action.
Can there be radii on the insides of the fingers to get rid of stress concentrations and radii on the outsides to reduce weight and print time? Can the edges of the fingers that grab through the hole be angled to be vertical when extended? This makes a bigger gap for aligning with the hatch hole edge and reduces weight. The horizontal part of the fingers that fit through the hole should be the same height and thickness as the part of the finger that points down if the strength is to be the same. The finger tips can be very small because they carry no bending load.
regarding radii, yes, absolutely. In Fusion, I find it easier to leave fillets till last so that I can more easily edit the core shapes.
The finger shape is dictated by the retracted position because of frame perimeter issues. they need to lay pretty flat when retracted. I can shorten the tips to give more clearance but it's a tradeoff for holding.
The dimensions of the horizontal and vertical parts of the fingers were originally the same at 5/16" and can be changed back to that. I made the top thicker because I wanted to strengthen the joint between the finger top and the back vertical element. I was thinking there would be torque at that joint when the fingers closed on the hatch.