• aditya mehrotra.

chip updates: combining a kangaroo, a ship pulley, and a wild robot arm into a leg [updates]

So we're going to take a slightly new design philosophy with this version of the chip robot dog. Combining a whole bunch of ideas we have been bouncing around for a while.

The above video is from a great robotics YouTuber and it's about rolling-joint block and tackle arm joints on things like the AMBIDEX arm (below) and how we can use pullies for high back-drivable reductions on robots.

The above is LIMS2-AMBIDEX which is a pretty cool Yong-Jae Kim design from NAVER Labs and KOREATECH.

It would be interesting to combine this kind of drive mechanism with the structure of the FESTO robot kangaroo and see what we can come up with. Complex but cool mechanical design, but the advantage is it would be very light and well packaged. This spider-plate design would look good w/ the cable drive as well.

So essentially the plan is to make a super light-weight system where we get rid of the versa planetary gearboxes completely. We use just the NEOs connected straight to block + tackle systems with spider-plate water jet plates for limb structure. We'll CAD this on the train today.

But before we start CAD'ing anything we need to start by finding some small pullies and some large ones...

(1x) https://www.amazon.com/20pcs-Groove-Pulley-Bearings-4136mm/dp/B012KKTBRU/ref=pd_bxgy_2/139-8621867-0459260?_encoding=UTF8&pd_rd_i=B012KKTBRU&pd_rd_r=9b896c62-c8cc-4aac-9a07-265e97027b87&pd_rd_w=w55Ya&pd_rd_wg=yWUyV&pf_rd_p=fd3ebcd0-c1a2-44cf-aba2-bbf4810b3732&pf_rd_r=CT287AD6RP4NEY1FNRPN&psc=1&refRID=CT287AD6RP4NEY1FNRPN --> tiny bearings

(8x) https://www.mcmaster.com/9466T84/ --> large aircraft cable bearings

(1x) Some aircraft cable --> https://www.amazon.com/MUZATA-Railing-Stainless-Aircraft-Balustrade/dp/B08JL9WN9D/ref=sr_1_3?dchild=1&keywords=aircraft+cable+black&qid=1622558163&sr=8-3

Here are some initial drawings of the leg design on paper and some basic math and geometry considerations. We're aiming for a 20:1 gear ratio w/ a 4:1 belt reduction and a 5:1 pulley diameter reduction.

Above is the design of the knee joint and the basic principals were mostly taken from the Skyentific YouTube channel design and the Hackaday project. So special shoutout to them for the overall design concept. The nice part about this whole design is if we want to change the reduction, it's relatively easy to do so by changing pulley diameters and etc etc. We won't want to add or take off pullies later as that would be a pain. This is a modified rolling joint essentially with a cable drive. It should look very cool, have a high degree of back drivability (as demonstrated), and also a very small amount of slop when tensioned properly. We probably want to come up with some form of eyelet screw tensioning system so we aren't relying on cable clamps and things for tensioning.

Above is a design for the screw-based string tensioner, we'd need two of these on different locations on the robot. We'd use cable clamps or similar to tie the cables through the eyelets and such. Calculating the appropriate pre-load on these is fairly simple. We probably won't use a toque wrench we'll just feel until it's tight enough, but below is an example preload consideration.

Also need to consider the stresses in the cable that come from wrapping it around a pulley, but we'll just say these are accounted for in the cable rating and the bend radius rating of a cable.

Essentially when tensioning, the stress that comes from axial elongation + the stress from the maximum loading on the cable must be less than the yield strength of the cable. We're planning on buying aircraft cable/steel cable 1/8" to 3/16" in diameter. They have around 400lbs capacity which is around 2000N which means we're likely more than good on both yield and ultimate failure. Usually we'd calculate this based on F/A of the cable and the Young's modulus of the steel in question.

This cable is a fairly good option: https://www.mcmaster.com/3332T52/

  • ultra flexible cable

  • 400lbs rated capacity

  • 1/8" diameter cable

  • extra lubrication

Problem is the recommended pulley diameter is > 3 in which won't work for the smaller parts of our design. The 1/16 version of this is rated at around 90lbs and has a bed radius recommendation of ~1.5 inch. Won't work for that 0.6inch small pulley really. But we could consider this since the 90lbs should be much more than we need. It just may not be as smooth as something else. We could also consider other types of line like fishing line but maybe the 1/16th steel cable will be OK. We can test it and find out. Since fatigue is also a consideration, small kevlar rope may work better as well.


It looks like KEVLAR line may be the way to go in this case, but we're asking MENG the resident cables expert in the lab to see what might be an option.

And there's ONE ISSUE w/ all of this:

Literally it's SO wide because of the choice of main pulley. This upper leg is almost 4 inches wide in the middle and the lower leg that would have to go around it would be even more. So we're going to re-think this design a little and see if we can't make it more compact. We're rivaling the versaplanetaries in terms of size we want to be like 3in wide around there.

If we use these pullies and a different block and tackle system we can make this system much much thinner. https://www.amazon.com/20pcs-Groove-Pulley-Bearings-4136mm/dp/B012KKTBRU/ref=pd_bxgy_2/139-8621867-0459260?_encoding=UTF8&pd_rd_i=B012KKTBRU&pd_rd_r=9b896c62-c8cc-4aac-9a07-265e97027b87&pd_rd_w=w55Ya&pd_rd_wg=yWUyV&pf_rd_p=fd3ebcd0-c1a2-44cf-aba2-bbf4810b3732&pf_rd_r=CT287AD6RP4NEY1FNRPN&psc=1&refRID=CT287AD6RP4NEY1FNRPN

And actually... a lot cheaper, these big pullies from McMaster are expenssssiiveeeee. We're going to look at a rolling joint design used by AMBIDEX:

This is the kind of system we're thinking about for the block and tackle system. One on each side of the joint. The only other thing to consider is the range of motion of the joint when we design and build it is limited by the distance between the pullies. The torque produced at the joint is also affected by this distance. For now let's start drawing out those dimensions.

Soo this is a classic rolling joint that would use 20 total little pullies (maybe 21 depending on how we wire it instead of 20:1 it's 19:1 or something). We'll start CAD'ing this soon but for now I think we've done a good amount of figuring out for the day. The above drawing shows how this should be laid out with the upper leg going straight off to the right and the lower leg going straight down in this configuration.

#updates #omgrobots #yay #cable_drive #built_by_nerds

2 views0 comments
© copyright 2019 | aditya mehrotra