category: SummerBlogs This creates a floating frame with caption below For best results, attach .jpg or .png which has been resized about right
Summer blog - update every few days with new findings.
This week I continued working on my spines and improving the hand with Wilson. I started the week by making a new design for the phalanges. Each phalange would have around 81 spines per phalange and are angled. Although making several spines is made easy with the pattern tool on solid works, making them at an angle is not. In order to make the angled holes I had to first make a plane at angle above the surface and then calculate the distance from the plane to the surface I had created. I also had to make calculations to calculate the travel of the spine at an angle. To ensure that I had made my calculations correctly I used an assembly to see whether or not the spines travelled as much as I wanted them to. I already had the model of the spines on Onshape so I drew a cylinder that functioned as the outer brass tube. In order to insert the spines into the phalange I used a cylindrical mate and then patterned the spines throughout. Although I found that my calculations were correct. I discovered that it would be impossible for me to insert my spines in the final row as they were at angle and would have to go through the back wall in order to be inserted. Therefore, I had to modify my design to include more space between the back wall of the phalange and the last row of spines. Although the assembly tool was useful for this I had to again to some calculations to calculate the exact distance. I then printed this model and started to make more brass inserts for the part. However during the week we realized that this was a very time consuming process so we have now ordered teflon tubes to use instead.
Lately I have been working on creating a functional spine tile with brass rods. I am still in the process of deciding whether or not to use telescoping tubes instead of just having the plaster as a taper. However, getting the manufacturing process down has been hard. The inner brass rods must be scratch free to properly slide through the outer rods. However, this is damaged by the cutting process of the rods. Since Brass is really soft, clamping the rods can either damage the shape of the rod or create scratches on the outside. It is also difficult to cut all of the rods at the exact same size as it does not have a stop. Therefore, I designed a jig on solid works that looked like a vice but had a spot and an exact opening to fit the hacksaw through. However, right before I was about to print it, Alex told me he had bought a mini electric saw that allowed me to cut at the same spot everytime. But to ensure that I was cutting the brass rods at the exact same length I cut a piece of would that I then double stuck onto the saw stand. This gave me pretty consistent results with the length of the brass tubes and made it super simple to replicate the manufacturing process. Nonetheless, I still got burrs in the brass rods even though I went slowly. Next time I cut the rods, I plan on including a support at the bottom to ensure that there are no burrs
Cutting Brass rods is harder than expected! So far I have not succeeded on making this process super fast. Although I got a functional quad of telescoping spines I still need to perfect the process to make a full spine tile.
Today I worked on redesigning the new pneumatically actuated spines and putting together the system I had. Although on Friday I had tried assembling my spines in the form printed structure. However, I soon realized that there were several issues with my set up. Firstly, even despite re-drilling the holes in the form print, there was still too much friction between the spines and the structure and the spines would not come out even when pushed by the air in the air compressor. I thought that this might be due to leaking. Thus I then decided to add the rubber but this only added more friction and when adding the top of the assembly the seal was not airtight and not strong. Although on Monday I tried using an objet printer printed part. This was surprisingly more successful as the plastic was slippery and the spines flowed better. I have ordered teflon tubing and also brass tubing to see how these might work on allowing the spines to move better inside the assembly. I will try these tomorrow
Day 2, 3 &4
On Tuesday I worked on post processing my screws and starting to CAD a precision plate for the rolling diaphragm setup. However, after I was done with that I begun working on a different way to engage the spines in Wilsons hand. After talking with Mark we came up with a few ideas on how to make differently manufacture these spines, especially using pneumatics have them engage rather than the tiny springs that are in shizuan's hand. Thus, after this discussion I began to try and test whether this could be done. I started by laser cutting some parts of acrylic to encase the spines in, through which I would then blast air through to see if they where strong enough. However, I encountered a series of issues including the fact that the different layers of acrylic would not be air tight and there was nothing holding the spines from going all the way through except for the thin layer of acrylic of a smaller diameter at the end. However, that thin layer of acrylic reduced the travel distance of the spine. I also had issues with the rubber, as I couldn't drill or cut holes through it that provided little friction. To fix this I moved on to CADing a casing for the spines to 3D print and also using the UV laser to cut holes into the rubber. Although the 3D printer is still running, I tested the spines in the acrylic and they fit in smoothly. Also, to reduce friction, I added silicone Grease which made the spines flow in the rubber perfectly. I also used tubing to make a thicker part in the spines and prevent them from going all the way out. I am excited to try the new setup by using both a channel that directly shoots air at the spines and another that fills up a balloon to push the spines outwards.
On Monday I worked on CADing screws for the rolling diaphragm setup. Since the big PRL lab is closed the we cant adequately lathe the threads into the rods. There for we thought it might be a good idea to make our own custom threaded inserts and screws that fit exactly into the the aluminum rod we have and do not wiggle for proper alignment. This was a really good way of practicing and becoming more proficient at CAD.
Day 2,3 & 4
For the rest of this week we continued to troubleshoot the casting process and try to narrow down the variables that might be affecting the curing of the silicone. To do so we printed new molds in the PRL and changed different variables in each one and kept one as a control. Some of the molds we cleaned in our IPA bath that we made by buying fresh IPA, others we cured in the UV oven and others with just sunlight, and we also we laquered most molds.However, we were able to determine that none of these factors was the one affecting the curing of the molds since none of them cured. We are currently looking more closely at where exactly the faults in the casted diaphragms occur to see if there is a trend. Although most of our attempts to cast during the week where unsuccessful I am slowly becoming better at CADing and have learned how to use the 3D printer on my own. During the week I also talked to Wilson about what the best geometry for the hand would be we finally decided on a more bear claw like hand that has two longer phalanges rather than three and has a particle jamming hand.
Today I worked on making more fabric but using the kevlar thread instead, since it is thinner and could potentially allow us to make thinner rolling diaphragms. However, we first had to spool the kevlar thread onto a bar that would fit the sewing machine. We did this by putting the bar on an electric drill and letting it twist. I realized that the kevlar allowed me to do straighter and better stitches because it was really thin and a little more rigid. However, when the time came to put this fabric inside the mold, it fit too tightly. We then went to cast the rolling diaphragms and I got to cast them myself with the help of alex.
On Friday I went and worked on post processing the 3D prints for new molds. I learned about the different types of resin baths and the UV cure oven and how the different resins need to be cured differently. Alex also came up with the idea of cutting our own pattern into the silinylon. However, I did not get the chance to cut it myself because there was a long line at the prl due to a summer camp.
On Thursday we went to the prl to try and make a railing to feed the fabric perfectly straight on the sewing machine and have parallel lines to prevent bulging on the rolling diaphragm. I used the laser cutter to cut rails and bars to hold the fabric taught. Then alex also taught me how to cut the silinylon for the other type of rolling diaphragms that seems to be working. After we came back we realized that our cast of the kevlar mold with the paper did not come out well because it was only cast on one side and it peeled off. The silicone did also not cure with the new molds.
On Tuesday, I worked on different ways to make the fabric for the rolling diaphragm. We finally got the water soluble paper so I tried gluing the kevlar in parallel strips with the new jig I had made. The kevlar looked really parallel and seemed to look well. We also tried printing straight lines on the dissolvable paper and sewing onto it. However, we realized that this did not work to well because the stitches on the mesh became to loose when the paper was dissolved. After this Alex showed me the proper process to cast the silicone meshes. We also then assembled as a lab the new cupboard/stand for the 3D printer.
Today while Alex was out, I worked with Wilson on finally deciding the shape for the new climbing hand. After some brainstorming we decided to make the climbing hand look more like a bear claw but still remain two-sided. With one side having a granular jamming palm with 3 spiky fingers (like the ones in the old hand) and the other side having three ice axe like structures.
After working with Wilson, I then worked on learning about arduino, to help Alex with a device to push the two parts of the casting mold for the silicone together while it is being degassed. I learned about bread boards and how to code the motors to do different things. I also worked on CADing a container for the casting molds that would hold them in place while the two molds are being pushed together.
On Thursday I spent most of the morning redesigning the brush for the sewing of the kevlar. I redesigned it so that both sides and can be used and we can maximize the production of the fabric and minimize waste. This was definitely a learning experience as I got to use solid works and become more familiar with the different features like doing mirroring of of a plane. I found OnShape to be really useful for continuing to work on my pieces at home and I am slowly becoming more familiar with it. The new design is shown in the image below.
After Wilson came in, I started working on sanding down the new prototype for casting the silicone casing for the particle jamming hand. There had been issues with bubbles and casting before so I tried to sand it as smoothly as possible. When I was sanding the mold Mark taught me some tips for casting urethane. Degassing the mold to get rid of some air bubbles before casting is a good step and so is turning the vacuum on and off because the urethane might boil.
After sanding the mold, I had to drill holes for dowel pins in the mold. This turned out to be challenging because they had to be at exactly 20 degrees so that all the spikes are inserted at the same angle. To drill these tiny holes we had to turn the drilling machine at a 20 degree angle, which reduced the human error of having to tilt the mold at exactly 20 degrees. I also had to use a really tiny 1mm drill bit.
However, the job was not done with just drilling the holes for the dowel pins. We also had to make some calculations for the length of the dowel pins in the mold and find a way to cut them all at the same length. We tried snipping them at exactly 15.43mm every time but this was not very accurate and we failed. I then thought of using a block of wood that was exactly that length and drill holes into them where we could put the dowel pins in and then cut the remainder. Although it required several iterations to make this process successful and efficient by the end we improved it so that cutting 20 dowel pins only took a couple minutes rather than an hour.
Although I spent most of the morning on Friday continuing to design the brush for the fabric of the rolling diaphragm, the rest of the day was spent on doing lab cleanup.
On Monday, I spent more time looking at the grasping mechanisms of the bear as well as other animals like pangolins. I constructed a foam core model to see how the bear grasped onto things. It turns out that it is the long claws that allow the bear to better climb up things. After finishing up with my bear claw I helped Capella in cleaning parts that she had printed in the objet printer. It turned out to be really time consuming and they had to be carefully done so as to not ruin the sizing for the press fit.
After Wilson came in on Monday we picked up the test tubes from the TLTL with our urethane samples. As mentioned earlier, we had poured urethane at two different times into each of these tubes, varying the times when we poured the urethane for the second time. After opening the tubes up we realized that it wasn't the curing time between the pours that affected the bonding between the urethane layers. So probably it was the demolding spray in between the layers that affected the bond.
On Tuesday, I worked with Alex on trying to make different fibers for the rolling diaphragm. After talking with Mark we thought of different ideas to better weave the fabric for the rolling diaphragm. One of the ideas that came up was using a brush to weave the kevlar fibers perpendicular to each other onto the mesh and then cast it. Sewing them in the sewing machine was not working so perfectly because the fibers did not always get sewn perfectly perpendicular to each other. Therefore, using the comb and doing it by hand might work best. We also needed to find a way to keep the Kevlar fibers together before casting them since this seem to be an issue. Alex had already made a prototype of a brush from before so I spent the rest of the day using that prototype to find different ways to sew the kevlar onto the mesh.
Day 2 & 3
On day 2 I started working with Wilson on the climbing robot. He started by explaining his vision of how the robot hand would look like, with one side having a particle jamming pad and the other having an Ice pick like side to cling onto holds. After explaining how the hand worked, I practiced prototyping it with legos and rubber band.
After prototyping, we went to the TLTL to do some urethane curing testing. I learned how to cast the urethane used for the pad and use the degassing machine. The particle jamming pad when manufactured at two different times and put together has not been curing together properly so we tried to figure out at what times the urethane best cures together. We therefore poured some urethane in multiple tubes and then poured more urethane at multiple time intervals.
The next day, I started looking the human hand articulation to look at grasp. I was inspired by the sticky bot to use push and pull wires to articulate the hand as well as a paper on the DLR hand (http://journals.sagepub.com/doi/pdf/10.1177/0278364912459209). However, although my prototype articulated properly, it enabled me to see how the human hand may not be the most optimal mechanism for a climbing robot. It also has many degrees of freedom with twisting and bending movements, which are useful for pinching and grasping things but may not be necessary for climbing. Therefore, I started looking at the pangolin and bear claw as a sources of inspiration. After looking at a paper on bear claws and human feet (https://www.fws.gov/lab/idnotes/idg11_bearhumanfeet.pdf) , I learned that bear paws and human feet are very similar in structural anatomy and the main difference is the presence of claws. However, I have decided to make a foam core prototype to better understand how the bear paw works, which I am currently working on. I think it may be interesting to have articulated bear claws for the climbing hand combined with a particule jamming pad on where the paw pad would be on a bear as opposed to having them on opposite ends of the hand.
After doing some prototyping and readings, I worked on putting together the prototype for Wilson's claw. I drew and cut some rods and a handle for the claw and then we put together the 3D printed structure at the TLTL where I got to work with the band saw and power drill.
After the clean up today, we also did some brainstorming on how to best put the hooks into the silicone molds because currently sometimes the urethane breaks. It was really interesting looking at how to improve the molds and the ideas the came about, including perhaps using a fluid based gripper instead of particle jamming or using a twisting fibre that allows claws to come together around a rock. I am currently reading papers on particle jamming to find a coarse yet smaller particle than coffee to improve the gripping.
Today I helped Stephanie with improving the pulley system in the SuperSCAMP. The string kept on slipping off the pulley and therefore the legs of the SCAMP did not move. At first I thought of using silicone to increase the coefficient of friction between the pulley and the string but as the pulley rotated, the string rubbed the silicone off and eventually the string would slip again. I then started designing a pulley with grooves, on solid works, to prevent the string from slipping but before I printed it, Prof.Cutkosky suggested using the same pulleys we had but instead wrapping the string around several times as this would increase the friction exponentially.
As I tried to do this, I quickly encountered some challenges, the string we were using was too thick and did not fit without overlap around the pulley twice and the pulley was V-shaped allowing the strings to become tangled more easily. Moreover, the string was not feeding in tangentially so this made the string slip off the pulley. To fix this I changed to using thinner strings and tested Kevlar and Spectre fibre, which despite being smoother did not slip when they were turned around the pulley twice. To allow the strings to come in tangentially to the pulley, I used bolt nuts and glued them to the frame. However, this did not work as the string had to feed tangential both horizontally (on the sides of the pulley) and outwards (in the direction of the face of the pulley). I then modified this to be made out of a other pieces which were longer and would allow proper alignment. This improved the feeding of the string into the pulley but did not prevent the strings from getting tangled. Therefore, tomorrow, I will try and design an wider pulley that is not v shaped and spirals to prevent the strings from getting tangled and see how it works.