This will be an update on the things I’ve learned molding quadrupeds over the last couple of months and some previews of the new robots I’ll be experimenting with in the next few weeks. To start, I’ve had the chance to run a gaggle of design experiments ranging from small changes to the particular silicone I’ve been casting, to more radical changes to how the whole plionics manufacturing process comes together.
I’ve discovered that molding complex channels of tubing can be extremely difficult, and the CAD equally infuriating. I’m discovering some automatic routing tools in SolidWorks that could streamline the process, but there might be another solution that sidesteps that whole mess entirely. It’s possible to cast around silicone tubing that’s already connecting up all the interior geometry. So, what I’d have to do to get the design working is build the cores with little fastenings for plugging in tubing and make sure all the tubes have enough clearance to get past one another. I’m anticipating the world of reality doesn’t let me off the hook that easily, but it’s a start.
I’ve tried a few experiments with molds containing fixtures that hold all the wax pieces in place for the first step of the casting process, and then get replaced for a second pour of silicone to give the robot its final surface. They’ve been successful enough that I’m basing the three new bots I’ve designed on this method.
In the past I’ve had trouble with the cores (pieces that fit into the robot mold that eventually get removed to form the hollow volumes inside the silicone skin) drifting around and creating an irregular wall thickness in the final part. Since the bots are pneumatic that can make a huge difference to their motion and needs to be reigned in if this is going to approach being a practical technology for doing important stuff. I’m seeing two solutions in the near future. The first is casting all the cores in a hard substance that is soluble in something nontoxic or melts at a lower temperature than the silicone. I’m eyeballing pewter and a few other hobby casting metals for the purpose, but they stand a chance of both throwing off the rubber’s cure and adding a layer of design complexity when it comes to meshing all the core pieces together without much room for bending the pieces to get them into place. The second, and more intriguing, option is creating fixtures inside the mold for keeping the wax cores where they’re supposed to be. Jim over at Viridis3d is also tinkering with options with printing directly in a soluble material, so here’s hoping that gets off the ground.
Here’s a picture of the fixtures on the bottom plate of my first quadruped design. It did a pretty excellent job of keeping all the different wax bodies in place for the first pour of silicone. You can see the casting that resulted from that pour to the right of this paragraph. The next step was to clap on the replacement mold piece and do a pour from that side. Here’s a pic of the replacement mold with its removable sprue. Given that the coming revised quadruped design is a whole lot more complex, including interior ribs like the rest of my tentacles, it seems like fixturing is an essential development in getting this tech working.
Oh, and as a little addition to the bathtub molding I was describing in an earlier post, I was able to cast really good waxes at home once the molds were properly ventilated. I used a custom venting tool I’d made specifically for perforating silicone molds years ago in college. It’s just a piece of sharpened brass tubing with a miniature file handle stuck to the end.
Check out some more photos from the latest quadruped experiments on Flickr (scroll to the bottom of the page to see the latest pics).
I have two prototypes currently getting printed that I’m hoping can shed some light on whether these bots can be applied to problems out there in the real world. I’ve designed two takes on an orthotic arm cuff. The plan is to create a series of soft pliant robots to slip over the elbow and control arm extension. One uses geometry similar to the tentacles that I’ve already gotten to work pretty reliably. The other has a wide inflating patch that should help spread the force across the upper and fore arm.
After talking with my dad (an orthopedic surgeon specializing in arthroscopic surgery) it seems like getting the elbow to extend is an essential step in getting someone mobile even if they have severe nerve damage. Apparently if you’re able to move both elbows reliably, and have enough strength to lift yourself, you can be fully paralyzed below the midriff and be reasonably mobile through a combination of braces, wheelchairs, and french crutches. Being able to get around autonomously, without the need for a full time assistant, makes a huge difference in peoples’ quality of life and makes it much more likely that a person can participate in everyday life like going to the grocery store or taking the bus instead of Access-A-Ride.
The advantage to using a soft orthotic device, as opposed to a hard motorized one, is how easy these are to manufacture (the difference between making one and a hundred soft robots is simply how many times you pour silicone into the mold, in principle), that a single device could last a child through many stages of growth (as opposed to needing to be refit for multiple custom devices as they grow), and that they have the potential of distributing the force of moving the arm around more evenly leading to less fatigue and pressure point injuries over time.
I’ve got a set of new renders of these designs and the molding that generates them over on Flickr.