My Robot Will Crush You With Its Soft Delicate Hands!

Soft robots are slowly trickling out of universities and labs into everyday life. Amazon is experimenting with installing soft grippers on robotic arms to pick any product off a warehouse shelf. DARPA just funded an extensive program to build soft exoskeletons for soldiers to enhance how much they can lift and how long they can march. My lab, Super-Releaser, is developing robotic spacesuit components for NASA as a subcontractor on a SBIR grant. On paper they might seem too complex to whip up at home, but if you’re the kind of person who loved Creepy Crawlers and have access to a 3d printer you can make your very own soft robots. Let’s take a step back to explain what a soft robot is and what they're good for. Most robots out there are made from hard parts like steel gears and plastic housings with the occasional rubber wheel or timing belt thrown in. When they respond to their environment it’s usually by reading sensors and using a processor to change their behavior accordingly. This responsiveness is called compliance. There’s another way to get compliance out of an engineered object, though: make the object soft. Everything from goat hooves to octopus tentacles, starfish suckers to human muscles, use softness and springiness to their advantage. When your robot responds to the environment by bending, say around the thing you’re trying to grip, getting a specific output, like putting that thing in a box to get packed up and shipped off, becomes a lot simpler on the computation side of things. Adding compliant mechanisms to your engineering toolbox can add huge problem solving power to any robotics problem you're trying to tackle. In this talk I’m going to describe how I go about fabricating soft robots. There are lots of methods out there – from heat sealing, to stitching, to direct printing – but the method I prefer is casting. I like casting my robots from 3d printed molds because I can test multiple designs in parallel (just print out the different molds with, say, different wall thicknesses or numbers of ribs inside the actuators, all at the same time). It also allows me to experiment with designs in a context that matches the material I’d be using if production were scaled up for mass manufacture. Finally, if I decide I really like the robot I’ve created, making a dozen of them is just a matter of casting that same mold a dozen times. I’m going to go over the materials I use and where to find them online, how to go about designing your own soft robot, and some interesting problems in soft robotics that are just waiting for solutions. I’m going to be taking the audience from building the simplest actuators, to methods for fastening parts, to getting airtight seals even at high pressures, to putting everything together into a single-piece walking quadruped.