Moving without a single rigid link
A pneu-net actuator is a silicone body cast around a network of internal air chambers, with strain-limiting material on one face. Inflate a chamber and the unconstrained side stretches while the strain-limited side doesn't, forcing the whole segment to curl. String several of these chambers together and inflate them in sequence, and the body can push, curl, or crawl forward, locomotion generated entirely by soft material deforming, with nothing that could be called a joint anywhere in the system.
For this build, inflation was manual rather than pump-driven: syringes fed each chamber directly, which meant the actuation sequence and timing were controlled by hand during testing rather than by an onboard controller, a deliberate simplification that kept the focus on the soft mechanics themselves rather than the pneumatic control system around them.
Clip 03 · friction test in rice bed
Why a rice bed
Locomotion by chamber inflation only works if the body can grip its surroundings enough to push off: too little friction and it just squirms in place. The rice bed wasn't a demo prop; it was a controlled way to test how much surface friction the actuator needed to actually translate chamber expansion into forward motion, before relying on results from a single fixed test surface.
My role: the part that isn't CAD
My contribution to this build was almost entirely physical rather than digital: I wasn't heavily involved in modeling the chamber geometry, and instead led fabrication and assembly of the soft components. That meant mixing and curing silicone to spec, working molds for the chamber networks, and assembling the cured sections into a working body. It's a different skill set from CAD-driven design work: silicone behaves differently batch to batch, cure time and mix ratio both move the final material properties, and a lot of the real learning was hands-on rather than something you can simulate beforehand.