Motion 3Historically, many machines (especially robots) were designed to mimic the motions of humans and other animals, but to add power, speed, "endurance," and reproducibility to those motions. [1] The robots on industrial assembly lines, for example, extend the capabilities of the workers that originally carried out assembly by hand using simple tools, by adding power, complex tools, indifference to environmental conditions, and mechanical "endurance". Similarly, four-wheeled robots are loosely derived from fourlimbed animals, and aerial drones can be traced in design backwards in time through manned aircraft, to birds. Conventional machines-especially those fabricated from metal, ceramics, and structural polymers (so-called "hard" machines)-can carry out almost arbitrarily complex motions using pulleys, cables, gears, and electric or hydraulic actuators. To achieve controlled motion, however, they also normally require complex systems for active controls (networks of sensors, actuators, and feedback controllers). [2,3] Some of these "hard" systems are exquisitely and highly developed, but can be heavy, energy inefficient, dangerous to humans, and expensive.We are exploring soft actuators and robots-machines modeled after simpler animals (e.g., starfish, worms, and squid) having no hard internal or external structures, and fabricated entirely or predominantly in soft, compliant polymers. [4,5] The first generation of these systems-originally sketched by Suzumori, [6][7][8] and then realized and elaborated by us, [5,[9][10][11][12][13] and by others [4] -use pneumatic actuators, comprising networks of micro-channels; in our systems, differential expansion of these pneumatic networks (PneuNets) by pressurization using air produces motions (especially bending, curling, and variants on them) that are already established as useful in grippers, and interesting for their potential in walkers, tentacles, and a number of other soft, actuated systems. [14] 4 Although the design of the first of these systems has been relatively simple, the motion they produce on actuation can be surprisingly sophisticated: for example, a representative structure-the "finger" or "tentacle" of a gripper-curls non-uniformly, starting from its tip and proceeding to its stem, although the pressure applied in the PneuNet is approximately uniform throughout the system of inflatable channels. [10,11] This motion reflects a non-linear property of soft materials and structures, referred to as a "snapthrough instability". [15][16][17][18][19] Although nonlinear properties of materials are often considered a disadvantage, this type of non-linearity, illustrated by snap-through, and other complex mechanical characteristics of soft systems, are proving to be useful, and to offer new capabilities to effectors, machines, and robots, because they enable a range of motions of sufficient complexity that-although they might be possible to replicate in a hard robotic system [20] -it would be complicated and expensive to do so. This paper demonstrates the ut...