"Mechanisms as Minds: What Camping Tents and Caterpillars Can Teach Us About Soft Robotics"
John Rieffel, Ph.D.
Computer Science Department
March 22, 2010
12:50 - 1:40 pm
367 Votey Hall
Conventionally, roboticists have chosen animals with relatively rigid morphologies, such as dogs and cockroaches, as a source of inspiration. This choice is natural and convenient, because the underlying biomechanics are relatively well understood, and because rigid-bodied robots are relatively easy to design, build, and control.
There are, however, inherent limitations to these designs: they cannot quickly change shape, they cannot navigate uneven terrain, and they cannot squeeze through holes and cracks. In order to build a new generation of robots without these limitations, engineers must cast their net wider, to the realm of soft-bodied animals, such as octopi and caterpillars. Unfortunately, soft bodied systems, both natural and artificial, are rife with the kinds of complex dynamics, such as resonance, anisotropy, and dynamical coupling, that engineers strive to avoid. In my talk I will argue that dynamical complexity can be a boon rather than a bane to robot designers, and that, paradoxically, as some systems become more mechanically complex they become easier to control. I will present several examples of robotic systems in which mechanisms act as minds, and bodies act as conduits for information.
Aside from helping us better understand the biomechanics of completely soft animals such as caterpillars, these ideas are crucial for developing control schemes for the emerging field of soft robotics.