A biomimetic soft robotic pinna for emulating dynamic reception behavior of horseshoe bats.

Encoding of sensory information is fundamental to closing the performance gapbetween man-made and biological sensing. It has been hypothesized that the coupling of sensingand actuation, a phenomenon observed in bats among other species, is critical to accomplishingthis. Using horseshoe bats as a model, we have developed a biomimetic pinna model with asoft actuation system along with a prototype strain sensor for enabling motor feedback. Theactuation system used three individually controlled pneumatic actuators per pinna which actuateddifferent portions of the baffle. This prototype produced eight different possible motions thatwere shown to have significant effects on incoming sound and could hence function as a substratefor adaptive sensing. The range of possible motions could be expanded by adjusting the fill andrelease parameters of the actuation system. Additionally, the strain sensor was able to representthe deformation of the pinna as measurements from this sensor were highly correlated withdeformation estimates based on stereo vision. However, the relationship between displacementsof points on the pinna and the sensor output was nonlinear. The improvements embodied inthe system discussed here could lead to enhancements in the ability of autonomous systems toencode relevant information about the real world.