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

There have been many pneumatic actuator designs created over the past several years. One such design consists of multiple, completely separated air cells that when pressurized, expand and push off one another causing the actuator to bend. With this chamber design, the actuator requires a small change in volume to deform, effectively leading to improved reliability and increased actuation speed. Other design characteristics such as cross section geometry, inner/outer wall thickness ratio, distance between cells, cell height, and material selection can be modified to optimize the actuator’s performance. Incorporating this design into the bat robot allowed for greater soft robotic ear deformation, however it is too large for this application. To see if it was possible to reduce the size of the actuator while maintaining its functionality, several adjustments were iteratively made to previously described mold designs. Manufacturing remained a two-step process, but removable side walls and a filter-ventilation system were added to allow for the silicone, used to cast the actuator, to fully cure. Ultimately, the actuator produced maintained its expected functionality and was produced at a smaller scale than originally thought possible.

A crucial development in the field of biomimicry is accurately recreating the kinematics of various organisms. In the example of a bat robot in development inspired by the greater horseshoe bat, Rhinolophus ferrumequinum, this takes the form of accurately recreating fast motions of the pinnae and noseleaf observed during echolocation. In order to accurately recreate these baffle shapes it is important to choose a material that is malleable but will retain its originally molded state. An actuation system was developed to replicate the ear movement seen in the bat when using biosonar. Several actuation designs were tested, including a physical motor connection, shape memory alloys, and soft-robotic pneumatics. In addition to the actuation system, a feedback system was developed in order to accurately control the robot and provide information which could be used to determine the effectiveness of the robot. Additionally, the final challenge to assemble all the components, i.e., the baffle shapes, actuator mechanism, and feedback control mechanism, in a way that recreates the baffle motions in an effective and accurate manner.