A model of the cockroach antenna links tactile features to distinct motifs on a soft sensor

The parsing of rich spatiotemporal information to guide decision-making is a hallmark of biological systems. However, the mechanisms behind how biological sensors process vast amounts of information remain poorly understood. A broad class of animals rely on touch sensation for perception. Among these, the American cockroach P. americana uses a pair of soft antennae for tactile exploration and guidance. We hypothesize that antenna mechanics is tuned to breakdown tactile features into a lexicon of spatiotemporal patterns (strain motifs). To test this hypothesis, we developed a mechanical model of the antenna (flagellum). Inspired by the morphology of the flagellum, our model included a sequence of rigid links interconnected by hinge joints. From mechanical testing data, we modeled the stiffness and damping of these joints using nonlinear optimization. We simulated the mechanical response of the f lagellum using the MuJoCo physics engine and demonstrated the model’s high fidelity to the actual antenna. Our simulation revealed how antenna mechanics represent a diverse set of tactile features into a lexicon of strain motifs. These motifs may assist in decision-making during slower tactile exploration and rapid course control. Our mechanical model is inspiring a soft, distributed robotic sensor whose shape and dynamics can be modulated.

Download