Decoding frequencies generated during spider vibration sensing through robophysical modeling

Orb-weaving spiders detect web intruders through vibration sensors in their legs. They have been observed to assume different leg postures and repeatedly crouch their legs during vibration sensing. This behavior is hypothesized to be a form of active sensing—by modulating the vibration dynamics of the spider-web-target system, the spider may better detect prey. However, whether this is true and how it works is poorly understood due to challenges in measuring the whole system's vibration dynamics with actively behaving spider and prey. Here we developed a robophysical model of the spider-web-prey system. We measured the resulting vibrations of the web and the spider robot's legs, when the spider robot and/or the prey robot were actuated at various frequencies. In all scenarios, actuation of either the spider robot, the prey robot, or both induced an added frequency component to the web vibration, which likely corresponds with the spider-web-target systems' natural frequencies. We hypothesize that this added frequency component can be modulated by the legs to enhance sensing. We are performing further experiments and developing a dynamic simulation to better understand how both robots' behaviors influence this component to better understand active vibration sensing on the web.