A robotic model of efficient prey finding in the gleaning bat Micronycteris microtis.

The common big-eared bat (Micronycteris microtis) gleans its prey directly from plant surfaces. Previous experiments have shown that this bat exploits the specular reflection effect to discriminate between occupied and empty leaves. Using this effect requires the bat to position itself such that it can ensonify the leaves at sharp angles. If the bat can perceive the orientation and position of individual leaves, it could position itself to take advantage of the specular reflection. However, this would be a highly inefficient foraging strategy. The bat would first have to inspect each leaf to determine its position and orientation. Given that, under natural conditions, the vast majority of leaves do not feature prey, this would require the bat to spend most of its time collecting information on empty leaves. Here, we propose a strategy that allows a bat to exploit the specular reflection effect without inferring the position and orientation of individual leaves. We implement this strategy on a robotic arm equipped with a sonar head. The robot is tasked with finding a 3D-printed dragonfly on one of a set of artificial leaves. The robot follows an echo amplitude gradient and abandons this search whenever the echoes become too weak. Importantly, the robot does not actively find or locate individual leaves. We show that the proposed sensorimotor model can exploit the specular reflection effect to efficiently and effectively locate prey. Our results increase the plausibility of M. microtis using the specular reflection effect under natural conditions.