Meeting Abstract

42.1  Thursday, Jan. 5  A bird’s eye view of path planning: Is there a simple rule for flying in a cluttered environment? LIN, H*; ROS, I/G; BIEWENER, A/A; Harvard University; Harvard University; Harvard University huaiti.lin@gmail.com

For a bird, flying through cluttered environments is challenging due to the complexity of 3D obstacles and high speed of flight. A variety of navigation strategies have been shown on the scale of migration and homing behaviors. However, relatively little is known about how birds negotiate nearby obstacles when flying in a cluttered environment. To address this question we tracked pigeons (C. livea) flying through a vertical pole forest in which the obstacle distribution can be manipulated between trials. Interestingly, pigeons did not adjust their flight path until ~1.5m before the forest. Once in the forest, the average flight speed dropped from ~6m/s to ~3.5m/s and the wing-beat frequency increased from ~7Hz to ~8.8Hz. Head saccades were rare during these flights, indicating that the overall optical flow might be more important than frontal vision in guiding maneuvers. Furthermore pigeons aimed for large openings in the forest only when these paths were visible in the angular sense. This observation suggests that pigeons’ depth perception may be insufficient to reconstruct an ‘obstacle map’. Instead, pigeons might rely on angular positions of the obstacles and optical flow for close range path planning. Combining the flight trajectories and obstacle pole positions, we recreated the pigeon’s visual field during flight. The obstacle poles can be represented as looming vertical bars across the bird’s visual field. We constructed a simple navigation model that steers a pigeon toward the largest frontal opening and tested different range estimating strategies (e.g. looming cues) given a specific set of experimental visual inputs. Such a simple navigation model explains a large number of pigeons’ flight trajectories through the cluttered environment. The results could provide insight for controlling high-speed unmanned aerial vehicles with simple visual feedback.