Artificial potential field implementation of flying animal gap-aiming behavior in 3D

This paper presents an artificial potential field implementation of flying animal gap-aiming behavior to autonomously navigate collision-free in an unknown and disordered indoor environment. Current methods for autonomous navigation on small unmanned aerial systems use obstacle detection for relative navigation or bio-inspired techniques, but lack the ability to operate in three dimensions or would not perform in an environment with closely spaced obstacles. Using flying animals as inspiration, a behavior-based robotics approach is taken to implement and test their observed gap-aiming behavior in three dimensions. Simulations are run to determine the best approach for perceiving gaps in the environment for action from three available options: largest, closest, or all gaps in the field of view. The results of each approach are compared by the time, straightness, and safety of the flight path produced. The results of the simulation experiment show the implementation can successfully produce the desired gap-aiming behavior. However, the best approach could not clearly be determined because there was no statistical significance between the values for two of the three measured metrics. Four avenues to improve the gap-aiming behavior are discussed and left to future work.