Animal-inspired agile flight using optical flow sensing

Abstract

There is evidence that flying animals such as pigeons, goshawks, and bats use optical flow sensing to enable high-speed flight through forest clutter. This paper discusses the elements of a theory of controlled flight through obstacle fields in which motion control laws are based on optical flow sensing.Performance comparison is made with feedback laws that use distance and bearing measurements, and practical challenges of implementation on an actual robotic air vehicle are described. The related question of fundamental performance limits due to clutter density is addressed.

Optical flow sensing and the inverse perception problem for flying bats

“…This simple argument is very similar to the way in which τ was introduced in [14] and [18], but with an extremely important difference. The analysis in [14] and [18] was carried out for a forward-looking imaging system, whereas here we have treated a side looking imaging system.…”

“…Sebesta and Baillieul [18] showed how the well-known optical parameter τ could be used to guide the motion of a moving optical sensor. It is worth redoing this analysis with a special emphasis on the eye geometry of the M. velifer.…”

“…It is worth redoing this analysis with a special emphasis on the eye geometry of the M. velifer. In [14] and [18], τ was described in terms in planar rendering of observed animal trajectories (per Section II), models of motion in the plane are appropriate.) The direction of motion is aligned with the vehicle body frame x-axis, and the feature O is observed with a pinhole camera system whose camera axis is aligned with the negative body frame y-axis.…”

“…The analysis in [14] and [18] was carried out for a forward-looking imaging system, whereas here we have treated a side looking imaging system.…”

“…Terms such as "looming" and "time-to-collision" have appeared infrequently in the literature of mobile robot control, with notable exceptions being [15], and [18]. Nevertheless, there is a growing literature dealing with biological sensorimotor control in which vision based sensing of the time-to-transit key feature points is seen as playing a central role in regulating motion behaviors ( [12], [14], [26]).…”