Optical flow and motion detection for navigation and control of biological and technological systems

Abstract

We review neural and computational measurement of optical flow. Insects are sensitive to visual motion which has been established by elecrophysiological and behavioural studies over more than half a century. We discuss the elementary motion detector theory of how optical flow is detected in the nervous system and the limitations of this theory in predicting some insect visual behaviour that is apparently driven by a measurement of optical flow. The use of optical flow techniques in aerial robotics applications that mimic the challenges faced by flying insects are compared. The limitations of optical flow measurements in computing distance to targets or speed of travel is observed and the merits of proposed solutions discussed. Finally we examine a range of optical devices that were designed to overcome the limitations of monocular optical flow and consider the new limitations that they create which have prevented their wide acceptance in robotics.

Keywords:

• Optical flow
• elementary motion detector
• insect vision
• vision
• navigation
• unmanned aerial vehicle
• drone
• aircraft
• robot
• optics
• optic flow

Acknowledgements

The author would like to thank Mandyam Srinivasan, Eric Warrant and Akiko Mizutani for some helpful discussions about the state of the art in biological motion detection. This work was supported by Tyche, the Defence Science and Technology Group’s Trusted Autonomy initiative.

Notes

No potential conflict of interest was reported by the author.