For use in autonomous micro air vehicles, visual sensors must not only be small, lightweight and insensitive to light variations; on-board autopilots also require fast and accurate optical flow measurements over a wide range of speeds. Using an auto-adaptive bio-inspired Michaelis–Menten Auto-adaptive Pixel (M2APix) analog silicon retina, in this article, we present comparative tests of two optical flow calculation algorithms operating under lighting conditions from 6×10−7 to 1.6×10−2 W·cm−2 (i.e., from 0.2 to 12,000 lux for human vision). Contrast “time of travel” between two adjacent light-sensitive pixels was determined by thresholding and by cross-correlating the two pixels’ signals, with measurement frequency up to 5 kHz for the 10 local motion sensors of the M2APix sensor. While both algorithms adequately measured optical flow between 25 ∘/s and 1000 ∘/s, thresholding gave rise to a lower precision, especially due to a larger number of outliers at higher speeds. Compared to thresholding, cross-correlation also allowed for a higher rate of optical flow output (99 Hz and 1195 Hz, respectively) but required substantially more computational resources.

This paper presents the autopilot and the behavior of a "simulated bee" traveling along two different tunnels using both visual and control events. The computational gain of an event-based PID controller compared to its time-based version is usually discussed because the event detector is computationally expensive. By combining visual and control events, the newly suggested event-based autopilot requires very low computational resources. In particular, the event detector which computes the control error and tests its magnitude is activated only when a new contrast is detected by the optic motion detectors that assesses the optic flow, i.e. only when the magnitude of the optic flow error could have changed. This new event-based control strategy used faithfully the visual information already available in the optic flow sensor to reduce even further the computational cost. The "simulated bee" was equipped with: (i) a minimalistic compound eye comprising 10 or 8 local motion sensors (depending of the tunnel configuration) measuring the optic flow magnitude, (ii) two optic flow regulators updating the control signals whenever specific optic flow criteria changed and (iii) three event-based controllers taking into account both error signals and visual events, each one in charge of its own translational dynamics. The "simulated bee" managed to travel safely along the tunnels without requiring any speed or distance measurements, using very low computational resources, by (i) concomitantly adjusting the side thrust, vertical lift and forward thrust only when both a visual contrast and a change of optic flow control error were detected, and (ii) avoiding collisions with the surface of the tunnels and decreasing or increasing its speed, depending on the clutter rate perceived by motion sensors.

Abstract-In this study, a quasi-panoramic bio-inspired eye dedicated to optic flow measurement on board micro flying robots is presented. It will allow future micro flying robots to mimic honeybees' navigational tasks which work without any metric sensors. An innovative optic flow-based algorithm was tested in the horizontal plane to measure the robot's incidence angle when travelling along a wall. Experimental results achieved while using a moving texture showed that our algorithm is able to measure the local visual heading with respect to a tapered wall with an error of 3• and an accuracy of 3• on average. Our bio-inspired eye will be implemented in the near future on board a micro quadrotor to "sense and avoid" obstacles in a GPS-denied environment.