Visual versus visual-inertial guidance in hawks pursuing terrestrial targets

Abstract

The flight behaviour of predatory birds is well modelled by a guidance law called proportional navigation, which commands steering in proportion to the angular rate of the line-of-sight from predator to prey. The line-of-sight rate is defined with respect to an inertial frame of reference, so proportional navigation is necessarily implemented using visual-inertial sensor fusion. In Harris' hawks, pursuit of terrestrial targets is even better modelled by assuming that visual-inertial information on the line-of-sight rate is combined with visual information on the deviation angle between the attacker's velocity and the line-of-sight. Here we ask whether a new variant of this mixed guidance law can model Harris' hawk pursuit behaviour successfully using visual information alone. We use high-speed motion capture to record n=228 attack flights from N=4 Harris' hawks, and confirm that proportional navigation and mixed guidance using visual-inertial information both model the trajectory data well. Moreover, the mixed guidance law still models the data closely if visual-inertial information on the line-of-sight rate is replaced with purely visual information on the apparent motion of the target relative to the background. Whilst the original form of the mixed guidance law provides the best model of the data, all three models can model the behavioural data phenomenologically, whilst making different predictions on the physiological pathways involved.

Obstacle avoidance in aerial pursuit

“…The full dataset contains the following subsets: (i) n=128 obstacle-free training flights collected over 8 days; followed by (ii) n=16 obstacle familiarization flights collected the next day; then (iii) a set of n=106 obstacle-free test flights; and (iv) a set of n=154 obstacle test flights; where (iii) and (iv) were collected over 15 days on which the presence or absence of obstacles was randomized (see Materials and Methods for details). The n=106 obstacle-free test flights are reported and included in an analysis of unobstructed pursuit elsewhere [7], but the flights with obstacles are described here for the first time. We simulated the measured data computationally using several alternative models of the guidance dynamics, matching the hawk’s simulated flight speed to its measured flight speed, and modelling its horizontal turning behavior using the mixed guidance law in Eq.…”

“…Although it is possible that other guidance laws [15] might explain our hawks' pursuit behavior as well as the mixed guidance law we have fitted (Eq. 1), our modelling demonstrates high repeatability in the guidance parameters fitted across hundreds of flights collected under varying experimental conditions (Fig.…”

“…The birds included one 7-year old female (Ruby) that had been included in a previous study [11], plus three first-year males (Drogon, Rhaegal, Toothless) that had not previously chased a target. A subset of the flights without obstacles that we report are described and analyzed elsewhere using a related method [15], but the flights with obstacles are reported here for the first time. Each bird usually flew after the lure four times per day, taking off spontaneously from the falconer's gloved fist when the lure began moving.…”

“…The full dataset contains four subsets: (i) a set of n=128 obstacle-free training flights collected over 8 days; (ii) a set of n=16 obstacle familiarization flights collected the next day; then (iii) a set of n=106 obstacle-free test flights; and (iv) a set of n=154 obstacle test flights; where (iii) and (iv) were collected over 15 days on which the presence or absence of obstacles was randomized (see Methods). The n=106 obstacle-free test flights are reported and included in an analysis of unobstructed pursuit elsewhere [15], but the flights with obstacles are described here for the first time.…”