Fig 2 - uploaded by Suzanne Amador Kane
Content may be subject to copyright.
Schematic illustration of headcam images and trajectories for various pursuit-evasion strategies. (A) Schematic headcam image, showing the orientation of the horizontal and vertical camera and visual angles (θ, χ). Black arrows indicate the optical flow due to self-motion of the predator at constant velocity. The intersection of the dashed lines indicates the center of motion at (0 deg, 0 deg). An angle, γ, defines the orientation of the prey's velocity (red arrows) with respect to the local optical flow field. Labels indicate a rabbit not tracked (NT) by interception strategies (γ≠0 deg), a pheasant tracked by classical pursuit (CP) and a rabbit tracked by constant absolute target direction (CATD; γ=0 deg). Earth-frame trajectories for each pursuit strategy are depicted as follows: (B) CP; (C) the geometry for the constant bearing (CB) criterion used in CATD; (D) CATD; (E) CPE; (F) finite-feedback implementations of proportional navigation (PN) in which the predator's trajectory (gray curve) can oscillate around the optimal bearing angle trajectory (CB, red curve). The baseline vector, R, is shown as a blue arrow. v e , prey velocity; v p , constant predator velocity; ϕ, bearing angle between R and v p ; β, angle between v e and R; α, the constant angle of R relative to the fixed Earth frame. D is adapted from Ghose et al. (Ghose et al., 2006); F is adapted from Shaw (Shaw, 1985).
Source publication
Video filmed by a camera mounted on the head of a Northern Goshawk (Accipiter gentilis) was used to study how the raptor used visual guidance to pursue prey and land on perches. A combination of novel image analysis methods and numerical simulations of mathematical pursuit models was used to determine the goshawk's pursuit strategy. The goshawk fle...
Contexts in source publication
Context 1
... reviewing pursuit-evasion strategies and their appearance on headcam video, we consider the basic geometry of headcam images for a predator moving at constant velocity, v p , with its head axis along v p ( Fig. 2A). The resulting optical flow field radiates outward from the predator's center of motion ( Lee and Kalmus, 1980). Possible images of moving and stationary prey are also shown in this figure; the prey's position is described using horizontal and vertical camera angles (θ and χ) that map on to the goshawk visual angles. We can define an ...
Context 2
... time-optimal pursuit strategy for stationary prey is classical pursuit (CP), in which the predator's velocity always points toward the prey (Nahin, 2012) so the prey's image remains stationary at the center of motion ( Fig. 2A). However, in general CP is inefficient when the prey moves (Fig. 2B). CP has been observed for bees ( Zhang et al., 1990), flies (Land, 1993;Trischler et al., 2010), beetles (Gilbert, 1997) and bats following conspecifics ( Chiu et al., 2010) and chasing slow prey ( Kalko and Schnitzler, ...
Context 3
... time-optimal pursuit strategy for stationary prey is classical pursuit (CP), in which the predator's velocity always points toward the prey (Nahin, 2012) so the prey's image remains stationary at the center of motion ( Fig. 2A). However, in general CP is inefficient when the prey moves (Fig. 2B). CP has been observed for bees ( Zhang et al., 1990), flies (Land, 1993;Trischler et al., 2010), beetles (Gilbert, 1997) and bats following conspecifics ( Chiu et al., 2010) and chasing slow prey ( Kalko and Schnitzler, ...
Context 4
... explain interception, in which the predator moves toward the prey's estimated future location (Collett and Land, 1978;Lanchester and Mark, 1975), it is useful to define the baseline vector, R, pointing from predator to prey, the bearing angle, ϕ, between R and v p , and the angle β between prey velocity v e and R (Fig. 2C). For constant v e and |v p |≥|v e |sinβ, time-optimal interception is possible if the predator maintains its bearing angle at: This is called the constant bearing decreasing range (CB) criterion (Nahin, 2012). If the prey maneuvers infrequently, then the CB criterion can be applied during each constant velocity interval. This strategy ...
Context 5
... 1 e p When hawks attack: animal-borne video studies of goshawk pursuit and prey-evasion strategies ϕ varies but R has a constant orientation in the Earth frame, as indicated by the angle α in Fig. 2D. ( Ghose et al., 2006;Reddy et al., 2006). The predator can implement CATD by maneuvering to keep the prey's image at constant visual angle (determined by the instantaneous value of ϕ ο ) for fixed predator head orientation (CATD in Fig. 2A). Deviations in ϕ from the ϕ ο set point serve as the predator's control signal, stimulating it ...
Context 6
... strategies ϕ varies but R has a constant orientation in the Earth frame, as indicated by the angle α in Fig. 2D. ( Ghose et al., 2006;Reddy et al., 2006). The predator can implement CATD by maneuvering to keep the prey's image at constant visual angle (determined by the instantaneous value of ϕ ο ) for fixed predator head orientation (CATD in Fig. 2A). Deviations in ϕ from the ϕ ο set point serve as the predator's control signal, stimulating it to accelerate to compensate. For stationary prey (|v e |=0), Eqn 1 gives ϕ ο =0, and CATD reduces to CP. For classical evasion, i.e. the prey flees directly away from the predator (β=0 deg), Eqn 1 gives ϕ ο =0 and CATD reduces to classical ...
Context 7
... in ϕ from the ϕ ο set point serve as the predator's control signal, stimulating it to accelerate to compensate. For stationary prey (|v e |=0), Eqn 1 gives ϕ ο =0, and CATD reduces to CP. For classical evasion, i.e. the prey flees directly away from the predator (β=0 deg), Eqn 1 gives ϕ ο =0 and CATD reduces to classical pursuit-evasion (CPE) (Fig. 2E). CATD is used by dragonflies ( Combes et al., 2012;Olberg, 2012), bats ( Ghose et al., 2006;Ghose et al., 2009) and humans (Fajen and Warren, 2004). CATD is a motion camouflage strategy Reddy et al., 2006) because the predator perceives no prey motion on its visual field and vice versa; the only motion cue is looming (retinal ...
Context 8
... of CATD can be determined from empirical data in several ways. The constant orientation of R in 3D predator and prey trajectories is a definitive test for CATD. In predator headcam video, CATD can be demonstrated if two requirements for a collision course are met: constant prey visual angle and γ=0 deg (CATD in Fig. 2A). Deviated pursuit (DP) is the more general case of constant prey visual angle for ϕ≠ϕ ο and γ≠ deg, so it results in curved predator trajectories even for prey that remain motionless or move at constant velocity (Shima, 2007), as observed for flies tracking fixed targets ( Osorio et al., ...
Context 9
... Reddy et al., 2007), and bats and insects use finite feedback implementations with a time delay between evasive maneuvers and predator responses ( Ghose et al., 2006;Land, 1993;Srinivasan and Zhang, 2004). Depending on the feedback constant and prey motion, the prey's velocity and visual angles on the headcam image can oscillate about a set point (Fig. 2F), or PN can result in a gradually arcing predator trajectory qualitatively similar to those seen for CP and ...
Context 10
... (prey, lure, perch, etc.) in a retinal fixation area at the center of its visual field; during searches for prey, moving objects (prey, the falconer, etc.) were tracked via ~2 Hz head saccades. The retinal fixation area agreed with the center of motion when the goshawk flew toward its target starting either from mid-air or from a perch (cf. Fig. 2A and Fig. ...
Citations
... Practitioners must pass a detailed examination, pass a state-mandated equipment and facilities inspection and serve a two-year apprenticeship under a mentor. Despite its long history, falconry and the use of falconry techniques have yet to be considered as an alternative mobile platform for research, excepting cases where the raptor itself forms the subject of study (Brighton et al., 2017(Brighton et al., , 2021Kane et al., 2015;Kane & Zamani, 2014). ...
Unoccupied aerial vehicles (UAVs; drones) offer mobile platforms for ecological investigation, but can be impractical in some environments and the resulting noise can disturb wildlife.
We developed a mobile alternative using a bird‐borne platform to record the behaviour of other animals in the field. This unit consists of a lightweight audio and video sensor that is carried by a trained Harris's hawk Parabuteo unicinctus.
We tested the hypothesis that our bird‐borne platform is a viable option for collecting behavioural data from mobile animals. We recorded acoustic and video data as the hawk flew through a dense group of Brazilian free‐tailed bats Tadarida brasiliensis emerging from a cave, with a test case of investigating how echolocation calls change depending on spatial position in the bat group.
The HawkEar platform is an alternative for collecting behavioural data when a mobile platform that is less noisy and restrictive than traditional UAVs is needed. The design and software are open source and can be modified to accommodate additional sensor needs.
... In water, sh hunters may pursue prey using various variations of the deviated pursuit strategy ("following"), guided by visual cues (Peterson and McHenry 2022;McHenry et al. 2019). "Parallel pursuit" (Kane et al. 2015) is another pursuit strategy commonly employed by ying hunters to plan their path to intercept prey. It requires the hunter adopting a new angle φ whenever the prey changes its heading or relative velocity (Fig. 1C). ...
As a highly sequentially programmed behavior driven by innate desire, one of the most challenging parts of preying is how the hunter can pursue and capture an escaping prey that is also running for its own survival. Although presumed, it remains uncertain how the experience of competing with escaping prey can enhance preying performance. Here, we developed an interactive platform to study the preying behavior in mice using an escaping bait. This robotic bait was magnetically controlled by a closed-loop system that continuously attempted to evade an approaching threat (e.g., a hungry mouse). By recording the time costs, trajectories and other parameters of both mouse and the escaping bait, we found that mice were not only able to perform preying tasks of varying difficulties, but also that they could improve their preying efficiency over trials, mainly due to the improvements in the pursuit phase. Further investigation revealed that the enhancement in pursuit performance primarily resulted from changes in pursuit strategy and the optimization of velocity control. In conclusion, this study reveals that experienced mice can optimize their pursuit strategies to improve their preying efficiency, and the transition from novice to veteran can be used to study the biological mechanisms of behavioral flexibility in mice.
... Of late, evolutionary and genetic modelling [24,25], models which leads to the concerted evasive manoeuvres observed in nature. An optimised 49 information transfer network could be emulated in robotic swarms for far-reaching 50 implications in the military and the disaster rescue sectors [30]. 51 Interactions between adversarial species, aptly named predator-prey interactions, are 52 a major talking point in the biophysics community due to the sheer complexity involved. ...
... The reviews by Sih et al. [45], 59 Bailey et al. [46], and more recently Hansen et al. [47] accentuate the benefits of group 60 hunting while emphasising the dearth of comprehension of the underlying physics 61 involved in such events. The attack patterns of predators have been explored in the 62 literature to some extent [10,24,[48][49][50]; however, there is a lack of studies highlighting 63 the effect of predator attacks using a combination of these strategies. 64 The current work focuses on explaining the aftermath of repeated predator attacks 65 on a generic prey flock using an agent-based model. ...
... In our work, the predators follow the simplest 389 "pure pursuit" approach, where the predator agent dynamically chases the moving prey 390 by accelerating towards it. However, there are more complex schemes, such as deviated 391 pursuit and parallel navigation [50], which can also be used to express predator pursuit. 392 Palmer and Packer classify the predators' hunting strategy into ambush and 393 coursing [10]. ...
Collective behaviour is a ubiquitous emergent phenomenon where organisms share information and conduct complicated manoeuvres as a group. Dilution of predation risk is presumed to be a major proponent contributing towards the emergence of such fascinating behaviour. However, the role of multiple sources of predation risk in determining the characteristics of the escape manoeuvres remains largely unexplored. The current work aims to address this paucity by examining the response of a flock to multiple persistently pursuing predators, using an agent-based approach employing a force-based model. Collective features such as herding, avoiding and split-and-join are observed across a wide spectrum of systemic conditions. The transition from one response state to another is examined as a function of the relative angle of predator attack, a parameter exclusive to multi-predator systems. Other concomitant parameters, such as the frequency of attacks and compatibility of target selection tactics of the predators, have a significant effect on the escape probability of the prey (i.e., the success rate of escape manoeuvres). A quantitative analysis has been carried out to determine the most successful combination of target selection while also focusing on beneficial ancillary effects such as flock splitting. The long-term dynamics of the system indicate a faster decay of prey numbers (higher prey mortality) at higher coordination strength due to a monotonically decreasing relation between coordination strength and prey speed supplanted by coincidental synchrony of predator attacks. The work highlights the non-additive nature of the effects of predation in a multi-predator system and urges further scrutiny of group hunting dynamics in such systems.
... A more general situation is when angle φ has been maintained at a nearzero bearing, as shown in Fig. 5B, which is known as "deviated pursuit" (Shneydor 1998). "Parallel pursuit" (Kane et al. 2015) was another pursuit strategy that most ying predators used to plan their path in order to intercept the prey. It requires the predator adopting a new angle φ whenever the prey changes its heading or relative velocity (Fig. 5C). ...
Most animals must reserve their limited intelligence for the most important situations, such as predation and escape, in order to have a better chance of survival. As a highly sequentially programmed behavior driven by innate desire, one of the most challenging parts of predation is how the predator can pursue and capture an escaping prey that is also running for its own survival. This requires the predator to synthesize environmental and prey information to make dynamic decisions in real time to guide appropriate behavior. However, it is still largely unclear whether and how mice can cope with such challenge. Here, we developed a real-time interactive platform to study the pursuit behavior during predation in rodents. An artificial prey was magnetically controlled by a closed-loop system that attempts to escape an approaching predator (e.g., a hungry mouse) in real time. By recording the time costs, trajectories and other parameters of both predator and prey, we found that not only were the mice able to complete predation tasks of varying difficulty, but that they could also improve their predation efficiency over trials, mainly due to the improvements in the pursuit phase. Further investigation revealed that the increase in pursuit performance may not entirely achieved by physical improvement, but rather by optimization of velocity control as well as a change of navigation strategy. In conclusion, this study reveals that mice are capable of making dynamic decisions during predatory pursuit, and the transition from novice to veteran can be used to study the biological mechanisms of dynamic decision making in mice.
... These behaviors are generally thought of as complex phenomena in which two or more agents interact in environments, which change from moment to moment, yet many studies have shown that the rules of behavior (e.g., which direction to move at each time in a given situation) can be described by relatively simple mathematical models consisting of the current state (e.g., positions and velocities) [5,4,22,15,48,12,16,18]. In other words, these behaviors can be modeled by standard reinforcement learning methods for a finite Markov decision process (MDP) in which each sequence is a distinct state. ...
... There is a separate output unit for each possible action, and only the state representation is an input to the neural network. The inputs to the neural network are the positions of oneself and others in the absolute coordinate system (x-and y-positions) and the positions and velocities of oneself and others in the relative coordinate system (u-and v-positions and u-and v-velocities), which are determined based on variables used in the chase and escape models in biology [5,4,22,15,48,12,16,18]. We assumed that delays in sensory processing are compensated for by estimation [47] and the current information at each time was used as input as is. ...
Cooperative hunting has long received considerable attention because it may be an evolutionary origin of cooperation and even our sociality. It has been known that the level of organization of this predation varies among species. Although predator-prey interactions have been studied in multi-agent reinforcement learning domains, there have been few attempts to use the simulations to better understand human and other animal behaviors. In this study, we introduce a predator-prey simulation environment based on multi-agent deep reinforcement learning that can bridge the gap between biological/ecological and artificial intelligence domains. Using this environment, we revealed that organized cooperative hunting patterns with role division among individuals, which is positioned as the highest level of organization in cooperative hunting of animals in nature, can emerge via a simplest form of multi-agent deep reinforcement learning. Our results suggest that sophisticated collaborative patterns, which have often been thought to require high cognition, can be realized from relatively simple cognitive and learning mechanisms and that the close link between the behavioral patterns of agents and animals acquired through interaction with their environments.Keywordsemergencecollaborationmulti-agent systemsdeep reinforcement learningpredator-prey interactions
... Collision avoidance [1][2][3][4] and target pursuit [5][6][7][8] are challenging flight behaviors for any animal or autonomous vehicle, but their interaction is even more so. [9][10][11] For predators adapted to hunting in clutter, the demands of these two tasks will often come into conflict, 12 requiring effective reconciliation to avoid loss of the target or a hazardous collision. Technical approaches to autonomous obstacle avoidance commonly combine active mapping 13 and path planning 14 algorithms, but these approaches are computationally costly 15 and are unlikely to be effective during closed-loop pursuit of a target maneuvering through clutter. ...
Pursuing prey through clutter is a complex and risky activity requiring integration of guidance subsystems for obstacle avoidance and target pursuit. The unobstructed pursuit trajectories of Harris' hawks Parabuteo unicinctus are well modeled by a mixed guidance law feeding back target deviation angle and line-of-sight rate. Here we ask how their pursuit behavior is modified in response to obstacles, using high-speed motion capture to reconstruct flight trajectories recorded during obstructed pursuit of maneuvering targets. We find that Harris' hawks use the same mixed guidance law during obstructed pursuit but appear to superpose a discrete bias command that resets their flight direction to aim at a clearance of approximately one wing length from an upcoming obstacle as they reach some threshold distance from it. Combining a feedback command in response to target motion with a feedforward command in response to upcoming obstacles provides an effective means of prioritizing obstacle avoidance while remaining locked-on to a target. We therefore anticipate that a similar mechanism may be used in terrestrial and aquatic pursuit. The same biased guidance law could also be used for obstacle avoidance in drones designed to intercept other drones in clutter, or to navigate between fixed waypoints in urban environments.
... We composed a guidance control model that would predict the predator's flight during Phase 3 of pursuit (Chase) based on our initial assumptions and prior work in this field [4,6,29,40,41]. Our mixed model considered the combined effects of pure pursuit and proportional navigation components [42], as well as each strategy independently. If a predator's pursuit was governed solely by pure pursuit, its steering command would be proportional to the angle δ: ...
... A gain of zero for either of the models' components was included to test the hypothesis that one strategy alone could explain the predator's steering. The selected range for K PP , K PN and τ were informed by previous work in this field [6,19,29,40,41,43,44], and the acceleration limits were set by our manual inspection of the predators' observed acceleration range in this study. We varied the starting point of the simulation from t = 0 to t = 0.2t end to account for transient effects in the predator's dynamics between Phases 2 (Repositioning) and 3 (Chase). ...
A predator's capacity to catch prey depends on its ability to navigate its environment in response to prey movements or escape behaviour. In predator–prey interactions that involve an active chase, pursuit behaviour can be studied as the collection of rules that dictate how a predator should steer to capture prey. It remains unclear how variable this behaviour is within and across species since most studies have detailed the pursuit behaviour of high-speed, open-area foragers. In this study, we analyse the pursuit behaviour in 44 successful captures by Corynorhinus townsendii, Townsend's big-eared bat (n = 4). This species forages close to vegetation using slow and highly manoeuvrable flight, which contrasts with the locomotor capabilities and feeding ecologies of other taxa studied to date. Our results indicate that this species relies on an initial stealthy approach, which is generally sufficient to capture prey (32 out of 44 trials). In cases where the initial approach is not sufficient to perform a capture attempt (12 out of 44 trials), C. townsendii continues its pursuit by reacting to prey movements in a manner best modelled with a combination of pure pursuit, or following prey directly, and proportional navigation, or moving to an interception point.
... Peregrines and other falcons Falco spp. have been found to intersperse longer periods of stabilized gaze with short saccadic head movements during the midphase of a pursuit, but fixate their target at a consistent location on their retina during its terminal phase [29][30][31]. Target fixation has also been observed in a northern goshawk Accipiter gentilis chasing terrestrial prey [29], but unpublished data from Harris' hawks [31] suggest that they lie at neither extreme of the mechanization continuum. Instead, Harris' hawks allow their target to drift a short distance across the retina, before making a fast saccadic motion to re-fix the retinal coordinates of the target. ...
... have been found to intersperse longer periods of stabilized gaze with short saccadic head movements during the midphase of a pursuit, but fixate their target at a consistent location on their retina during its terminal phase [29][30][31]. Target fixation has also been observed in a northern goshawk Accipiter gentilis chasing terrestrial prey [29], but unpublished data from Harris' hawks [31] suggest that they lie at neither extreme of the mechanization continuum. Instead, Harris' hawks allow their target to drift a short distance across the retina, before making a fast saccadic motion to re-fix the retinal coordinates of the target. ...
The aerial interception behaviour of falcons is well modelled by a guidance law called proportional navigation, which commands steering at a rate proportional to the angular rate of the line-of-sight from predator to prey. Because the line-of-sight rate is defined in an inertial frame of reference, proportional navigation must be implemented using visual-inertial sensor fusion. By contrast, the aerial pursuit behaviour of hawks chasing terrestrial targets is better modelled by a mixed guidance law combining information on the line-of-sight rate with information on the deviation angle between the attacker's velocity and the line-of-sight. Here we ask whether this behaviour may be controlled using visual information alone. We use high-speed motion capture to record n = 228 flights from N = 4 Harris' hawks Parabuteo unicinctus, and show that proportional navigation and mixed guidance both model their trajectories well. The mixed guidance law also models the data closely when visual-inertial information on the line-of-sight rate is replaced by visual information on the motion of the target relative to its background. Although the visual-inertial form of the mixed guidance law provides the closest fit, all three guidance laws provide an adequate phenomenological model of the behavioural data, whilst making different predictions on the physiological pathways involved.
... These stages could conceivably also apply to aerial predators hunting flocks of small birds or insect swarms. The pod (i) swam in a 'wide line' during search, (ii) formed a 'tight line' to drive the prey directionally, before (iii) 'circling' the prey to trap (Ghose et al., 2009;Combes et al., 2012;Kane et al., 2015) to pursue and capture targeted prey (orange oval) or the classical pursuit (CP) strategy (Kalko, 1998;Shubkina et al., 2012)it does not perform both. CP is a pure pursuit strategy where the predator aims at the current position of the prey target. ...
... For example, experimental work with archerfish (Toxotes jaculatrix) showed how the specifics of characteristic archerfish 'shooting' behaviour altered in the presence of predator conspecifics (Jones et al., 2018). Ground-breaking work on individual predator pursuits and prey evasion has explored hunts through the simple rules animals follow in overt reaction to a stimulus (Ghose et al., 2009;Kane, Fulton & Rosenthal, 2015;Hein et al., 2018Hein et al., , 2020Wilson et al., 2015Wilson et al., , 2018Storms et al., 2019;Papadopoulou et al., 2022). This has set the stage for empirical investigations of wild group hunts and in Fig. 6 we provide potential scenarios to be investigated in the future, illustrating how additional predators could alter individual strategies. ...
Group-hunting is ubiquitous across animal taxa and has received considerable attention in the context of its functions. By contrast much less is known about the mechanisms by which grouping predators hunt their prey. This is primarily due to a lack of experimental manipulation alongside logistical difficulties quantifying the behaviour of multiple predators at high spatiotemporal resolution as they search, select, and capture wild prey. However, the use of new remote-sensing technologies and a broadening of the focal taxa beyond apex predators provides researchers with a great opportunity to discern accurately how multiple predators hunt together and not just whether doing so provides hunters with a per capita benefit. We incorporate many ideas from collective behaviour and locomotion throughout this review to make testable predictions for future researchers and pay particular attention to the role that computer simulation can play in a feedback loop with empirical data collection. Our review of the literature showed that the breadth of predator:prey size ratios among the taxa that can be considered to hunt as a group is very large (<100 to >102 ). We therefore synthesised the literature with respect to these predator:prey ratios and found that they promoted different hunting mechanisms. Additionally, these different hunting mechanisms are also related to particular stages of the hunt (search, selection, capture) and thus we structure our review in accordance with these two factors (stage of the hunt and predator:prey size ratio). We identify several novel group-hunting mechanisms which are largely untested, particularly under field conditions, and we also highlight a range of potential study organisms that are amenable to experimental testing of these mechanisms in connection with tracking technology. We believe that a combination of new hypotheses, study systems and methodological approaches should help push the field of group-hunting in new directions.
... have been used to analyse aerial attack behaviours in hawks and falcons (Kane et al., 2015;Kane & Zamani, 2014;Ochs et al., 2016). This approach allows us to investigate a bird's behaviour in its natural habitat, but is subject to the extreme limitations of pixel count, dynamic range and field of view of any camera small enough to mount on the head. ...
... Payload is conventionally limited to ≤ 5% of a bird's body mass on welfare grounds (Fair et al., 2010), but much more stringent limits may be required to ensure natural behaviour if the load is carried on the head (Kane & Zamani, 2014). The 20 g cameras that have been used previously (Kane et al., 2015;Kane & Zamani, 2014) are twice the weight of many small birds, and therefore only suitable for very large species such as raptors. Even so, it is currently not possible for a small camera to cover a bird's full field of view at an appropriate optical or sampling resolution. ...
... Even so, it is currently not possible for a small camera to cover a bird's full field of view at an appropriate optical or sampling resolution. For example, the vertical field of view (31 • ) of the head-mounted camera used to study goshawks and falcons (Kane et al., 2015;Kane & Zamani, 2014) wouldn't cover the vertical extent of the binocular overlap (100 • ) of the birds in this work, namely Harris' hawks (Potier et al., 2016). Furthermore, the possibilities for analysing head-mounted video data are also impacted by the cameras' low frame rates (30 Hz was used in Kane et al., 2015, andZamani, 2014), and the motion blur associated with low shutter speeds and rolling shutters (Kane & Zamani, 2014). ...
Unlabelled:
Birds of prey rely on vision to execute flight manoeuvres that are key to their survival, such as intercepting fast-moving targets or navigating through clutter. A better understanding of the role played by vision during these manoeuvres is not only relevant within the field of animal behaviour, but could also have applications for autonomous drones. In this paper, we present a novel method that uses computer vision tools to analyse the role of active vision in bird flight, and demonstrate its use to answer behavioural questions. Combining motion capture data from Harris' hawks with a hybrid 3D model of the environment, we render RGB images, semantic maps, depth information and optic flow outputs that characterise the visual experience of the bird in flight. In contrast with previous approaches, our method allows us to consider different camera models and alternative gaze strategies for the purposes of hypothesis testing, allows us to consider visual input over the complete visual field of the bird, and is not limited by the technical specifications and performance of a head-mounted camera light enough to attach to a bird's head in flight. We present pilot data from three sample flights: a pursuit flight, in which a hawk intercepts a moving target, and two obstacle avoidance flights. With this approach, we provide a reproducible method that facilitates the collection of large volumes of data across many individuals, opening up new avenues for data-driven models of animal behaviour.
Supplementary information:
The online version contains supplementary material available at 10.1007/s11263-022-01733-2.
