Collision-avoidance and landing responses are mediated by separate pathways in the fruit fly,Drosophila melanogaster

Abstract

SUMMARYFlies rely heavily on visual feedback for several aspects of flight control. As a fly approaches an object, the image projected across its retina expands, providing the fly with visual feedback that can be used either to trigger a collision-avoidance maneuver or a landing response. To determine how a fly makes the decision to land on or avoid a looming object, we measured the behaviors generated in response to an expanding image during tethered flight in a visual closed-loop flight arena. During these experiments, each fly varied its wing-stroke kinematics to actively control the azimuth position of a 15°×15° square within its visual field. Periodically, the square symmetrically expanded in both the horizontal and vertical directions. We measured changes in the fly's wing-stroke amplitude and frequency in response to the expanding square while optically tracking the position of its legs to monitor stereotyped landing responses. Although this stimulus could elicit both the landing responses and collision-avoidance reactions, separate pathways appear to mediate the two behaviors. For example, if the square is in the lateral portion of the fly's field of view at the onset of expansion, the fly increases stroke amplitude in one wing while decreasing amplitude in the other, indicative of a collision-avoidance maneuver. In contrast, frontal expansion elicits an increase in wing-beat frequency and leg extension,indicative of a landing response. To further characterize the sensitivity of these responses to expansion rate, we tested a range of expansion velocities from 100 to 10 000° s-1. Differences in the latency of both the collision-avoidance reactions and the landing responses with expansion rate supported the hypothesis that the two behaviors are mediated by separate pathways. To examine the effects of visual feedback on the magnitude and time course of the two behaviors, we presented the stimulus under open-loop conditions, such that the fly's response did not alter the position of the expanding square. From our results we suggest a model that takes into account the spatial sensitivities and temporal latencies of the collision-avoidance and landing responses, and is sufficient to schematically represent how the fly uses integration of motion information in deciding whether to turn or land when confronted with an expanding object.

Towards Computational Models and Applications of Insect Visual Systems for Motion Perception: A Review

“…Biologists have also explored similar looming sensitive visual neurons in other animals; these include fruit flies (drosophila) [42,45,215,138] and arthropods like crabs [131]. For instance, the lobula giant neurons (LGNs) in crabs have been identified as looming detectors that are located in the lobula layer and correspond to reactive collision avoidance behaviours [149,212,18].…”

“…Most Importantly, such a visual strategy can be used to conduct various forms of insect behaviours such as landing, e.g. [215,13,184] and terrain following, e.g. [193,184,141,53] and tunnel crossing or travelling, e.g.…”

Monocular distance estimation with optical flow maneuvers and efference copies: a stability-based strategy

“…A second difference is that the stability-based strategy explicitly allows for the observation that a visual stimulus itself can trigger the landing responses -even for tethered insects [4,40]. The reason for this is that the strategy depends on the control stability and hence the system bandwidth.…”

Avoiding obstacles while intercepting a moving target: a miniature fly's solution

“…Although in this paper, expansion is discussed with response to a single distinct object, it is not necessarily treated as such in the animal’s neural pathways, as this would require continuous feature correspondence over time ( Fermüller and Aloimonos, 1993 ). Instead, obstacle avoidance could be generated by more standard optomotor pathways, such as those already described in flying insects navigating around their environment ( Krapp and Hengstenberg, 1996 ; Land, 1999 ; Lee, 1980 ; Pix et al, 2000 ; Schuster et al, 2002 ; Srinivasan et al, 1996 ; Tammero and Dickinson, 2002 ), implemented in modern robotics ( Raviv and Joarder, 2000 ; Serres and Ruffier, 2017 ; Srinivasan et al, 1999 ), and which guide our own navigation of the world ( Warren et al, 2001 ). In our setup, the obstacle is orders of magnitude closer to the animal than other objects in a similar FOV (i.e.…”