Many flying animals fly through cluttered environments, requiring sophisticated aerial manoeuvres. The rates at which optical information of obstacles is converted into control outputs are impressive and inspire the development of advanced obstacle avoidance systems for unmanned aerial vehicles. Birds of prey, such as hawks, are known for their capabilities to fly through narrow gaps. As an initial step to find out how birds use optical information to control their flight, we performed an experiment where birds had to fly through a narrow gap, at varying distances along the trajectory. A bird could judge the distance before take-off and adjust every wingbeat so that it passes the gap exactly when it has the wings folded. Alternatively, it could fly up to the gap and only then make an adjustment to avoid collision.
We trained four Harris's hawks (Parabuteo unicinctus) to fly between two perches 12m distance apart and raised 1.5m above the ground in a room of 20m length, 6m wide and >3.5m height. Vertical posts of 2m height (0.08m diameter) were mounted on 1.5m long motorized tracks, to form an obstacle along the flightpath. The gap was 33cm wide, which is about a third of the wingspan. The position of the gap was varied randomly between each flight, ±0.75m along the flight direction.
The room was equiped with a motion capture system, with 20 cameras covering the volume between the perches. The birds carried a backpack with reflective markers, from which we could reconstruct the flight trajectories and body accelerations.
From a preliminary analysis of a small subset of the data it appears that the birds do not adjust the timing of the wingbeats early in the trajectory. The figure shows data for 17 flights of one bird: the location of peak lift for the three last wingbeats for different gap locations (take-off at -6m and landing at 6m). Not surprisingly, the last wingbeat before the glide phase, is strongly affected by the gap, occasionally postponed to after passing through the gap.
More data needs to be processed, but this analysis suggests the birds only adjust their flight pattern at the last moment resulting in a delayed last wingbeat.
Further experiments will provide information on how and when birds judge the distance to the gap.
We are deeply indebted to Mark Parker and Lucy Larkman for the training and handling of the birds for these experiments. Also the assistance of Caroline Brighton and Lydia France has been invaluable.