Each panel represents a single obstacle test flight and plots the hawk’s measured flight trajectory (dark blue) in pursuit of the lure (dark red) up to the point of capture (black dot). The measured data are compared to simulations of the hawk’s trajectory (light blue) under the refined mixed guidance law (Equation 1), with best-fitting parameters N=0.75, K=1.25s-1, and tau=0.010s, assuming discrete application of a deviation angle bias targeting a clearance of 0.6 m from the nearest edge of an upcoming obstacle (black cross), applied once at take-off in respect of the first obstacle and once at a distance of 4.0 m from the second obstacle. In cases where the gap between obstacles was <1.2 m, this mid-course steering correction was assumed to target the center of the gap, instead of the usual clearance of 0.6 m from the nearest obstacle. The displayed values of epsilon show the mean prediction error for the corresponding simulation. The dashed line plots the continuation of the simulation without the second steering correction applied to show the effect of its application on obstacle avoidance. Hanging rope obstacles are plotted as gray dots. The lefthand panels show the four longest flights; the righthand panels show the four flights with the lowest mean prediction error relative to the total distance flown, for flights >9 m in length. Grid spacing: 1 m. See also Figure S1.
