QUICK SEARCH:   [advanced] Author: Keyword(s):
Home     Help     Feedback     Subscriptions     Archive     Search     Table of Contents

Help viewing high resolution images Return to article

Fig. 1. Signal processing performed by the virtual fly. We implemented two visual pathways in the virtual fly: one for target fixation (A) and one for speed control (B). A further module that receives input from both pathways determines the virtual fly's actual position in the next simulation step (C). In each simulation step the fixation controller, converts the error angle according to the characteristic curve shown in C, weighted by G_p, and the retinal velocity, weighted by G_v, into angular velocity of the pursuing virtual fly (t_n+1). The output of the virtual fly's speed controller depends on retinal target size according to the characteristic curve shown in the box and determines the absolute value of the fly's speed vector for the next simulation step [s(t_n+1)]. First-order low-pass temporal filters are applied to the outputs of both visual pathways, lumping together inertial effects, neuronal processing and muscular reaction time. The filtered outputs from each pathway form the `intended' vector of locomotion of the virtual fly. A third module determines the virtual fly's velocity in the next simulation step as the sum of the actual fly velocity and the `intended' velocity vector weighted by the movement coefficient M. Six of the free model parameters were taken from our preceding study (Boeddeker and Egelhaaf 2003): the two first-order low-pass filter time constants acting on fixation (_f=15 ms) and speed control (_v=80 ms), the movement coefficient (M=0.0455), and three parameters characterising the transfer function of the speed controller (S_g=0.8 m s^–1, S_v=67, and *= 0.0865). The gain factor for yaw rotation, depending on retinal target position (G_p), was set to 0.1 and the gain factor for yaw rotation, depending on retinal target velocity (G_v).

Return to article