Aisbq 113

Having designed the control system, the next stage was to test whether these motion camouflage controllers would be able to fool humans! We did this by performing a novel psychophysical experiment that masqueraded as a computer game competition. The competition was based on a 3D computer game (Missile Defence) of the popular first-person shoot-em-up genre: purpose-written to allow comparison of subjects’ success in detecting different approach strategies. In the game, the player takes the role of the prey. The player flies along the centre of a straight tunnel (see Figure 2) whilst attempting to shoot missiles (fired at the player). The missiles represent our predators employing different strategies to approach the prey. Three different approach strategies were investigated. The first was motion camouflage. The second was a homing approach where, at each time step, the missile moved in the direction of the prey. In the third, a direct-interception ap-proach, the missile moved in a straight line to intercept the prey as quickly as possible: these missiles had access to the path of the prey in advance. The results of the experiment, conducted on 30 volunteers, showed that motion-camouflaged missiles were, in general, able to get closer to the player than missiles using the other strategies: they approached to an average distance-to-prey that was just ~60% of the missiles using the homing strategy and ~50% of those attempting direct interception. This psychophysical experiment therefore served two purposes: to further validate the design of the control system using a real task; and to provide the first evidence suggesting that humans are susceptible to motion camouflage. The study and modelling of the hoverfly’s stealth behaviour has proved useful in that it has No.113 Quarterly