Abstract
As people steer a car to intercept a moving target, we have shown that they keep the target in a roughly constant direction with reference to the car's heading direction, consistent with the constant target-heading strategy (Zhao, Straub, & Rothkopf, 2017). This pattern of steering usually results in a curved interception path, which is suboptimal due to its longer traveled distance compared with a straight interception path. In the current study, we examined whether participants can learn a better interception solution. Participants (N=8) steered a car (moving at a fixed speed of 7 m/s) to intercept a moving target in virtual environments. In the learning sessions, they intercepted the target in four target conditions, two target movement directions (horizontally left/right) by two target speeds (4.5 or 5.5 m/s). They learned in an environment consisting of a textured ground plane, a blue sky with clouds, and surrounding background image, which provided rich visual information about optic flow and allocentric reference frames. After five learning sessions on different days, participants were tested in two test sessions on the same day with 20 target conditions in each session, four target movement directions (horizontally left/right and approaching from the left/right) by five target speeds (4, 4.5, 5, 5.5, or 6 m/s). In the first test session, participants intercepted the target in the same environment as in the learning sessions; in the second test session, they intercepted the target in an environment consisting of only a ground plane of solid green and a grey sky, which provided no visual information about optic flow or allocentric reference frames. The results show that participants learned to intercept the moving target more efficiently by steering a straighter interception path. Moreover, the learned steering pattern can be generalized to the new target motion and the new environment.
Meeting abstract presented at VSS 2018