Purchase this article with an account.
Michael K. McBeath, Thomas G. Sugar, Dennis M. Shaffer; Comparison of active versus passive ball catching control algorithms using robotic simulations. Journal of Vision 2001;1(3):193. doi: https://doi.org/10.1167/1.3.193.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Past work on how baseball players determine where to run to catch fly balls has identified several visual control variables, (1) maintenance of optical acceleration cancellation (OAC) for the ball, and (2) maintenance of a linear optical trajectory (LOT) for the ball. We use a robotic simulation as an instantiation proof of these control variables, and we test aspects of how well a system can perform when limited to the use of these parameters. The first issue that we examined is how well the system performs using a passive versus an active optical strategy. For the passive strategy, the robotic simulation moved so as to maintain constant optical ball movement upon its fixed “retinal” CCD image plane. For the active strategy, it moved so as to maintain constant camera movement while fixated on the ball. The findings demonstrated superior performance for the active strategy. The second issue that we examined is a reactive versus proactive optical strategy. For the reactive strategy, the robotic simulation waited for a control variable to deviate before it initiated changes in movement. For the proactive strategy, it tried to anticipate and produce control variable deviations for which it could then compensate. For example, if a ball is headed to the right, the robotic simulation can initiate movement to the right and continue in that direction until it produces optical curvature rather than waiting for the optical curvature before initiating movement. The results support a proactive strategy. The findings of both studies are consistent with behavior exhibited by human fielders in which (1) they do not maintain constant movement of the image of the ball on the retina, and (2) they typically begin running in roughly the correct direction before the control variables demand that they do so. Our use of a robotic testing platform provides an exciting new way to instantiate and test proposed human control algorithms in a setting with real world constraints.
This PDF is available to Subscribers Only