Perception/Action: An Holistic Approach II

摘要

This final report reviews three years of research focused on the coordination ofperception and action. Human performance has been evaluated within the framework of a closed-loop system where perception and action are intimately coupled. Four problems have been studied: the control of locomotion, dynamic occlusion, depth perception, and minimally invasive surgery. Studies of the control of locomotion have shown that for control of altitude there was an interaction between the flow structure (splay or depression angle) and the event dynamic (hover or forward flight). Results showed that in hover conditions, depression angle specifies altitude changes most reliably; but in forward flight conditions, splay angle specifies altitude changes most reliably. These results are explained as a function of singal-to-noise ratios within the optical flow field. Initial work to evaluate control of collision is also discussed. The dynamic occlusion task was used to evaluate effects of mode of observation (active versus passive) on performance. This task was chosen because of the ability to control for information differences. Results showed that the observation mode had little effect. Previous research that has shown an advantage for active observers appears to be due to information differences, not to the observation mode. Research on depth perception showed that difference that had previously been found for response mode (walking versus matching) could be attributed to the response frame (egocentric versus exocentric). Subjects were generally accurate (little effects due to foreshortening) when using an egocentric response frame. Finally, preliminary work to evaluate performance in minimally invasive surgery is outlined. This work included field observations of surgery as well as laboratory studies.