Neural Models of Spatial Orientation in Novel Environments
Annual rept. 1 Sep 1992-31 Aug 1993
BOSTON UNIV MA CENTER FOR ADAPTIVE SYSTEMS
Pagination or Media Count:
Completed work on this grant analyzes problems of 3-D vision, visual search, spatial mapping, planning and action, all of which utilize spatial- representations to control adaptive behaviors in real time. Highlights include significant contributions towards solving the classical figure-ground problem for biological vision, the motor equivalence problem for flexible arm movement control including tool use, the problem of self-organizing body-centered spatial representations for movement planning and spatial orientation, and the problem of carrying out fast visual search for targets among multiple distractors. New research directions include projects which have been developed to frontally attack core problems concerning how a rapidly moving agent can self-organize spatial representations, use these representations for real-time movement planning, and transform spatial movement plans into appropriate motor commands for movement control and real-time navigation. Specific projects include retinal image processing, Formation of egocentric maps of object positions from optic flow, detection of moving objects from optic flow, integration of egocentric and allocentric representations for autonomous navigation investigation of spatial reference frames and transformations between frames for real-time flexible speech articulator control.