VISUAL COMPUTATIONS FOR MOTOR CONTROL

电机控制的视觉计算

• 批准号: 6259411
• 负责人: DAVID C KNILL
• 金额: $ 28.56万
• 依托单位: UNIVERSITY OF ROCHESTER
• 依托单位国家: 美国
• 财政年份: 2001
• 资助国家: 美国
• 起止时间: 2001-01-01 至 2003-12-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6259411
• 关键词: behavior test; biomechanics; cues; goal oriented behavior; human subject; interactive multimedia; judgment; learning transfer; limb movement; orientation; psychomotor function; robotics; sensory mechanism; surface property; vision; visual depth perception; visual feedback; visual pathways; visual perception

DESCRIPTION (Adapted from applicant's abstract): The goal of the proposed research is to understand how the Visio-motor system integrates different cues for three-dimensional object geometry to guide motion of the hand during goal directed movements; particularly, pointing, prehension and object placement. The proposed experiments will investigate how the visual system integrates binocular and monocular cues to surface orientation and depth for both the planning and on-line control of hand movements. We will use the results of these experiments to compare the visual strategies used to control different motor behaviors, to control different motor components of the same behavior, and for different phases of motor control (planning vs. on-line control). The results will inform the question of which motor control functions rely on common visual computations and which rely on task-specific computations. We will also compare the visual strategies used to combine cues for motor control with those used to make perceptual judgments, as a west of the hypothesis that independent visual mechanisms subserve these broadly defined task domains. We will further investigate how feedback obtained during performance of goal-directed movements adapts the visual computations used to derive object information for guiding hand movements. Specifically, we will artificially adjust the correlations between different 3D cues and physical target object geometry (e.g. surface orientation) and measure induced changes in the cue weights used to estimate object geometry for motor control. We will measure transfer of learning between different visuo-motor tasks and between visuo-motor tasks and perceptual tasks as a further test of the task-specificity of visual computations. Experiments will be performed in a laboratory comprised of a virtual display system for display of visual target information, a robot arm for placement of physical targets for subjects' reaches and a real-time 3D motion tracking system for measuring hand movement kinematics. Novel computational analyses are described for quantifying the relative contributions of visual cues to motor control, both for motor planning and for on-line control of the hand during a movement. The research will elucidate the functional relationships between visual processes subserving different functions, both at the broad level of perceptual judgment vs. visuo-motor behavior and at a more refined level of one motor task vs. another and motor planning vs. on-line motor control.

描述（改编自申请人的摘要）：拟议的目标 研究的目的是了解视觉运动系统如何整合不同的线索 用于在进球期间引导手的运动的三维物体几何形状 定向运动;特别是指向、移动和对象放置。 这些实验将研究视觉系统如何整合 双眼和单眼线索的表面方向和深度， 规划和在线控制手部动作。我们将使用 这些实验来比较用于控制不同的视觉策略， 运动行为，控制同一行为的不同运动组件， 以及用于电机控制的不同阶段（规划与在线控制）。的 结果将告知运动控制功能依赖于什么的问题 常见的视觉计算并且依赖于特定于任务的计算。我们 我还将比较用于联合收割机线索运动控制的视觉策略 与那些用来做出感性判断的人，作为一个西方的假设， 独立的视觉机制辅助这些广泛定义的任务域。我们 将进一步研究如何在执行过程中获得反馈， 目标导向的运动适应用于导出对象的视觉计算 指导手部动作的信息。具体来说，我们将人工 调整不同3D提示和物理目标对象之间的相关性 几何形状（例如表面方向）和测量线索中的诱导变化 用于估计用于运动控制的对象几何形状的权重。我们将测量 不同视觉运动任务之间的学习转移， 视觉运动任务和知觉任务作为进一步的测试， 视觉计算的任务特异性实验将在 包括用于显示可视目标虚拟显示系统的实验室 信息，机器人手臂，用于放置对象的物理目标， 和用于测量手部运动的实时3D运动跟踪系统 运动学新的计算分析描述了量化的 视觉线索对运动控制的相对贡献，无论是运动规划 并且用于在运动期间在线控制手。这项研究将 阐明视觉过程之间的功能关系， 不同的功能，无论是在广泛的水平上的知觉判断与。 视觉运动行为，并在一个运动任务与另一个更精细的水平 以及运动规划与在线运动控制。