| There are many sources of delay in movement control systems, whether human or robotic, which hamper efforts to maintain stability and good performance. Sensory, planning, and control stages of visual-motor processing can be computationally intensive and time consuming. Transmission delays are significant both in the human nervous system and in remote control of a telerobot in space. This thesis describes research to understand how the human neuromotor control system achieves its performance, and how some of these principles can be applied to telerobotic control of a remote manipulator.; The phenomenon of discrete control of human movement may be indicative of the need for feedforward movement processing to maintain stability, and the need to avoid having to compute complex inverse muscle and limb dynamics. Understanding the timing characteristics of discrete movements is essential to understanding how the human neuromotor system maintains stability and performance.; Discrete movement responses are elicited in two sets of experiments, using double-step and quasi-random targets. From double-step experiments one can measure the ability to change a movement plan according to new visual input. It is shown that there is a limited ability to change a plan if new input arrives early enough in processing.; Subjects use predominantly discrete movements, identifiable as peaks in speed traces, when tracking unpredictable quasi-random targets. Surprisingly, these movements are as short as 100 ms. The question of whether these movements are created by independent movement processing events is discussed. The response delay, as measured by responses to sudden directional changes in ramp targets embedded in the quasi-random targets and by cross-correlation calculation, is consistently about 230 ms. The significance of these measurements to the understanding of movement timing is discussed.; A supervisory telerobotic control system was designed and implemented which applies some of the principles of human motor control, including hierarchical design and model-based vision processing. The system uses vision feedback in real-time control of a locomoting robot. Feedforward of the robot's expected position improves the robustness of the vision system's processing.; Finally, an unusual application of human visual-motor control ability, endoscopic surgery, is discussed. Experiments show that limitations of both visual display, including the lack of a stereo endoscope image, and endoscopic instruments, primarily due to their reduced kinematic degrees of freedom, significantly hamper surgeons' performance. Related experiments, analysis, and observations are integrated to understand the fundamental problems of performance in endoscopy and some possible solutions.; Themes of human and robotic visual-manual control will be tied together by comparisons throughout the thesis. |
