Modeling the function of the cerebellum in scheduled linear servo control of simple horizontal planar arm movements
Author(s)
Massaquoi, Steven G
DownloadFull printable version (15.5Mb)
Other Contributors
Massachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science.
Advisor
Munther Dahleh.
Terms of use
Metadata
Show full item recordAbstract
A new set of models has been developed that suggests that the intermediate and lateral anterior cerebellum mediate scheduled linear servo control of simple single-joint and two-joint arm movements. These models propose that the problem of stabilization in the presence of delays is handled by transmitting part of the feedback loop information as "wave variables". Wave variables are specialized linear combinations of command and feedback signals that ensure passive, and therefore stable communication despite delays and changes in loads. The signal integration employed in wave-variable teleoperation may be viewed as implementing a rudimentary linear internal model of limb dynamics. The models propose that for any given simple movement, the bio controller is linear and that therefore compensation for nonlinearities in multi joint limb dynamics is incomplete. It is demonstrated that at least in the two-joint case, this gives rise to specific acceleration-dependent trajectory tracking errors that are characteristic of human experimental movement data. The problem of managing the different dynamic demands of different simple movements may be addressed by the scheduling of a set of linear controllers. During movement simulation, the models generate internal signals that closely resemble neurophysiological waveforms recorded in primates. The qualities of the degraded movement control associated with several simulated lesions of the models also appear to correlate well with the various types of ataxia and tremor that occur following natural and experimental damage to the cerebellar system at neuro anatomically corresponding locations. A state-space representation of the single-joint model suggests an ad hoc reduced order model that reproduces a major part of the behavior of the full single-joint model. The reduced order model gives insight into the stability and performance characteristics of the full model, as well as into the contributions of certain cerebellar pathways to normal movement control.
Description
Thesis (Ph.D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1999. Includes bibliographical references (p. 225-240).
Date issued
1999Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer SciencePublisher
Massachusetts Institute of Technology
Keywords
Electrical Engineering and Computer Science.