Eric Ronco and Olivier J.-M. D. Coenen A gain scheduling feedback motor control system. Neural Control of Movement Conference Abstracts, 2003

Sony CSL authors: Olivier Coenen, Eric Ronco

Abstract

The motor control system apparently satisfy two sets of data at the antipode of each other. On one hand, patients that suffer from sensory neuropath y and are deprived from proprioceptive feedback through the spinal cord are still able to achieve complex movements using external feedback, such as vision. On the other hand, animals that underwent spinal cord transection maintain the ability to execute complex movements, such as walking on a treadmill. Hence, the first suggests that spinal feedback is not necessary for motor control, whereas the second demonstrates that complex motor programs are precisely located within the spinal cord. We present a control model that attemps to reconcile these two extremes. The model is an adaptive feedback control system where the feedback loop (a spinal stretch reflex model) is driven by the final and desired muscle lengths. These are computed from desired final movements of the limbs and provided by higher (than the spinal cord) brain areas. A learned and adaptive mapping is used to achieve the correspondance between muscle lengths and limb positions relative to the body. Feedback gains in the model are adapted according to the movement context. This ``gain scheduling'' could be achieved by the higher brain areas by modulating the sensitivity of the muscle spindle afferents. Finally, another lookup table is used to store the feedback gains according to the context and type of movements to perform. We suggest that storage and context modulation for the parameters may be a role of the cerebellum. A benchmark reaching movement task is considered to compare results obtained by our model with results obtained by a common motor control model put forward in the literature, an inverse model controller. Simulations are performed with a model of a six-muscle human arm attached to a robot manipulendum. Similar results are obtained with both models and are consistent with human behavioral data. The special case in which the spinal proprioceptive feedback is abolished (no muscle spindle afferent) is also considered. With this structure, simulations suggests that reaching movements are still possible.

Keywords: motor control, spinal cord, model, model

BibTeX entry

@INPROCEEDINGS { ronco:03b, AUTHOR="Eric Ronco and Olivier J.-M. D. Coenen", BOOKTITLE="Neural Control of Movement Conference Abstracts", TITLE="A gain scheduling feedback motor control system", YEAR="2003", }