Quantization of Human Motions and Learning of Accurate
Movements
Human arm movements are smooth and can be described well using
mathematical models maximizing smoothness. A close examination of
slow or accurate movements shows that the motion is, however,
segmented in submotions, suggesting that the motion is generated
in a discrete way. We investigate how the control, using visual
and kinesthetic feedback, deals with this motion quantization and
performs typical tasks.
- Ted
Milner showed evidence for submotions in movements
with accuracy constraints. In his experiment, subjects
had to move a peg into a hole 0.2 m away, for holes of
different diameters. He found that the motion velocity
was modulated with the hole diameter, and the velocity
profile had small oscillations. He then showed that the
movement can be represented well as the superposition of
smooth submovements which may correspond to visual
corrections.
- I developed a mathematical model for the learning of
accurate human arm movements based on the idea that the
movement is the superposition of smooth submovements. The
intrinsic deviation of arm movements is considered, and
visual and kinesthetic feedbacks are integrated in the
motion control. Movements optimal with respect to time
and accuracy are learned by repeating movements a
reasonable number of times. The model is consistent with
the jerky movements performed by babies, and may explain
how the adult motion behavior emerges from the baby
behavior. Comparison with Ted's measurements of adult
movements shows that the kinematics of accurate movements
are well predicted by the model.
- I am examining how smooth movements emerge from movements
with a visible segmentation. A visual illusion is used in
order to perturb the motions, and the hand trajectory and
the EMG of active muscles are observed during adaptation
to this illusion.
Generating movements with discrete submovements provides a
very economic way of planning smooth movements. Our study may
also lead to efficient motion planners for machines and computer
animated figures.
Related publications
- T.E. Milner and M.M. Ijaz (1990) The Effect of
Accuracy Constraints on Three-Dimensional Movement
Kinematics, Neuroscience 35:487-496
- T.E. Milner (1992) A Model for The Generation of
Movements Requiring Endpoint Precision, Neuroscience
49(2):365-374
- E. Burdet and T.E. Milner (1998) Quantization of
Human Motions and Learning of Accurate Movements,
Biological Cybernetics 78(4): 307-318
back to Etienne BURDET Homepage
June 1998, by E. Burdet