[robotics-worldwide] PhD position on "A predictive model of pathologic walking gaits"

Pierre-Brice Wieber Pierre-Brice.Wieber at inrialpes.fr
Tue Mar 20 08:33:23 PDT 2007

** Title

A predictive model of pathologic walking gaits

** Environnement

This position is offered at the “Rhone-Alpes” Research Unit of INRIA,  
near Grenoble and Lyon, in France. The Unit counts more than 500 people
within 25 research teams.

The starting date will be between september 2007 and december 15, 2007.

Duration: 3-years position.
Salary and benefits : 1529€ NET / month + full health insurance and
social benefits included - salary will be upgraded to 1611 € NET /month
the 3rd year.

INRIA Project Team: BIPOP (http://bipop.inrialpes.fr)

Advisor: Pierre-Brice Wieber

Collaborations will be undertaken with the Interdisciplinary Center
of Scientific Computing of the University of Heidelberg and the
Laboratoire de Mecanique des Solides of the University of Poitiers

THE LINKS HERE: http://bipop.inrialpes.fr/join_us.html

** Subject

Numerical models of human walking as well as numerical methods of
optimal control are now reaching enough maturity [1, 2, 3, 4] for
considering seriously the development of fully functional predictive
models of normal and pathologic walking gaits that could become in
the near future powerful tools in therapies of gait disorders [4].
The purpose of such a model would be to help predicting the outcome
of a given medical act on the gait disorder of a person and therefore
help the medical staff deciding which medical act should be
preferred. A driving problem for this thesis could be for example the
surgical treatment of spasticity disorders in young children. Such a
treatment regularly consists in modifying the kinematics of the
musculoskeletal system by applying surgery to the muscles, tendons
and bones of the person so as to relieve the disorders induced by the
spasticity. The predictive model we're looking for would help
deciding which surgery to apply and how.

The first step in developing such a predictive model would be to
decide of the underlying musculoskeletal model to consider, and
especially of the muscle contraction model [5] that could cope
properly with spasticity disorders. This physiological model will
form the basis of the predictive model itself and will need to be
balanced therefore between completeness and complexity. The second
and major step is to define the functional model of walking, that is
to effectively predict how a given physiological model will walk. The
building block here will most probably be optimal control [2]: we
suppose that walking amounts in one way or another to minimizing some
measure of discomfort and fatigue. The question then is to find which
optimal control problem can adequately predict the walking gaits
which are observed on a given set of patients.

There will be a major need here for a general identification
procedure to tune the model to each patient: knowing how a person
walks, with measures coming for example from traditional motion
capture methods and electromyography, tuning the different lengths
and coefficients of the physiological model, but most of all tuning
the different parameters of the functional model (the optimal control
problem) so as to obtain a "prediction", a numerical "explanation" of
why this person walks the way he or she walks. This problem of
parameter identification will mostly induce a problem of inverse
optimal control: find which optimal control problem gives as a
solution the gait that we can observe for a given person. This is
both a theoretical and a numerical problem.

Once this tuned predictive model is obtained, the final goal will be
to apply it to the medical problem presented earlier, that is to
check the outcome of different surgical acts on this numerical model
of the person's gait, propose the results of these simulations to the
medical staff and validate their appropriateness.

Requirements: parameter identification, optimal control, numerical

[1] P.-B. Wieber, Modelisation et Commande d'un Robot Marcheur
Antrhopomorphe, PhD thesis from the Ecole des Mines de Paris, 2000
[2] K. Mombaur, Stability Optimization of Open-loop Controlled
Walking Robots, PhD thesis from the University of Heidelberg, 2001
[3] The HuMAnS toolbox (Humanoid Motion Analysis and Simulation),
[4] D. Popovic et al., Optimal Control of Walking with Functional
Electrical Stimulation: A Computer Simulation Study, IEEE
Transactions on Rehabilitation Engineering, march 1999
[5] H. El Makssoud et al., Modelling of the Skeletal Muscle under
Functional Electrical Stimulation, International Conference of the
Functional Electrical Stimulation Society, 2003

More information about the robotics-worldwide mailing list