Force-feedback devices are used in Virtual Reality to produce a tactile
sensation on the user immersed in a virtual environment. Such devices are
actually strong mechanical systems able to apply forces onto the user's body
such that his/her positions and postures are constrained to the objects of
the environment. A person in a virtual environment of an airplane cockpit,
for example, can have his/her real arms motion blocked by instruments and
walls in certain positions.
A new force-feedback dataglove (pictures above) is now available at the
VRlab. This glove is not able to prevent someone to make his/her arms pass
through a wall, but in a situation of grasping an object, for example, one
can have the motion of his/her hand limited by the object geometry.
Our lab traditionally works in virtual human bodies representation
and motion. Now we are working in deformable models for biological tissues
(muscles, cartilage, etc.) based in real material properties and biomechanical
parameters. However, such properties and parameters present large variations
from one person to another, and consequently their importance to the final
behavior of the tissue is not well defined in the biomechanics literature.
The use of force-feedback applied on virtual samples of bio-tissues can
help us to configure our deformable models to better fit the real behavior
of each different tissue.
The problem
Our biomechanical model will deform according to the external forces acting
on it (collision with other objects, gravity, etc.) and the forces produced
internally in the tissue as a consequence of the response (reaction) to the
external ones.
Thus, the problem to be studied by the student is linking inputs and outputs
from the model to the glove and vice-versa.
In this work, the samples of tissue will be simple geometric primitives,
like cubes and spheres, and the test situations will be firstly limited to
relatively simple situations: touch with one finger (the sample would be on
a planar surface), or pressing between two fingers.
Student’s work
The student is asked to implement a set of functions to send forces
from the glove to the deformable model, and compute and send to the glove
the response forces from the model eventually produced due to collision
between the hand and the deformable object (collisions will be detected using
an existent library). In other words, the forces produced by the user hand
should be passed to the model that will, in turn, deform its geometry and
produce reaction forces that should be correctly interpreted by the glove
in the form of forces to be applied onto the user hand.
The first task would be calibration of the glove in order to respect the
dimensions, position and scale of the virtual tissue model in relation to
the real glove and user. In this task, a dinamometer will probably used to
validate the applied and produced forces.
The student should also be involved in setting the appropriate user-interface
for the sample size and type, and experiment different settings for the tuning
of the suitable deformation parameters.
Some resulting demo should be:
- the interactive identification of such parameters for a given tissue (we
suggest mashmallow, styrofoam, or some kind of "mousse" for
packaging).
- then using the deformation model alone, try to predict some deformation
for different configurations of applied force(s)
- then compare the result of the simulation with what really happens when
applying this pattern of forces (interactively with the glove).
The outcome should be to validate/invalidate our deformation model (in a
static context).
