|
The use of robotic manipulators to augment upper extremity function has been limited by the
availability of a means to control multiple degrees of freedom simultaneously to generate smooth
and efficient movements. This development project employs multi-degree of freedom
human-computer interfaces in an effort to convey a child’s input to the motion of a manipulator.
Design and Development
Implementation of four different controllers:
Stylus
- Six degrees of freedom
- Revolute joints, same configuration as iARM
manipulator
- Input: one-to-one angles matching. Does not
require inverse kinematics
- Toggle button for grasping
- Forward kinematics used in the collision
detection algorithm for interaction with objects
Spaceball
- Six degrees of freedom
- Input: Isometric forces and torques that are
interpreted as end-effector linear and angular
velocities.
- Inverse kinematics: Inverse Jacobian to
determine joint angles.
- Manipulability of the end-effector to handle
singularities
- Forward kinematics used in the collision
detection algorithm for interaction with objects
- Toggle buttons for grasping and stopping the
simulation
Joystick
- Three degrees of freedom plus toggle button
to switch between translation and rotation (i.e.,
position and orientation)
- Input: Normalized angle value that is
interpreted as the value of displacement or
rotation.
- Inverse kinematics for a 6 dof revolute joint
manipulator
- Forward kinematics used in the collision
detection algorithm for interaction with objects
- Toggle button for grasping and stopping
simulation
- Algorithm for singularities still pending.
Flock of Birds
- Magnetic-field tracking position device
- Six degrees of freedom
- Input: Position and angles of a magnetic
sensor related to a reference frame determined
by the position of a transmitter.
- Inverse kinematics for a 6 dof revolute joint
manipulator.
- No collision detection algorithm implemented
Current Work
- Extend reaching and grasping abilities to accomplish a large number of tasks in an
unstructured world by making use of a variety of control algorithms including direct multi-joint
position control, resolved Cartesian rate control via force input, and hybrid Cartesian/joint
methods for isometric force input ball (Spacemouse 5000®) and position sensing six degree of
freedom stylus (Immersion Probe®) controllers. January 2007.
- Implement two alternative controllers for evaluation, Flock of Birds (Ascension Tech) and three
degree of freedom joystick. January 2008
- Create training environment and evaluate controllers using box of blocks function test and
Fitts' law paradigm test. January 2008.
Experimental design
- Shutter glasses are used to create the 3D image,
Virtual Reality Modeling Language (VRML), position
sensing six degree of freedom stylus, isometric force
input ball, and Matlab are the main technological
components of the three dimensional virtual training
system.
- Box of blocks function test and Fitts' law paradigm
test
- The experiments were conducted on 6 subjects, 4
female, two male. Two-day experiment.
- The first day consisted of all of the real life
experiments, and two fifteen minute training session
in virtual reality, one session per user interface (Stylus
and Spaceball). The second session consisted of
exact duplicates of the real life tests in virtual reality.
- Data collected from position tracking devices and
the programs controlling the visual representations
Results
- Fitts’ law shows that the movement time is
dependant on ID.
- The most significant source of variation is the
input device
- Stylus presented a more realistic approach to the
normal human movements whereas Spaceball was
characterized by continuous overshooting in
reaching the targets, mainly due to the non
familiarity of the subject with the behavior of the
device.
- Box of blocks displayed a better behavior of the
subject when using the Spaceball than Fitts’ Law
test, in terms of time and accuracy; this could be
attributed to the fact that in a more realistic
environment the awkwardness of the movements
are less relevant.
|