Graduate Exam Abstract
Biyun Xie
Ph.D. Preliminary
December 3, 2018, 3:00 pm - 5:00 pm
Mechanical Engineering Conference Room
Fault tolerant design and motion planning of redundant robots
Abstract: A family of optimally fault
tolerant robots for equal joint
failure probabilities has
previously been designed to
optimize a measure of local
fault tolerance. In this work,
the structure and global pre-
and post-failure dexterity
performance of this family of
robots are further explored.
First, the characteristics of
the kinematic properties,
described by Denavit and
Hartenberg parameters, of
these robots are analyzed,
and used to illustrate the
structural correlations
between these robots. Then,
the global pre- and post-
failure dexterity performance
of these robots are studied
and the optimal robot
designs are obtained.
Furthermore, the methods to
design optimally fault tolerant
robots for different joint
failure probabilities are
developed. First, the
equations for the null space
and the canonical form of a
Jacobian that optimizes the
fault tolerance measure for
arbitrary joint failure
probabilities were
determined. Then, classes of
optimally fault tolerant robots
for an arbitrary set of joint
failure probabilities were
designed. Last, a method to
design optimally fault tolerant
robots for cases where the
joint failure probabilities
change was proposed. After
the optimally fault tolerant
robots are designed, a
motion planning algorithm to
maximize the task completion
probability for a set of point-
to-point tasks is presented.
This proposed approach first
develops a method to
calculate the probability of
task failure for an arbitrary
trajectory, where the
trajectory is divided into small
segments, and the probability
of task failure of each
segment is calculated based
on its failure scenarios. Then,
the A* algorithm was applied
to find the optimal trajectory
with minimum probability of
task failure. A novel
component of the proposed
approach was the method
used to compute the
heuristic function that
guarantees a lower-bound on
the remaining path cost.
Adviser: Anthony A. Maciejewski
Co-Adviser: N/A
Non-ECE Member: Jianguo Zhao, Mechanical Engineering Department
Member 3: Peter M. Young, Electrical and Computer Engineering
Addional Members: Ali Pezeshki, Electrical and Computer Engineering
Publications:
[1] B. Xie and A. A. Maciejewski, "Kinematic design of optimally fault tolerant robots for different joint failure probabilities," IEEE Robotics and Automation Letters, vol. 3, no. 2, pp. 827-834, 2018.
[2] B. Xie and A. A. Maciejewski, "Structure and performance analysis of the 7! robots generated from an optimally fault tolerant Jacobian," IEEE Robotics and Automation Letters, vol. 2, no. 4, pp. 1956-1963, 2017.
Program of Study:
ECE555
ECE666
ECE611
MATH560
MATH561
N/A
N/A
N/A
