96. CONVEX COMBINATION OF SPARSE CONTROL POLICIES IN FAST HUMAN MOVEMENTS

Department: Electrical & Computer Engineering
Faculty Advisor(s): Robert Hecht-Nielsen | Clark C. Guest

Primary Student
Name: Mehrdad Yazdani
Email: myazdani@ucsd.edu
Phone: 858-229-2558
Grad Year: 2012

Abstract
Point-to-point ballistic human movements (i.e. fast movements) are a class of movements characterized by straight paths, bell-shaped velocity profiles, and smoothness. We propose sparse bang-bang optimal control policies that can achieve such movements. These optimal control policies are accomplished by minimizing the infinity-norm of the jerk and higher-order derivatives of end-effector position with known initial position, final position, and duration of movement. These optimal control policies are then combined as an optimal convex combination. We compare the results of such an architecture with human motion data recorded with a manipulandum. We propose that such sparse bang-bang control policies are inherently simple for the central nervous system to implement, and physiological experiments support the possibility that some parts of the CNS use sparse control policies. Furthermore, while many computational neural models of movement control have used a bang-bang control policy without justification, our study shows that the use of such policies is not only convenient, but optimal.

Related Links:

  1. http://www.sciencedirect.com/science/article/pii/S0893608011002978
  2. http://ieng9.ucsd.edu/~myazdani/

« Back to Posters or Search Results