Walking Rehabilitation

Abstract
In this project we examine a method to control the stepping motion of a paralyzed person suspended on a treadmill using a robot attached to the torso and hips. A leg swing motion is created by moving the hips without contact with the legs. The problem is formulated as an optimal control problem for an underactuated articulated chain. The optimal control problem is converted into a discrete parameter optimization and an efficient gradient-based algorithm is used to solve it. Motion capture data from a human subject is compared to the results from the dynamic motion optimization. Our results indicate that it is possible for the robot to create a gait for the paralyzed person that is close to that of an unimpaired subject.

Body Weight Supported (BWS) Training (The picture shows a partial weight bearing gait therapy device LITEGAIT I by Mobility Research, Inc. )

One Step: one-half of the gait cycle (The .mpg files can be played by Windows Media Player.)

• Real motion obtained from the video-based motion-capture system:
  • Case 1 - 0.433 sec: (Side View) , (Front View) , and (Top View)
  • Case 2 - 0.467 sec: (Side View) , (Front View) , and (Top View)
  • Case 3 - 0.500 sec: (Side View) , (Front View) , and (Top View)
  • Case 4 - 0.567 sec: (Side View) , (Front View) , and (Top View)

  

• Under-actuated model w/motion captured pelvis movement:
  • Case 1 - 0.433 sec: (Side View) , (Front View) , and (Top View)
  • Case 2 - 0.467 sec: (Side View) , (Front View) , and (Top View)
  • Case 3 - 0.500 sec: (Side View) , (Front View) , and (Top View)
  • Case 4 - 0.567 sec: (Side View) , (Front View) , and (Top View)

  

• Fully actuated model (optimal result):
  • Case 1 - 0.433 sec: (Side View) , (Front View) , and (Top View)
  • Case 2 - 0.467 sec: (Side View) , (Front View) , and (Top View)
  • Case 3 - 0.500 sec: (Side View) , (Front View) , and (Top View)
  • Case 4 - 0.567sec: (Side View) , (Front View) , and (Top View)

  

• Under-actuated model (optimal result):
  • Case 1 - 0.433 sec: (Side View) , (Front View) , and (Top View)
  • Case 2 - 0.467 sec: (Side View) , (Front View) , and (Top View)
  • Case 3 - 0.500 sec: (Side View) , (Front View) , and (Top View)
  • Case 4 - 0.567 sec: (Side View) , (Front View) , and (Top View)

  

• Under-actuated model (optimal result w/bounded stance hip internal/external rotation):
  • Case 1 - 0.433 sec: (Side View) , (Front View) , and (Top View)
  • Case 2 - 0.467 sec: (Side View) , (Front View) , and (Top View)
  • Case 3 - 0.500 sec: (Side View) , (Front View) , and (Top View)
  • Case 4 - 0.567 sec: (Side View) , (Front View) , and (Top View)

  

• Zero initial velocities (0.500 sec)
  • Under-actuated model w/motion captured pelvis movement: (Side View) , (Front View) , and (Top View)
  • Under-actuated model (optimal result): (Side View) , (Front View) , and (Top View)
  • Under-actuated model (optimal result w/bounded stance hip internal/external rotation): (Side View) , (Front View) , and (Top View)

References:
''Swinging From The Hip: Use of Dynamic Motion Optimization in the Design of Robotic Gait Rehabilitation,'' ( .ps file ) C,-Y. E. Wang, J. E. Bobrow, and D. J. Reinkensmeyer, IEEE International Conference on Robotics and Automation, Korea, May 2001.