Loading...

Design and Implementation of an Intention-based Control System for a Lower Limb Exoskeleton by Combining Biological, Kinematic and Interaction Force Signals

Salarkia, Morteza | 2018

626 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 50780 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Vosoughi, Gholamreza
  7. Abstract:
  8. The goal of this project is to propose an algorithm to detect human motion intent during level walking. The primary aim of this algorithm should be to reduce human-robot interaction forces by combining EMG and force signals. Human gait is composed of 6 different stages. During each stage, each muscle is activated differently so it is possible to detect human intent by a pattern recognition technique. However, due to the fact that EMG signals are unrepeatable and noisy, we propose that combining EMG, motion and interaction force signals may lead to more concise and repeatable estimates of human motion intent. As a result, this control approach is expected to reduce human robot interaction force to the minimum amount. In this project, the aim is to use neuro-fuzzy algorithms to combine different sensor data to generate control signals for an exoskeleton robot. The control loop consists of an outer torque loop with inner admittance based loop. To this end, the effect of each sensory data like EMG, encoders and force sensors are considered and the best combination of input data on the basis of cost, repeatability, the ability to reduce interaction forces are identified
  9. Keywords:
  10. Neural Networks ; Exoskeleton ; Electromyogram Signal ; Data Aggregation ; Diagnostic Control System ; Data Reconciliation ; Lower Limb Exoskeleton

 Digital Object List

 Bookmark

No TOC