To Map 3 DOF Waist Power Assistive Robot Motion Using Neural Network

Ali Othman, Syafiq (2015) To Map 3 DOF Waist Power Assistive Robot Motion Using Neural Network. Project Report. UTeM, Melaka, Malaysia. (Submitted)

Text To Map 3 DOF Waist Power Assistive Robot Motion Using Neural Network 24 Pages.pdf - Submitted Version Download (293kB)

Abstract

Wearable robot such as an exoskeleton can increase the performance of a user. It can be used to enhance the strength or improve the user endurance through various means. There are numerous type of exoskeleton with various design and waist assistive robot is one of it. Waist power assistive exoskeleton can be used to improve muscle endurance of the user by providing support torque at the back. To control exoskeleton, desired characteristic to sense physical motion of human is needed. There are several studies that use Electromyography (EMG) sensor to detect human intention. However, it is complicated and expensive. Thus, by using force sensor resistance, the measure of external muscle pressure is simple, cheaper, and has reliable way of sensing and giving feedback of muscle. The mapping method was proposed to map the exoskeleton according to the user’s intention. With the feature extraction and classification using Back-propagation neural network (BPN), the Force Sensor Resistrance (FSR)-Angle model was constructed to be used for pattern recognition. One healthy subject performed forward bending, forward bending to the right and to the left. The FSR signals of the forces from the front and the back of the subject body were collected. The mapping scheme reliability were evaluated in the experiments. The results indicated that mapping using BPN with 2 hidden layer shows better performance compared to 3 hidden layer BPN.

Actions (login required)

View Item