Project

# Title Team Members TA Documents Sponsor
19 FOAM PRESSURE-SENSOR BASED CONTROL METHOD FOR CONTROLLING PROSTHETIC HANDS
Honorable Mention
Jack Li
Yangge Li
Zhoushi Zhu
Amr Martini design_document4.pdf
final_paper1.pdf
photo1.jpg
presentation1.pptx
proposal1.pdf
video1.mp4
video2.mp4
video3.mp4
*We are collabarating with PSYONIC Inc. (http://www.psyonic.co/#ourstory) and they have agreed to give technical support and cover the cost for extra PCB orders (Separated from the courses timeline). One of our team member is a previous member in PSYONIC Inc.

Problem:

Nowadays, prosthetic hands are commonly controlled by Electromyographic (EMG) method which evaluates the electrical activity produced by skeletal muscles. However, the traditional EMG method is not accurate enough, because the measurements of that electrical signal suffer from high level noises. In addition, due to the physical layout and high cost of EMG sensors, the number of sensors is insufficient to acquire enough data to track the muscle movements precisely.

Solution:

In this senior design project, our goal is to develop a foam pressure-sensor based method as an alternative of the EMG method for controlling prosthetic hands, which includes a design of PCB to carry the electrodes array with its corresponding communication peripherals and programing of the communication protocol. The pressure sensor method is more accurate, less noisy and cheaper, and preliminary research (1) shows promising result to the method.

Technical Details:

In response to the question posted by professor Oelze, we have assembled some initial technical details.

Our project will consist of at most 10 pressure sensor modules and a master device. Each pressure sensor is based on a resistive working principle in which the interface resistivity between two surfaces changes according to the applied load. We will use metal trail on PCB as electrodes and use conductive foam as the sensor material. When load is applied the resistance between the electrodes will be changed and we can use the resistance change to sense pressure change.

Each pressure sensor module/PCB, as Professor Oelze suggested, will now contain 32 pressure sensors. However, exact number of sensors may be changed because of, for example time limitation, PCB limitation, or microcontroller limitation. Each pressure sensor module will also have a microcontroller (stm32, in order to interface with Psyonic and our team has worked with it before) to acquire and transmit the data. The analog voltage signals from the pressure sensors will be digitized by the microcontrollers’ embedded ADC. Since most of microcontrollers don’t have 32 ADC channels, we are going to use analog signal selectors or MUXs to select between data from different sensors.

The data from each module will be transmitted to master device using a communication protocol. In each reading of sensors, assume each digitized pressure sensor reading is 16 bits (12 bit data represented by 16 bit number), 320*16=5120 bits of data will be transmitted to the master. Therefore, we currently plan to use Controller Area Network (CAN) protocol for communicating between pressure sensor modules and master device. The CAN protocol can transmit as fast as 1 Mbit/s with high fault tolerance. Therefore, each transmission can be completed in about 5ms. In addition, SPI or I2C are also feasible choice for the communication protocol and both have their advantage and disadvantages.

The master device will mainly receive CAN message containing pressure sensor reading and process those data. On the master device, we are going to evaluate the data and classify them into command to the hand.

Although having lots of components, the cost of the project won’t be high. We are going to use the electronic packaging foam as the sensor material (already tested and meet our requirement), which is basically free. The cost mostly falls on PCBs, which we can get 10 pieces for $5 from PCBWay, and micro controllers. We expect cost of the components to be reasonable.

Answers to additional questions from Professor/TA:

Q: How are we going to the demo?

A: We will perform the demo on our own arm, here’s our plan:
1. The most basic goal for the demo is that we want to build a heat map of pressure for the muscle movement of different hand motion. We will classify those pattern into corresponding hand motion.
2. If things go well, we plan to control a real prosthetic hand borrowed from Psyonic Inc. With the method we developed, we hope to have the prosthetic hand mimic the movement of our hand.

Q: How do we know the solution applies to people without hand from the demo?

A: From previous research (1), they have shown that the method can achieve similar accuracy on both people with or without hand with the same system and our demo can prove its own effectiveness on people with hands.

Q: Muscle and tendon motions are much different when people do not have a hand. How do we account for that?

A: Even without tendons, the muscle will still bulge (1). As long as there exists a correlation between groups of hand motions and collected data, the amputees will be able control their specific prosthetic hand.

Novelty:
Even though our idea is inspired by the paper (1), we are doing more than what they have done by designing our own PCB and developing corresponding software and communication protocols. In addition, in that paper, the researchers are demoing using computer simulation but we are going to attempt to demo by controlling a real prosthetic hand.

(1) Castellini, C.; Kõiva, R.; Pasluosta, C.; Viegas, C.; Eskofier, B.M. Tactile Myography: An Off-Line Assessment of Able-Bodied Subjects and One Upper-Limb Amputee. Technologies 2018, 6, 38.

project idea: https://courses.engr.illinois.edu/ece445/pace/view-topic.asp?id=30680

rejected project: https://courses.engr.illinois.edu/ece445/pace/view-topic.asp?id=31316

Drum Tutor Lite

Zhen Qin, Yuanheng Yan, Xun Yu

Drum Tutor Lite

Featured Project

Team: Yuanheng Yan, Zhen Qin, Xun Yu

Vision: Rhythm games such as guitar hero are much easier than playing the actual drums. We want to make a drum tutor that makes playing drums as easy as guitar hero. The player is not required to read a sheet music.

Description: We will build a drum add-on that will tutor people how to play the drums. We will make a panel for visual queue of the drum and beats in a form similar to guitar hero game. The panel can be a N*10 (N varying with the drum kit) led bar array. Each horizontal bar will be a beat and each horizontal line above the bottom line will represent the upcoming beats.

There will be sensors on each drum that will fire when the drum heads is hit. The drums will be affixed with ring of light that provides the timing and accuracy of the player according to the sensors.

Of course with a flip of a switch, the drum could be a simple light up drum: when the player hits the drum, that particular drum will light up giving cool effects.

The system will be on a microprocessor. Or for more versatile uses, it could be connected to the computer. And a app will be written for the tutor.

Project Videos