1 of 20

Using EEG xARM approach for exoskeleton control

by Christopher Reyes & Dion Parra

2 of 20

3 of 20

4 of 20

Introduction

  • Worth Health Organization estimates that only between 5-15% of people who need prosthesis have access (Laurence, 2022)
  • Nearly 75% of amputations caused by diabetes and cardiovascular complications (M. Newton, 2022)
  • Cost of Prosthesis can vary depending on use ranging from below $10,000 up to $70,000 (primecareprosthetics.com)
  • People abandon prosthesis due to weight, cost, or difficulty of use (Engdahl et al., 2015)

5 of 20

Introduction

  • Electroencephalographic (EEG) interfaces are most useful for direct links to brain activity
    • Lack long term durability, and fast movements
  • Neural Interface technology for humans to communicate with robotic devices (RDs), such as prosthetic limbs and exoskeletons (Kastalskiy, 2018)
  • Limits neuromuscular control of robotic devices such as prosthetics

6 of 20

Review of literature

  • Neuromuscular Interface for Robotic Control (Kastalsky et.al, 2018)_-
    • Where able to find ways to make systems that could interpret EEG systems faster and longer without error
    • Errors decreased as days went on
    • More advanced algorithms would be needed to help process EEG

7 of 20

Review of literature

  • Self Contained Neuroskeletal Arm Prosthesis (Ortiz-Catalan et,al 2020)-
    • Prosthesis was successful
    • Smoother movement and no prior training was required
    • Needed a neural interface that provided faster feedback
    • Faster learning method

8 of 20

Engineering Goal

Problem Statement Goal

Prosthesis control and movement is difficult with EEG

Making interface for xARM robot movement/control with EEG can help develop solutions

9 of 20

Materials

Muse Headband

  • xArm Uno – Robotic arm used for prosthetic simulation.�
  • Muse 2 EEG Headband – Captured brainwaves (focused on beta waves linked to motor intent).�
  • Laptop – Handled coding, data streaming, and device interfacing.�
  • USB/Bluetooth – USB used initially for reliable connection.

xARM

10 of 20

Methods & Materials

  • PyCharm IDE: Chosen for its robust Python environment management and built-in terminal access. This enabled seamless installation of dependencies and execution of control scripts.
  • Python 3.x Environment Setup: Encountered and resolved issues with pip installation in Windows Command Prompt. Isolated virtual environments were created to ensure conflict-free development.

​

11 of 20

Methods & Materials

  • 1. Fixing Python Installation Issues

We initially ran into problems using Python’s package installer pip in the command prompt. After troubleshooting and updating system paths, we were able to install all necessary packages.

  • 2. Installing xArm Control Libraries

The xArm’s official software was outdated and couldn’t connect through USB. We switched to using updated libraries from GitHub. The following commands were run in the PyCharm terminal to get everything working:

​

12 of 20

Intended Methods & Materials

Muse

Python connection

Laptop

Connection

xARM

receiving/sending info

receiving/sending info

Connection between systems

13 of 20

Similar Methods & Materials

T. H. Hsieh, "An EEG-based Approach for Exoskeleton Control," Final Report of College Undergraduate Research Project, 2011

14 of 20

Similar Methods & Materials

T. H. Hsieh, "An EEG-based Approach for Exoskeleton Control," Final Report of College Undergraduate Research Project, 2011

Start

Receive EEG data

EEG post- processing

Neutral State?

State 1?

Send Command 1

Keypress?

End

Yes

Yes

Yes

State 2?

No

Send Command 2

Yes

15 of 20

Expected Results

  • EEG headset should be able to detect messages sent from the brain, and the arm should be able to process and correctly act out within timely manner
  • Low latency, and good feedback
  • The system between the EEG headset and the xARM to work as intended
  • More flexibility in users movement and position

16 of 20

Similar Results

  • Was successful in processing EEG and to listen to the commands given

​

  • Both NXT Robot, and Exoskeleton in this study were able to move, and operate with the headset commands
    • Response time was delayed, and you had to be seated

T. H. Hsieh, "An EEG-based Approach for Exoskeleton Control," Final Report of College Undergraduate Research Project, 2011

17 of 20

Discussion - Future Research and Implications

  • If successful, this technology could begin to be applied to more sophisticated machines and software
    • Would allow for less abandonment of prosthesis, faster response time, and could be the closest to attempting to replicate human limbs
  • Could help reduce the cost of prosthesis, and exoskeleton for people in need

​

18 of 20

Discussions-Next Time

  • Begin to try more advanced software and machines
  • Obtain Insight- 5 Channel EEG Wireless Headset
    • Contains more user friendly software, and have more ways to connect with outside sources
  • Obtain another “exoskeleton” or “prosthetic” close to how EVE works by Automata
    • Easy to program, and to connect to outside sources

19 of 20

References

20 of 20

EMG

EEG

Input Module

Processing Modules

Executor modules

NXT robot

NAO robot

Exo skeleton

Input device driver

Data preparation

Output interface

Configuration

Input Interface

Classification

Algorithm

Output Interface

Configuration

Input Interface

Command

implementation

Output device

driver

Configuration

Bio potential Data

Normalized Data

Pattern number

Command sequence

Methods & Materials

GUI Controlled module configuration