Utilizing Hall Effect Sensors and Artificial Neural Networks to Classify Soft Magnetic Actuator Pose

By Arhan Sankhla
Slide 1: Title slide for research presentation on utilizing Hall Effect Sensors and Artificial Neural Networks to classify soft magnetic actuator pose

Slide-1

Arhan Sankhla, REU Student

College of Engineering, Rice University

Research Mentor: Dr. Erik Engeberg, Professor, Department of Biomedical Engineering

NSF REU: Sensing and Smart Systems

Slide 2: Table of contents showing six sections - Background, Research Gap, Techniques, Actuator Manufacturing Process, Set Up, and Results

Slide-2

Table of contents

01 Background

02 Research Gap

03 Techniques

04 Actuator Manufacturing Process

05 Set Up

06 Results

Slide 3: Background section explaining soft robotics applications and soft pneumatic actuators

Slide-3

Background

Soft robotics' high dexterity and safety make them ideal for biomedical applications. Ideal for gripping things, locomotion, and other biomedical devices, where the environment is highly dynamic and sensitive to physical interaction.

Soft pneumatic actuators are the dominant technology in soft robotics due to its low cost and mass, fast response time, and easy implementation

Slide 4: Background definitions for Ferrofluid, Dragonskin, NPR Tool, Hall Effect Sensor, and Pose

Slide-4

Background

Ferrofluid - a liquid made of nanoscale magnetic particles suspended in a carrier fluid

Dragonskin - basically stretchy/rubbery feeling silicone

NPR Tool - is a machine learning technique that uses artificial neural networks (ANNs) to recognize patterns in different types of data. ANNs are computational systems that are modeled after the human brain and can learn to associate input patterns with output categories.

Hall Effect Sensor - a sensor that measures the intensity of the magnetic field in volts

Pose - position of the actuator in space

Slide 5: Research Gap describing challenges with integrated sensing and intelligent control in soft pneumatic actuators

Slide-5

Research Gap

Despite recent breakthroughs, soft pneumatic actuators and robots experience challenges related to integrated sensing and intelligent control.

Mainly there is no effective way to measure where the actuator is in space (position) and what the shape is of the actuator at that time

These gaps are significant because without ways to actively and effectively control these actuators their use in biomedical and other important applications will be diminished

Slide 6: Techniques describing the methodology using Hall Effect sensors, different ferrofluid ratios, and neural network training

Slide-6

Techniques

Use a hall effect sensor which measures the intensity of the magnetic field along with a soft actuator made out of different ratios of magnetic material (ferrofluid) mixed with silicon

Measure the displacement of the actuator based on the hall effect sensors reading of the magnetic field at specific pressures

Train a neural network with the data from 3 different pressures for each different ratio of ferrofluid to silicon

Test the neural network with training data to see which ratio of ferrofluid to silicon is best for determining the shape/pose of the soft actuator

Slide 7: Actuator Manufacturing Process showing 4-step process with ratios and curing time

Slide-7

Actuator Manufacturing Process

  1. Mix DragonSkin and Ferrofluid in a 85:15, 70:30, and 50:50 ratio
  2. Pour into mold
  3. Place tubing inside mold
  4. Let cure for 24 hours
Slide 8: Set Up diagram showing experimental apparatus with Hall Effect sensor, actuator, and measurement equipment

Slide-8

Set Up

The experimental setup shows a Hall Effect sensor positioned to measure the magnetic field intensity as the soft actuator changes position. The actuator is connected to pneumatic control equipment that allows for precise pressure adjustments. The setup includes measurement equipment to record both the Hall Effect sensor readings and the actuator displacement at various pressure levels.

Slide 9: Additional Set Up information showing detailed experimental configuration

Slide-9

Set Up

The experimental setup demonstrates the positioning of the Hall Effect sensor relative to the soft actuator. The configuration allows for consistent measurement of magnetic field changes as the actuator undergoes deformation at different pressure levels. The setup ensures repeatable measurements across different ferrofluid to Dragonskin ratios.

Slide 10: Results showing Hall Effect Signal Data graph and Confusion Matrix for neural network classification

Slide-10

Results

Hall Effect Signal Data

The graph shows Hall Effect sensor voltage measurements over time for different actuator configurations. The data demonstrates varying signal patterns corresponding to different ferrofluid ratios and pressure levels.

Confusion Matrix

The confusion matrix displays the neural network's classification accuracy for different actuator poses. The matrix shows how well the trained neural network can distinguish between different actuator positions based on Hall Effect sensor readings.

Slide 11: Results summary describing methodology using Matlab NPR tool and findings about ferrofluid ratios

Slide-11

Results

Use Matlab NPR tool (Neural Net Pattern Recognition) to train on the pressures selected for each actuator

Input experimental data for each pressure and output the classification accuracy of the neural network for each different combination of Ferrofluid:Dragonskin actuator

Figure out whether or not the amount of ferrofluid in the actuator impacts the hall effect sensor's ability to get consistency in data

Make recommendations on the amount of ferrofluid to use when making Soft pneumatic actuators and whether hall effect sensors work for control/sensing of actuators

Slide 12: References listing two academic papers on soft medical robotics and fluidic mass control

Slide-12

References

Hsiao, Jen-Hsuan & Chang, Jen-Yuan & Cheng, Chao-Min. (2019). Soft medical robotics: clinical and biomedical applications, challenges, and future directions. Advanced Robotics. 33. 1-13. 10.1080/01691864.2019.1679251.

H. Xu, P. Agarwal and B. Stephens-Fripp, "Constant Fluidic Mass Control for Soft Actuators Using Artificial Neural Network Algorithm," 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Prague, Czech Republic, 2021, pp. 1732-1739, doi: 10.1109/IROS51168.2021.9636630.

Last slide: Contains plain text stating 'End of presentation. Click the right arrow to return to beginning of slide show.'

End of Presentation

Click the right arrow to return to the beginning of the slide show.

For a downloadable version of this presentation, email: I-SENSE@FAU.

Additional Information
The Institute for Sensing and Embedded Network Systems Engineering (I-SENSE) was established in early 2015 to coordinate university-wide activities in the Sensing and Smart Systems pillar of FAU’s Strategic Plan for the Race to Excellence.
Address
Florida Atlantic University
777 Glades Road
Boca Raton, FL 33431
i-sense@fau.edu