Your browser doesn't support the features required by impress.js, so you are presented with a simplified version of this presentation.

For the best experience please use the latest Chrome, Safari or Firefox browser.

Eye-Tracking Robotic Arm

University of Leicester - Bioengineering Group

Contents

  1. Introduction
  2. Mechanical
  3. Electronic
  4. Control
    1. Eye-Tracking
    2. Object Detection
  5. Project Execution

Introduction

Amy Lymn

Project Aims

Mechanical Systems Design

Amy Lymn & Joe Ahuja

Mechanical Systems

Mainly Horizontal Displacement

Mainly Vertical Displacement

Concept Ideas

Concept 1

Measurements in metres

Concept 2

Measurements in metres

Concept 3

Grip Concepts

Development of Chosen Concept

Detailed Mechanical Design

Full Technical Drawings

Monitor Manufacture Process

Integration

Testing

Limiting Factors

Improvements

Conclusions

Electronic Control System

By Will Scott-Jackson

Design Rationale

Breakdown of Sub-Systems

Controller

A controller was required to implement the logic of the control system, it must:

Having considered the options, an Arduino Mega 2560 was selected:

This was selected because:

  • It has many digital I/O pins
  • 16 Analogue to Digital Converter pins
  • USB port can interface with PC via serial link
  • 16MHz of processing power
  • Can be programmed with an Open Source IDE
http://nicegear.co.nz/obj/images/arduino_mega.jpg

Actuator Systems

Four electrical motors were required to drive the four degrees of freedom. There were several factors that needed to be considered:

Actuator Systems (Research)

Actuator Systems (Research cont.)

Actuator Systems (Selections)

Based on the aforementioned research, several electrical motors were selected:

One Trinamic QSH-5718-51-28-101

One Astrosyn - MY3002 Size 11 Stepper Motor

Two RE 385 DC Motors

Actuator Systems Circuitry Development

In order to actuate these motors, specific driver circuity was required, although there were some factors to take into consideration:

Feedback Systems

Feedback Systems (Implementation)

The feedback system corrects errors caused by disturbances caused by motor slip etc.

User Interface

In order to develop this system into a standalone unit, several major components were required:

Progress to Date

Problems and Limitations

Further Improvements

Electronic Control System

Live Demo

Control Systems

Eye-Tracker

By Ian Chapple

Hardware

  • Build a cheap eye tracker
  • Normally cost >£10,000

  • Built for ~£70 using:
    • Playstation Eye Camera
    • Infra Red LED's

Software

Image Processing:

Data Mapping:

Contour Finding

Find the edges of objects in an image

Circle Finding

Locate circular objects from within an image

Glint Finding

Find brightest point in image

Data Mapping

Linear 3D map:

Use 2 sets of equations to locate coorinates (x,y):

Eye-Tracking

Video

Control

Object Detection

By James Reuss

Why is Object Detection Needed?

Because the system is unable to detect its environment!

How can this help?

What's the Problem?

All of these systems are expensive!

The Kinect

How does the Kinect work?

Pixel Organisation

RGB and Depth Images - Depth in millimetres

Point Clouds

The Algorithms

Need to convert the raw data into a usable form.

Then it is posible to detect the objects.

This requires the use of the following operations:

Surface Normals

Calculate the surface normal of each 3D point

using neighbouring points.

Plane Segmentation

As mentioned previously, this contains many steps

Point Clustering

Use Tags!

Now each point has a tag

Cluster Filtering

Now filter out tags that don't represent a plane

Dominant Plane Finding

Which plane cluster is the table?

NB: This assumes that there is a table

The RANSAC Algorithm

RANdom Sample And Consensus

The Convex Hull Algorithm

Point Inclusion

There is now a lot of information available:

Which points make up objects on the table?

Use a Point-In-Polygon algorithm

Object Point Clustering

K-Means Algorithm

A process by which:

Results

Before Calibration

Some Calibration

After Calibration

With Object Detection

Extended Calibration

Improvements

Future Work

Project Execution

By Joe Ahuja

Overview

Weekly Meetings

Online Group

File sharing group

Conflict Resolution

Budget

Project Timeline

Summary

Thank you for listening!

Use a spacebar or arrow keys to navigate