<kbd id='992781cdf5'></kbd><address id='992781cdf5'><style id='992781cdf5'></style></address><button id='992781cdf5'></button>

          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.


          University of Leicester - Bioengineering Group


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


          Amy Lymn


          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



          Limiting Factors



          Electronic Control System

          By Will Scott-Jackson

          Design Rationale

          Breakdown of Sub-Systems


          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

          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


          By Ian Chapple


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

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


          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):




          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:


          Before Calibration

          Some Calibration

          After Calibration

          With Object Detection

          Extended Calibration


          Future Work

          Project Execution

          By Joe Ahuja


          Weekly Meetings

          Online Group

          File sharing group

          Conflict Resolution


          Project Timeline


          Thank you for listening!

          Use a spacebar or arrow keys to navigate