Wearable Sensor Analysis for Gesture Recognition

Document Sample
Wearable Sensor Analysis for Gesture Recognition Powered By Docstoc
					 Wearable Sensor
Analysis for Gesture
   Recognition

    Supervisor:   Dr. Manolya Kavakli
    Student:      Alexey Novoselov
    St. ID:       41650883
                      Agenda
1: Background;
2: Goals;
3: Intended Outcomes;
4: Significance;
5: Approach;
6: Contemporary Technologies;
7: Piezo-Electric Technology;
8: Problems;
9: Experiments;
10: Output Data;
11: Initial Analysis;
12: Mathematical Approach;
13: Algorithmic Approach;
14: Conclusion
15: Future Work
                                2
               1: Background

 Virtual Reality is very promising scientific area;
 Needs tools for interaction;
 A lot of various technologies and devices;
 Sensor Jacket is one of them;
 No appropriate software for output analysis;



                                                 3
                   2: Goals

 Investigate contemporary Motion Capture
  technologies and techniques;
 Collect Sensor Jacket’s output data;
 Analyze it;
 Develop a mathematical model or an algorithm for
  Sensor Jacket;



                                                 4
        3: Intended Outcomes

 The list of justified experiments;
 Data, collected during the set of
  experiments;
 General characteristics of signals;
 Application of mathematical approach;
 Application of algorithmic approach;
 Mathematical model or algorithm for
  signal processing;
                                          5
            4: Significance

 Other Motion Capture systems not very
  convenient in use;
 Sensor Jacket is wearable;
 Sensor Jacket is simple;
 But is has no software for output
  analysis;
                                      6
                 5: Approach

 Develop and perform experiments;
 Collect data;
 Filter data;
 Analyze data;
 Develop a tool;

                                     7
   6: Contemporary Technologies

 Optical;


 Inertial;


 Mechanical;


                                  8
    7: Piezo-Electric Technology

 Piezo-effect;


 Piezo-electric sensor;


 Made of graphite and silicone rubber;
                                          9
              8: Problems

 Only static characteristic provided;

 Electric noise in the output channels;

 Strongly non-linear output signal;

 Speed dependent;

                                           10
9: Experiments




                 11
                     10: Output data
1600

1400
1400


1200
 1200


 1000
1000

 800
 800
 600
 600
 400

 400
 200


 2000
        1 9 17 25 33 41 49        161 181 201 81 89 97 105 113 121 129 137 401 421
        1 21 41 61 81 101 121 14157 65 73 221 241 261 281 301 321 341 361 381145 153


                                                                                       12
          11: Initial Analysis

 Minimal (Starting) value;
 Maximal (Peak) value;
 Steady Value;
 The overshoot;
 Difference between initial and final values;

                                             13
    12: Mathematical Approach

 Basic characteristics;

 Speed of the signal change;

 Angle of slope of the signal;

 Area of the signal;

                                  14
          13: Algorithmic Approach
1.   Determine the active sensors;
2.   Calculate the area of their signals;
3.   Calculate the speed of sensors’ signals change;
4.   Using the graphs (Figures 18-20), calculate the real speed of
     movement;
5.   Using the tables of active sensors, determine to which types
     of movement (AoF, AoS, or AStF) this motion belongs;
6.   Determine the approximate direction of movement;
7.   Calculate the average time between the start and the peak of
     transient process for each group of sensors, forming the
     movement type;
8.   Using the information from steps 4, 5, and 6, calculate the
     approximate distance that operator’s hand has passed during
     the motion in each direction;
9.   Calculate the final coordinates of operator’s palm using
     formulas;
                                                                 15
            14: Conclusion

 Technologies review;
 Designed and performed experiments;
 Collected, filtered, and analysed data;
 Mathematical approach did not succeed;
 Algorithm created;

                                            16
            15: Future Work

 More experiments;

 More sensitive data filtration;

 Use advanced mathematical techniques;

 Create more accurate and precise tool.


                                           17
Thank you!




             18