Word_Template.rtf



A Novel Method for Joint Motion Sensing on a Wearable Computer

|Aaron Toneyα, |M. Billinghurstβ |

|αDepartment of Electrical Engineering |βHuman Interface Technology Laboratory |

|University of Washington |University of Washington |

|Box 352-500 |Box 352-142 |

|Seattle, WA 98195, USA |Seattle, WA 98195, USA |

|joeboy@u.washington.edu |grof@hitl.washington.edu |

Abstract

Some wearable computing applications require low cost sensing devices that detect the deflection of joints during human motion. We present a novel non-invasive technique for measuring joint motion using low cost pressure sensors. We describe a specific example of this; a glove that can be used as a low cost, high resolution gesture input device for wearable computers.

1: Introduction

Body worn wearable computers enable new types of human-computer interaction because of their long term physical intimacy. Rather than the user manipulating an input device such as a chordic keyboard or mouse, the body itself can be used as the primary input mechanism. This requires the monitoring of physical and physiological factors. Previous researchers have shown examples of full body [1], foot [2] and physiological input [3] using a variety of body worn sensors. A common requirement of many such applications is a non-invasive accurate sensor for measuring joint deflection. In this paper we describe how small low-cost pressure sensors can be used to detect joint motion and present a specific application; a glove with multiple joint sensors that can be used as a gesture input device for wearable computers.

2: Theoretical Background

The helmholtz equation gives the relationship between pressure, volume, temperature, and molecular weight of molecules present molecules present.:

P V = nr T Eqn. 1

If we assume temperature is constant, then changes in volume result in changes in pressure. The volume of an elastic sealed cylindrical tube of fluid will change as it is bent. Since there is a fixed volume of gas in the sealed tube, we have the linear relationship:

P ∝ 1/V Eqn. 2

So we can measure the bend of an appendage by having a fluid filled tube mounted over the joint and measuring the change in tube volume with respect to angular displacement. Change in volume can be found by measuring the change in fluid pressure.

A bent tube of length L and volume V can be approximated by N straight segments of length L/N and volume V/N. For a total angular displacement of θт˚ degrees across a limb with N joint separated at lengths of {L1,L2,L3,…LN}, and constant cross sectional area Ą, the total volume in the tube is given by the sum of the volume of all the curved sections and all of the straight sections. The total volume of these sections is given by;

N

Vstraight = Ą ( Ln Eqn. 3

n=1

N θn˚ θт˚

VBent = ( [½Ą (tan θ ∂θ]( [½Ą(tanθ∂θ] Eqn. 4

n=1 θ(n-1) ˚ 0˚

N θn˚

VTotal = Ą ( [Ln +½ (tan θ ∂θ] Eqn. 5

n=1 θ(n-1) ˚

As both the straight and curved cylinder sections share the same cross sectional area, the ratio of volume change is proportional to the ratio of length change. Thus we can measure the length Ln and bend angle (tan θ of each joint using a pressure sensor.

The sensitivity of our angular measure relies on using the smallest total volume possible while generating the greatest change in volume possible for a given angular displacement. The diameter of the cylinder is critical in determining the volume’s dependence on curvature. The smaller the volume the greater the sensitivity.

3: The Data Glove

To demonstrate the viability of using pressure sensors to measure angular deflection we constructed a data glove. Fluid filled tubes were mounted along the fingers and pressure sensing electronics mounted on the back of the hand as shown in figure 1.

|[pic] |[pic] |

| |Finger Mounting |

| |Figure 1 |

Figure 2 shows a layout of the angles being measured for each finger in the glove. In the human hand the angular displacement C is in direct proportion to the displacement A. If we know the total length of the finger L1 + L2 + L3 from an initial calibration of the system, and the total curvature of the hand B, then we can compute L1, L3, A and C. So a single sensor can be used to measure the curvature of the finger.

Figure 1

Glove_finger_angles.ps

Four MPX2010 pressure sensors were used [ref]. These each have a typical voltage swing of 20mV, which was amplified to 2.5 Volts and noise filtered out using the simple LM351 OP amp RC circuit shown below.

[pic]

Figure 4 Transducer amplifier configuration

Pressure data is sampled from each finger at a rate of 240Hz and transmitted back to the computer at 9600bps. From a pool of seven people polled the fastest frequency to completely open and close the hand was 8Hz, with an average frequency of 4.5Hz. This corresponds to a worst case of one sample every 12°. However for speeds less then 2.6Hz, sampling will occur every 4°.

In order to test out the Glove a series of measurements were taken with the gloved hand grabbing onto cylindrical object of a known diameter. The results are plotted in Figure 5. Above a minimum size, the voltage measurement increases linearly with diameter showing that the glove can be used as an input device.

[pic] Figure 2 Data Glove finger physical overview

4: Conclusion

In this paper we have shown how inexpensive pressure sensors can be used to measure joint bending. While we have demonstrated the applicability of this in a dataglove, the technique can be applied for sensing one degree of freedom joint motion for any joint in the body. While these results are in themselves significant enough to call the prototype a success, it is worth noting that these are the accuracy measurements for the crudely constructed prototype. It is not hard to imagine that any commercially produced glove would greatly reduce the sources of error and increase accuracy significantly.

9: References

[1] M. Coniglio, Toika Ranch, The MidiDancer System. See “http;//~troika/mididancer.html”

[2] J. Paradiso, E. Hu Expressive Footwear for Computer-Augmented Dance Performance. In ISWC 97

[3] R.W. Picard, J. Healey, Affective Wearables. In ISWC 97

[4] Op Amp reference

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download