El liderazgo

For evaluation of the ShapeHand system performance, the ShapeRecorder API interface provided by the Measurand Company has been used. ShapeRecorder is able to display, capture, and export motion capture data from all the ShapeWrap devices. It can also be used for viewing and saving data from individual ShapeTape. Some basic instructions for the ShapeHand system are listed in Table 3.1, which perform the system initialisation, communication link setting up, data collection and application termination, respectively. During the evaluation, the user is required to wear the ShapeHand system as shown in Fig. 3-4, where two ShapeHand data gloves are worn on the user’s hands and the data

Stereoscopic Display Ethernet Connection Wireless Connection ShapeHand Data-glove Formatted Data Wireless Router Directional Wireless receiver Data Concentrator Data communication module Data output module Input module

concentrator is attached onto the forebody of the user with the help of Thoracic Harness. Moreover, the system could operate in the wireless mode after setting up the connection between the data concentrator and computer, thereby enabling the user to wander around freely.

Table 3.1: API commands for ShapeHand system

Figure 3-4. User equipped with ShapeHand system.

The outputs from ShapeHand are the position and orientation parameters of each finger joint. Two types of output are available from the ShapeRecorder software with one in numeric data formats (with file name extensions of ‘.C3D’,’.BVH’ and ‘.txt’), and the other one in a graphic format. As an example, a recorded ’.BVH’ file and a graphic output are presented in the following.

The ‘.BVH’ data file shown in Fig. 3-5 is decoded using a C/C++ program. From this

Commands Functions

Initialisation() ShapeHand data initialisation

ConnectToDataCollector() _{Connect to the data concentrator}

GetPortInfo(SERIAL_DEVICE) _{Identify and setup link to the detected tape}

GetHandData() _{Collect the hand data from data gloves}

figure, it is seen to contain five frames of hand motion data at the frame rate of 2.6 ms per frame. The output shows that the hand data have been recorded in a hierarchical structure which is based on the hand kinematic joint system [210]. Based on the hand joint structure and their offsets described in the ‘ROOT’ section (see Fig. 3-5(a)), which are the lengths of the hand phalanges, the ‘MOTION’ section (see Fig. 3-5(c)) records the position and rotation parameters of the joints corresponding to the channel IDs stated in the ‘ROOT’ section. For example, the first six data recorded in the ‘MOTION’ section in the line below the line of ‘Frame Time’ are corresponding to the channel IDs of ‘Xposition’, ‘Yposition’, ‘Zposition’, ‘Zrotation’, ‘Xrotation’ and ‘Yrotation’ in the line above the line of ‘JOINT thumbA’ in the ‘ROOT’ section, respectively,. It can be seen that according to the channel and Joint IDs, the ‘MOTION’ data structure starts with the three position and three rotation data of the hand base at the recording time, where the position normally remains as zero for it is the centre of the hand local coordinate system. It is then followed by 5 x 3 (5 finger digits, where each digit has three joints) groups of three-parameter direction data for each hand joint starting from the proximal joint of the thumb to the distal joint of the little finger.

Based on the recorded ’.BVH’file, the hand posture can be reconstructed as shown in

Fig. 3-6, where the hand model has been rotated -900 with respect to the x-axis (denoted

by XS) compared to the conventional right hand coordinate system in order to yield a

frontal view. It can be seen that the x-axis is towards the side of the narrow flat tape (in the

direction across the palm), the y-axis (denoted by YS) is perpendicular to the back of the

narrow flat tape (perpendicular to the back of the palm), and the z-axis (denoted by ZS)

runs along the length of the narrow flat tape (along each finger towards the finger tip), which forms a right hand coordinate. Furthermore, the origin is fixed at the bottom of the palm in the middle of the wrist.

(c)

Figure 3-5. ‘.BVH’ file output from ShapeRecorder.

Figure 3-6. Kinematic hand model where T1, T2 and T3 correspond to JOINT thumbA,

JOINT thumbB and JOINT thumbC in the ‘.BVH’ file. Similarly, I1, I2 and I3 to JOINT

indexA, JOINT indexB and JOINT indexC; M1, M2 and M3 to JOINT middleA, JOINT

middleB and JOINT middleC; R1, R2 and R3 to JOINT ringA, JOINT ringB and JOIINT

ringC; S1, S2 and S3 to JOINT littleA, JOINT littleB and JOINT littleC.

Furthermore, the graphic output of ShapeRecorder provides an animated approach to display the user’s hand configuration. As illustrated in Fig. 3-7, although the ShapeRecorder graphic output follows the hand movements of both hands, it does not provide the global positions and orientations of the two wrists. Furthermore, it can be seen that it has a video update rate of 62.5 Hz and a data transfer rate of 65.2 Hz. Incidentally, to obtain good tracking results of the hand, a thorough pre-calibration and a suitable glove size for the user’s hand are required, and more information can be found in the documentation provided by Measurand [206].

Xs Ys Zs

Figure 3-7. ShapeRecorder output for two hand gestures shown in the photograph.

Further testing has also been conducted to evaluate the hand joint angle measurement. By making the hand postures of ‘Homing’, ‘Open’, ‘Fist’ and ‘Claw’, with the hand joint angles programmed to be shown on the tool bar (see Fig. 3-8), the accuracy of the data glove measurement can be assessed. From Fig. 3-8, for the ‘homing’ hand posture, all four finger joint angles are seen to be approximately zero degree; for the ‘Fist’ hand posture, the pitch angles of all four finger joints are around their maximum values with the angles of the proximal and intermediate phalanges close to 90 degrees and the distal phalanges close to 45 degrees; for the ‘open’ hand posture, the pitch angles of all four fingers are close to 0 degree and the yaw angles between the four figures are at least 7 degrees; and for the ‘claw’ hand posture, the pitch angles are around the middle values between zero degree and their maximum values.

Figure 3-8. Joint angles for hand postures of (a) homing, (b) fist, (c) open, and (d) Claw, with the first 4x3 numbers on the data bar showing the pitch angles of each phalanx (starting from the proximal phalanx) on each finger (starting from the index finger) and

the last three numbers showing the yaw angles between four fingers.

Homing Open Claw Fist (a) (b) (d) (c)