Low-Cost Self-Calibration Data Glove Based on Space-Division Multiplexed Flexible Optical Fiber Sensor

Abstract

Wearable devices such as data gloves have experienced tremendous growth over the past two decades. It is vital to develop flexible sensors with fast response, high sensitivity and high stability for intelligent data gloves. Therefore, a tractable low-cost flexible data glove with self-calibration function based on a space-division multiplexed flexible optical fiber sensor is proposed. A simple, stable and economical method was used to fabricate flexible silicone rubber fiber for a stretchable double-layered coaxial cylinder. The test results show that the fiber is not sensitive to the temperature range of (20~50 °C) and exhibits excellent flexibility and high stability under tensile, bending and torsional deformation. In addition, the signal detection part of the data glove enables compact and efficient real-time information acquisition and processing. Combined with a self-calibration function that can improve the accuracy of data acquisition, the data glove can be self-adaptive according to different hand sizes and bending habits. In a gesture capture test, it can accurately recognize and capture each gesture, and guide the manipulator to make the same action. The low-cost, fast-responding and structurally robust data glove has potential applications in areas such as sign language recognition, telemedicine and human-robot interaction.

Keywords: data glove; flexible sensor; human-robot interaction; optical fiber; wearable devices.

Figure 1

(a) Structure diagram of data glove. (b) Analogy of flexible optical fiber and index finger behavior during glove bending. (c) Linkage between data glove and manipulator.

Figure 2

(a) The cross-section of flexible optical fiber. (b) The state of flexible optical fiber under tension, bending and torsion. (c) Manufacturing process of flexible optical fiber.

Figure 3

Mechanical and optical properties of the fibers. (a) Strain-normalized intensity curve of flexible fiber at wavelength 650 nm (the point “A” indicates that the fiber breaks only when the strain is greater than 90%). (b) Normalized spectra of flexible optical fiber with increasing strain from 0 to 50%. The inset shows the relationship between the normalized intensity at 650 nm and the applied tensile strain. (c) Dynamic tensile test of a flexible fiber being stretched 50 cycles at the strain range 0–50%. (d) Propagation loss of flexible fiber, measured in air by cutback method. Fibers for deformation sensing. Dynamic tensile test of flexible fiber being stretched 50 cycles with 50% applied peak strain at different temperatures, (e) 20 °C, (f) 30 °C, (g) 50 °C. (h) Comparison of the first cycle and the 50th cycle in tensile test at room temperature. The different color lines in (c,e–g) represent each stretch cycle.

Figure 4

Bending and torsion-sensing performances of the flexible optical fiber. (a) Definition of bending angle. (b) The flexible fiber is cyclically bent 50 times in the range of 0~120° bending angle. The different color lines represent each bending cycle. (c) The angle-normalized intensity curve at 10 intervals during the 10th to 100th cycles and the 1st cycle in the bending test. (d) The flexible fiber is repeatedly twisted 50 times in the range of 0~90° torsion angle. The different color lines represent each torsion cycle. (e) The angle-normalized intensity curves during the 1st, 50th and 100th cycles in the torsion test.

Figure 5

Flow chart of data collection and processing, including pretreatment—automatic calibration and data processing.

Figure 6

(a) Gray value change diagram of the whole palm closing action in three consecutive states of relaxing–bending–relaxing. (b) Gray value change diagram of gesture change (four processes of index finger bending).

Figure 7

The quantization process of finger bending of data glove and intuitive observation of gesture capture process. The data glove from the stretch state repeatedly makes the (a) Number 1, (b) Number 2, (c) Number 3 and (d) Number 4 gestures from the stretch state. The yellow area represents the data block of the five-finger stretch state, and the gray area represents the data block of the corresponding gesture.

Figure 8

Video screenshots of linkage between data glove and manipulator. The manipulator tracks the hand gestures of the data glove in real-time. (a) Number 1. (b) Number 2. (c) Number 3. (d) Number 4.