Physical activity improves general health and well-being. The benefits of physical activity include prevention of cerebral accidents, cardiovascular disease, and some cancers. These benefits can also be enjoyed in old age, when the incidence of many diseases and disabilities increases. Researchers have made numerous attempts to develop methods for measuring physical activity because precise measurements of physical activity will enable the assessment of physical activity in the same way as a physical activity diary. Self-report measures are often used to estimate physical activity. However, they have some disadvantages such as the difficulty of estimating activity patterns throughout the day [1] and fluctuations in health status [2] . Therefore, objective human physical activity estimation methods are attracting increasing attention because they can overcome the limitations of self-report measures. Particularly, measurement of the acceleration of the body of a subject provides information about the amount, frequency, and duration of physical activity [3] . An objective physical activity measurement method such as accelerometry can also assess free-living activity. Objective physical activity measurements [4] prove that participation in non-exercise physical activity such as housework and climbing stairs reduces mortality risk.

In particular, walking for exercise is especially effective for long-term weight loss [5] , increasing HDL (high-density lipoprotein cholesterol) [6] , reducing blood pressure [7] , and lowering the risk of heart disease and cancer initiation. In the last decade, gait analysis has attracted much attention from computer vision researchers motivated by its wide application potential. In vision-based human motion detection systems, motion features are typically extracted from image sequences. Subsequently, the magnitudes of the features are compared as an identification process. However, in conventional human motion detection methods, experimental conditions such as the distance between the subject and camera and the angle and pixel density of the camera significantly influence the recognition rate of the system. Moreover, the primary drawback of these methods is that they cannot detect anything in the dead zone of the camera. Furthermore, many image-processing systems require complex logic to compensate for the disturbances caused by the motion of nonhuman objects.

Alternatively, body-mounted accelerometers [8] -[10] are extensively used for monitoring human motion because these systems are inexpensive when compared with an optoelectronic motion capture system. Another advantage of body-mounted accelerometers is that the system can be used in an indoor/outdoor space. Additionally, numerous methods have been presented to measure human activity using different techniques such as an ultrasonic motion analysis system to temporally and spatially measure the amount of human activity, an electromagnetic 3-D orientation estimation system that uses the Earth’s magnetic field, and a wearable ultrasonic motion analysis system. However, in these measurement systems, some type of sensor or marker that remains in contact with the subject must be used. Therefore, these methods have not been applied in noncontact detection of human activity measurement.

In this paper, we present a new objective human physical activity estimation method that does not use a camera or an accelerometer. We have developed an effective noncontact and nonattached technique for the detection of human walking motion using human-generated body charge. This technique involves the detection of an electrostatic induction current of the order of approximately sub-picoamperes flowing through an electrode placed at a distance of 3 m from a subject. The absolute value of electrostatic induction current depends on the type of footwear of the subject and floor material. However, we confirm that this technique has sufficient sensitivity to detect the electrostatic induction current generated during the walking motion in daily life. This technique effectively explains the behavior of the waveform of the electrostatic induction current flowing through a given measurement electrode via a capacitance model for the human body. Moreover, a study wherein the leg of a normal subject was attached to a splint using bandages was conducted to determine the change in the subject’s gait. The waveform of the electrostatic induction current generated owing to the motion of walking was observed to vary under different fastening conditions (Cases 1-3 described in the experimental method section). Experimental results show that the cadence patterns of the induced current vary with the fastening conditions. The proposed technique enables the detection of even subtle differences in walking motion thus proving that the quality of walking motion can be estimated using the proposed technique.

This paper is organized as follows. Section 2 explains the principle of the detection for the electrostatic induction current generated by walking motion under noncontact and nonattached conditions. The experimental method is described in Section 3. The experimental results and discussion are presented in Section 4. The final section presents the conclusion of this paper.

Let us consider the example of a walking motion to explain the principle of the abovementioned measurement technique. The human body is electrically charged during walking [11] -[14] . In the case when a subject is standing or walking, we assume that two highly resistive layers exist between the feet of the subject and the floor, as shown in Figure 1. One layer is the sole of the footwear of the subject, and the other is the floor surface.

The reciprocal of capacitance of the feet relative to the ground may be calculated as the sum of the reciprocal of capacitance of the sole and the reciprocal of capacitance of the floor surface, as shown below:

The electric capacitance between the sole and floor can be expressed as follows:

where S is the area of the sole, and S_c is the contact area between the sole and floor. Further, is the delamination area of the sole against the floor, and x and x₀ denote the distances between the sole and the earth and the floor surface and the earth, respectively. In addition, ε_a is the permittivity of the air gap between the sole and the floor, and ε_f is the averaged permittivity of the floor materials between the floor surface and the earth. Further, S must be replaced with an area of the sole projected onto the floor. However, the projected area is exactly similar to S in the case of normal waking motion. Therefore, S is assumed to be a time-independent constant value in place of the projected area in the above equation.

On the other hand, the electric capacitance of the human body (C_B) during walking motion can be expressed as follows:

Therefore, the electric potential of the human body during the motion of walking can be expressed as follows:

where Q_B is the instantaneous charge of the human body during walking motion. The induced charge of a measurement electrode placed at a certain distance from the subject can be expressed as

where C is the capacitance between the human body and the measurement electrode in the wireless portable sensor, and V is the potential of the measurement electrode. From the above two equations, the induced current flowing through the measurement electrode can be expressed as follows:

We assume that the human body is a good conductor. The second term on the right-hand side of Equation (6) represents the current induced by the motion of the foot before it is lifted off the floor. When the toe of the right foot is lifted off the floor, S_c decreases. Therefore, I increases as predicted by the second term on the right-hand side of Equation (6). On the other hand, the first term on the right-hand side of Equation (6) represents the current induced by the motion of the foot before it is lifted off the floor. The first term is approximately proportional to the speed of the foot in a direction perpendicular to the floor. However, the first term of Equation (6) has a small effect on I because ε_a is less than ε_f. Additionally, the speed of motion in a direction perpendicular to the floor is less in the case of typical walking motion. Therefore, the influence of the first term of Equation (6) on signal intensity is small when compared with that of the second term of Equation (6) in the case of normal walking motion. Therefore, in the case of walking motion near the measurement electrode, the electrostatic induction current generated when the subject’s foot is in contact with and off the floor can be measured under perfect noncontact and nonattached conditions.

A schematic of the measurement system for the electrostatic induction current generated owing to the variation in the electric potential of a subject’s body is shown in Figure 2. The electrostatic induction current flowing through an electrode in the wireless portable sensor placed less than 5 m from the body of the subject is converted into voltage with an converter having a conversion ratio of 20 V/pA and comprising an operational amplifier (op-amp). The converter consists of two low input current op-amps, a feedback capacitance of approximately 0.2 pF, and a feedback resistance of 10 TΩ, as shown in Figure 3. The stray capacitance is formed by winding a tinned copper wire on the feedback resistance because the required electric capacitance is less than the minimum value of a marketed capacitor. The selected low-noise op-amp has an input offset voltage of 40 μV and an input offset current of 1 pA. The feedback resistor connected to the op-amp is a hermetically sealed large value resistor that can prevent stray current due to humidity. For measurements with high input resistance, a conventional guarding method is absolutely necessary for shielding op-amps; this prevents stray currents from entering sensitive nodes. Sensitive nodes are surrounded by a guard conductor that is maintained at the same electric potential as the sensitive node. In addition, induction currents generated by commercial power sources manifest in the form of noise. Therefore, a filtering system with a cutoff frequency

of 20 Hz and a 60-Hz notch filter are used. This measurement system is unaffected by noise from other electronic devices such as mobile phones and microwave ovens. The analog signals are subsequently converted into digital signals with a 16-bit analog-to-digital (A/D) converter.

The XBee protocol is used for wireless data transmission from the portable motion detection system to a personal computer. Data were acquired at a sampling frequency of 100 Hz, which is safe because the actual data transmission rate on XBee networks is as low as approximately 10 kbps. This sampling frequency is sufficient for detecting contact events. The measurement electrode is a square with a side of 2 cm. The wireless portable sensor was placed in the ceiling of an ordinary house with wooden floors, as shown in Figure 4. The subjects were asked to wear slippers and walk normally on the wooden floor. The subjects (A, B, and C) were 3 healthy men aged between 19 and 20 years. The subjects were asked to walk in a cycle of approximately 100 steps per minute to maintain the same cycle of walking. Moreover, the signal produced when the subject walked with the ankle and knee fastened to a splint with bandages to simulate a limp was measured. Subject A was asked to perform different tasks such as walk normally (Case 1), walk with the left ankle fastened to a splint using bandages (Case 2), and walk with the left ankle and left knee fastened to a splint with bandages (Case 3).

This paper presented the development of an effective noncontact and nonattached technique for the detection of human physical activity using human-generated body charge. This technique involved the detection of an electrostatic induction current of the order of approximately sub-picoamperes flowing through an electrode. A theoretical model was proposed for the electrostatic induction current generated during the walking motion. The proposed model was compared with the theoretical model and was shown to effectively explain the behavior of the waveform of the electrostatic induction current flowing through the measurement electrode. The normal walking motions of daily living were detected with a portable sensor located in a regular house. The obtained results showed that detailed information of physical activity such as a gait cycle could be estimated using our proposed technique. The electrostatic induction current was also measured when the subject walked with the left ankle and left knee fastened to a splint with bandages to simulate a limp. The obtained results show that the degree of fastening of the foot is directly reflected in the generated electrostatic induction current. This technique can detect even subtle differences in the walking style. Therefore, the proposed technique based on the detection of the signal generated owing to walking motion can be successfully employed to not only estimate the gait cycle and number of steps but also study the quality of walking motion.

This research was partially supported by the Grant-in-Aid for Scientific Research (C), 26350056, 2014.