| GISdevelopment.net ---> AARS ---> ACRS 2000 ---> Poster Session 1 |
Sensor integration for Personal Positioning System
Yusuke KONISHI, Ryosuke SHIBASAKI
Center for Spatial Information Science, Institute of Industrial Science,
University of Tokyo
4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan
Tel & Fax : (81)-3-5452-6417
E-mail:konishi@skl.iis.u-tokyo.ac.jp
Yusuke KONISHI, Ryosuke SHIBASAKI
Center for Spatial Information Science, Institute of Industrial Science,
University of Tokyo
4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan
Tel & Fax : (81)-3-5452-6417
E-mail:konishi@skl.iis.u-tokyo.ac.jp
Keywords:
positioning system, personal system, sensor, gyrocompass
Abstract
To realize monitoring system of human activity and transportation, or to provide the positional information services, it is necessary to develop the positioning technologies which can estimate user’s position and activities. As before, positioning systems using GPS or PHS were typical system as personal positioning system. However, the personal navigation systems using GPS or PHS need improvement in respect of availability, time-resolution and positioning accuracy. It is strongly demanded to develop personal positioning systems which can estimate one’s position continuously and with high time-resolution, especially in urban area where the positioning systems using GPS or PHS can’t work correctly. In this paper, we make a proposal of a personal positioning system which consists of some sensors.
1 Introduction
To realize monitoring system of human activity and transportation, or to provide the positional information services, it is necessary to develop the positioning technologies which can estimate user’s position and activities. Recently, many researches are being conducted on the personal positioning system, and many personal or hand-held positioning systems using GPS or PHS were being proposed. However, the personal navigation systems using GPS or PHS need improvement in respect of availability, time-resolution and positioning accuracy. GPS accuracy and reliability is limited in the area where the sufficient signal from GPS satellites can not be received, for example, in the valley between buildings, underground shopping malls, and so on. Similarly, the positioning systems using PHS can’t work correctly in the area where the signal from the base-stations can not be received, and the positioning accuracy is about dozens meters which is insufficient to use for personal positioning system. It is strongly demanded to develop the complementary positioning systems which can estimate one’s position continuously and accurately, especially in urban area where the positioning systems using GPS or PHS can’t work correctly.
In the case of car navigation, some complementary positioning systems were already de-veloped and reached practicable level. They are realized by counting number of wheel rotations, sensing steering angles, and matching positioning trajectory to map. On the other hand, there is no effective complementary positioning system for human under the present circumstances, although the development is strongly demanded.
In this paper, we make a proposal of a personal positioning system consists of some sensors which can complement GPS or PHS in urban areas.
2 Positioning System
2.1 Hardware
This system consists of some sensors and a note-PC. The outline of this system is illustrated in Figure 1.
Figure 1: Hardware
The sensor unit illustrated in Figure 1 consists of a pocket-size gyrocompass which can measure three-dimensional posture angle and acceleration (Japan Aviation Electronics, JIMS-30S), a magnetic sensor which can measure three-dimensional magnetic field (Honey-well, HMR2300), a digital barometer (Yoko-gawa M&C, MU101-AM1P), and a GPS receiver (Trimble, Lassen-SK8). The sensor unit is fixed to the lower back of a user, and signals are processed by the note-PC.
2.2 System
Sub-systems are constructed by the combinations of the sensors illustrated in 2.1. As subsystems, we propose Pedometer System, Direction Finder, Height Finder, and GPS. This personal positioning system is constructed by the effective combinations of these subsystems. The concept of this system is illustrated in Figure 2.
Figure 2: Concept of system
In developing this system, we focused the motion of walking human’s pelvis, and number of steps, length of a step, and walking direction are listed as important parameters of human’s walking. Therefore, we construct the Pedometer System to estimate number of steps and length of a step, the Direction Finder to estimate walking direction, and the Height Finder to estimate vertical movement.
Finally, this personal positioning system is constructed by the combination of these complementary sub-systems and GPS which can estimate absolute position.
3 Sub Systems
3.1 Pedometer System
In developing this system, we focused the motion of walking human’s pelvis.
3.1.1 Algorithm
At first, we measured the motion of pelvis when user walks 6 meters in 10 steps by the gyrocompass which is fixed to the lower back of a user. The time-series data of rotation an-gles around each axes measured by the gyro-compass are illustrated in Figure 3. The axes are defined in Figure 1.
Figure 3: The motion of walking human’s pelvis measured by the gyrometer
The experimental data illustrated in Figure 3 shows that number of steps is corresponding to number of waves which appeared in time- series data of pitch angle. Therefore, number of steps can be estimated by the time-series data of pitch angle measured by the gyrocompass.
Additionally, we measured the motion of pelvis when an identical user walks 14 meters in 22 steps, 20 steps and 17 steps. The each time-series data of pitch angle are illustrated in Figure 4.
Figure 4: The time-series data of pitch angle whose forms change as number of steps
The relationship between width of changing pitch angle and length of a step which is com- puted by Figure 4 is illustrated in Figure 5.
The experimental data illustrated in Figure 5 shows that length of a step is computed by the time-series data of pitch angle measured by the gyrocompass.
Therefore important parameters of human’s walking which includes number of steps and length of a step can be computed by the time- series data of the motion of pelvis measured by the gyrocompass.
Figure 5: The relationship between width of changing pitch angle and length of step
3.1.2 Experiment
To prove applicability of this system, we ex-amined how accurately this system can com-pute the walking distance. Each four subjects walks 15 meters in three different length of a step, and each walking distances are estimated by the Pedometer System. In this experiment, we use the same parameters to estimate length of a step from width of changing pitch angle. The experimental results is illustrated in Figure 6.
Figure 6: The relationship between number of steps and estimated walking distance
The experimental data illustrated in Figure 6 shows that the accuracy of estimated walking distance is different from each subjects. This means that we can improve the accuracy by set- ting up different parameters for each subjects.
3.2 Direction Finder
The time-series data of yaw angle measured by the gyrocompass has an important characteristic, gradual drift of the origin. The time- series data measured by the gyrocompass in a stationary state is illustrated in Figure 7.
Figure 7: The time-series data of yaw angle when the sensor is stable
Drift of the origin illustrated in Figure 7 is observed about 1 degree per minute as an average value, and about 3 degrees per minute as the biggest value. Because of this characteristic, estimate of direction by using only gyrocompass is very difficult.
The other hand, the magnetic sensor is so sensitive that the measured data is easily af fected by disturbance of magnetic field. The magnetic sensor which is adopted for this system can measure three-dimensional magnetic field by each x, y, and z axes’ magnetic sen sors. Offset of the origin and difference of sensitivity between each axes are occurred by the characteristics of each axes’ sensor and the disturbance of magnetic field around the magnetic sensor. The concept of these error is illustrated in Figure 8; this figure shows measured data of geomagnetic field when the magnetic sensor is rotated 360 degrees in horizontal plane.
To estimate the amount of offset of the origin and difference of sensitivity between each axes illustrated in Figure 8, we measured the magnetic field in 5 typical places of human activity in urban area; in an open space, beside a prefab, beside a ferroconcrete pole, in an elevator, and on an narrow corridor. In each places, we measured magnetic field when the magnetic sensor is rotated 360 degrees in horizontal plane. Off-sets of the origin computed by the measured data is illustrated in Figure 9, and difference of sensitivity between each axes computed by the measured data is illustrated in Table 1.
Figure 8: The characteristics of magnetic sen-sor
Figure 9: Offsets of the origin in each places Table 1: Differences of sensitivity in each places
Offset of the origin illustrated in Figure 9 and difference of sensitivity between each axes illustrated in Table 1 shows that there is disturbance of magnetic field in some specific places. In other words, the walking direction computed by the measured data of magnetic sensor has disturbance in some specific places where there is disturbance of magnetic field.
Then, we realize Direction Finder by combination of magnetic sensor and gyrocompass. The disturbance of estimated direction computed by measured data of magnetic sensor can be complemented by the time-series data of yaw angle measured by gyrocompass. This combination of magnetic sensor and gyrocompass can realize continuous and stable estimate of walking direction.
3.3 Height Finder
To monitor human activity in the underground shopping malls, buildings, and so on, the vertical movement is also important. However, the Pedometer System can estimate only horizontal movement, especially movement of walking. So, it is necessary to develop the Height Finder which can estimate vertical movement. In this system, we use barometer for estimate of vertical movement.
4 Conclusions
We proposed the Pedometer System for estimate of walking velocity, the Direction Finder for estimate of walking direction, and the Height Finder for estimate of vertical movement, which can complement GPS as personal positioning system. It is expected that this system can get one’s positional information continuously and in high time-resolution.
5 Future works
5.1 Sub-systems’ future works
We have some future works about each syb-systems.
5.1.1 Pedometer System
To improve the accuracy of estimated walking velocity, it is necessary to develop the algorithm to set up different parameters for each users. We are now developing some mechanism to set up different parameters for each users; the parameters are estimated by the time-series data measured by the sensors when user walks several times in the same distance, or the parameters are automatically improved while user using the positioning system.
5.1.2 Direction Finder
We have to develop the algorithm which can estimate walking direction by the combination of gyrocompass and magnetic sensor. It is also necessary to do many detailed examination to examine the characteristics of magnetic field in each places in urban area.
5.1.3 Height Finder
We have to develop the algorithm which can estimate vertical movement by using barometer. And, it is also necessary to develop the algorithm to estimate the horizontal movement by using estimated vertical movement, for example, movement by using an escalator.
5.2 Future works
At first, it is necessary to develop an algorithm which can automatically distinguish modes of activity, like walking, running, going up or down stairs, standing, sitting, and so on. And it is also important to develop an algorithm which can automatically identify the modes of transportation, like by elevator, escalator, car, bus, trains, and so on. If these algorithms are developed, and the positioning system can estimate one’s modes of activity and modes of transportation in real-time basis, the utility value of positional information will be more higher.
Next, it is necessary to combine this personal positioning system with map matching technology. It is expected that the combination is especially effective in buildings, underground shopping malls, and so on.
After discussing about these ideas, we will conduct detailed experiments to prove the accuracy and availability of this personal positioning system.
References
Abstract
To realize monitoring system of human activity and transportation, or to provide the positional information services, it is necessary to develop the positioning technologies which can estimate user’s position and activities. As before, positioning systems using GPS or PHS were typical system as personal positioning system. However, the personal navigation systems using GPS or PHS need improvement in respect of availability, time-resolution and positioning accuracy. It is strongly demanded to develop personal positioning systems which can estimate one’s position continuously and with high time-resolution, especially in urban area where the positioning systems using GPS or PHS can’t work correctly. In this paper, we make a proposal of a personal positioning system which consists of some sensors.
1 Introduction
To realize monitoring system of human activity and transportation, or to provide the positional information services, it is necessary to develop the positioning technologies which can estimate user’s position and activities. Recently, many researches are being conducted on the personal positioning system, and many personal or hand-held positioning systems using GPS or PHS were being proposed. However, the personal navigation systems using GPS or PHS need improvement in respect of availability, time-resolution and positioning accuracy. GPS accuracy and reliability is limited in the area where the sufficient signal from GPS satellites can not be received, for example, in the valley between buildings, underground shopping malls, and so on. Similarly, the positioning systems using PHS can’t work correctly in the area where the signal from the base-stations can not be received, and the positioning accuracy is about dozens meters which is insufficient to use for personal positioning system. It is strongly demanded to develop the complementary positioning systems which can estimate one’s position continuously and accurately, especially in urban area where the positioning systems using GPS or PHS can’t work correctly.
In the case of car navigation, some complementary positioning systems were already de-veloped and reached practicable level. They are realized by counting number of wheel rotations, sensing steering angles, and matching positioning trajectory to map. On the other hand, there is no effective complementary positioning system for human under the present circumstances, although the development is strongly demanded.
In this paper, we make a proposal of a personal positioning system consists of some sensors which can complement GPS or PHS in urban areas.
2 Positioning System
2.1 Hardware
This system consists of some sensors and a note-PC. The outline of this system is illustrated in Figure 1.
Figure 1: Hardware
The sensor unit illustrated in Figure 1 consists of a pocket-size gyrocompass which can measure three-dimensional posture angle and acceleration (Japan Aviation Electronics, JIMS-30S), a magnetic sensor which can measure three-dimensional magnetic field (Honey-well, HMR2300), a digital barometer (Yoko-gawa M&C, MU101-AM1P), and a GPS receiver (Trimble, Lassen-SK8). The sensor unit is fixed to the lower back of a user, and signals are processed by the note-PC.
2.2 System
Sub-systems are constructed by the combinations of the sensors illustrated in 2.1. As subsystems, we propose Pedometer System, Direction Finder, Height Finder, and GPS. This personal positioning system is constructed by the effective combinations of these subsystems. The concept of this system is illustrated in Figure 2.
Figure 2: Concept of system
In developing this system, we focused the motion of walking human’s pelvis, and number of steps, length of a step, and walking direction are listed as important parameters of human’s walking. Therefore, we construct the Pedometer System to estimate number of steps and length of a step, the Direction Finder to estimate walking direction, and the Height Finder to estimate vertical movement.
Finally, this personal positioning system is constructed by the combination of these complementary sub-systems and GPS which can estimate absolute position.
3 Sub Systems
3.1 Pedometer System
In developing this system, we focused the motion of walking human’s pelvis.
3.1.1 Algorithm
At first, we measured the motion of pelvis when user walks 6 meters in 10 steps by the gyrocompass which is fixed to the lower back of a user. The time-series data of rotation an-gles around each axes measured by the gyro-compass are illustrated in Figure 3. The axes are defined in Figure 1.
 |
 |
 |
Figure 3: The motion of walking human’s pelvis measured by the gyrometer
The experimental data illustrated in Figure 3 shows that number of steps is corresponding to number of waves which appeared in time- series data of pitch angle. Therefore, number of steps can be estimated by the time-series data of pitch angle measured by the gyrocompass.
Additionally, we measured the motion of pelvis when an identical user walks 14 meters in 22 steps, 20 steps and 17 steps. The each time-series data of pitch angle are illustrated in Figure 4.
 |
 |
 |
Figure 4: The time-series data of pitch angle whose forms change as number of steps
The relationship between width of changing pitch angle and length of a step which is com- puted by Figure 4 is illustrated in Figure 5.
The experimental data illustrated in Figure 5 shows that length of a step is computed by the time-series data of pitch angle measured by the gyrocompass.
Therefore important parameters of human’s walking which includes number of steps and length of a step can be computed by the time- series data of the motion of pelvis measured by the gyrocompass.
Figure 5: The relationship between width of changing pitch angle and length of step
3.1.2 Experiment
To prove applicability of this system, we ex-amined how accurately this system can com-pute the walking distance. Each four subjects walks 15 meters in three different length of a step, and each walking distances are estimated by the Pedometer System. In this experiment, we use the same parameters to estimate length of a step from width of changing pitch angle. The experimental results is illustrated in Figure 6.
Figure 6: The relationship between number of steps and estimated walking distance
The experimental data illustrated in Figure 6 shows that the accuracy of estimated walking distance is different from each subjects. This means that we can improve the accuracy by set- ting up different parameters for each subjects.
3.2 Direction Finder
The time-series data of yaw angle measured by the gyrocompass has an important characteristic, gradual drift of the origin. The time- series data measured by the gyrocompass in a stationary state is illustrated in Figure 7.
Figure 7: The time-series data of yaw angle when the sensor is stable
Drift of the origin illustrated in Figure 7 is observed about 1 degree per minute as an average value, and about 3 degrees per minute as the biggest value. Because of this characteristic, estimate of direction by using only gyrocompass is very difficult.
The other hand, the magnetic sensor is so sensitive that the measured data is easily af fected by disturbance of magnetic field. The magnetic sensor which is adopted for this system can measure three-dimensional magnetic field by each x, y, and z axes’ magnetic sen sors. Offset of the origin and difference of sensitivity between each axes are occurred by the characteristics of each axes’ sensor and the disturbance of magnetic field around the magnetic sensor. The concept of these error is illustrated in Figure 8; this figure shows measured data of geomagnetic field when the magnetic sensor is rotated 360 degrees in horizontal plane.
To estimate the amount of offset of the origin and difference of sensitivity between each axes illustrated in Figure 8, we measured the magnetic field in 5 typical places of human activity in urban area; in an open space, beside a prefab, beside a ferroconcrete pole, in an elevator, and on an narrow corridor. In each places, we measured magnetic field when the magnetic sensor is rotated 360 degrees in horizontal plane. Off-sets of the origin computed by the measured data is illustrated in Figure 9, and difference of sensitivity between each axes computed by the measured data is illustrated in Table 1.
| a. offset of the origin | b. difference of sensitivity |
 |
|
Figure 8: The characteristics of magnetic sen-sor
Figure 9: Offsets of the origin in each places Table 1: Differences of sensitivity in each places
| difference of sensibility (Byw/Bxw) |
|
| on an open space | 1.52 |
| beside a prefab | 1.50 |
| beside a ferroconcrete pole | 1.54 |
| in an elevator | 1.42 |
| on an narrow corridor | 1.74 |
Offset of the origin illustrated in Figure 9 and difference of sensitivity between each axes illustrated in Table 1 shows that there is disturbance of magnetic field in some specific places. In other words, the walking direction computed by the measured data of magnetic sensor has disturbance in some specific places where there is disturbance of magnetic field.
Then, we realize Direction Finder by combination of magnetic sensor and gyrocompass. The disturbance of estimated direction computed by measured data of magnetic sensor can be complemented by the time-series data of yaw angle measured by gyrocompass. This combination of magnetic sensor and gyrocompass can realize continuous and stable estimate of walking direction.
3.3 Height Finder
To monitor human activity in the underground shopping malls, buildings, and so on, the vertical movement is also important. However, the Pedometer System can estimate only horizontal movement, especially movement of walking. So, it is necessary to develop the Height Finder which can estimate vertical movement. In this system, we use barometer for estimate of vertical movement.
4 Conclusions
We proposed the Pedometer System for estimate of walking velocity, the Direction Finder for estimate of walking direction, and the Height Finder for estimate of vertical movement, which can complement GPS as personal positioning system. It is expected that this system can get one’s positional information continuously and in high time-resolution.
5 Future works
5.1 Sub-systems’ future works
We have some future works about each syb-systems.
5.1.1 Pedometer System
To improve the accuracy of estimated walking velocity, it is necessary to develop the algorithm to set up different parameters for each users. We are now developing some mechanism to set up different parameters for each users; the parameters are estimated by the time-series data measured by the sensors when user walks several times in the same distance, or the parameters are automatically improved while user using the positioning system.
5.1.2 Direction Finder
We have to develop the algorithm which can estimate walking direction by the combination of gyrocompass and magnetic sensor. It is also necessary to do many detailed examination to examine the characteristics of magnetic field in each places in urban area.
5.1.3 Height Finder
We have to develop the algorithm which can estimate vertical movement by using barometer. And, it is also necessary to develop the algorithm to estimate the horizontal movement by using estimated vertical movement, for example, movement by using an escalator.
5.2 Future works
At first, it is necessary to develop an algorithm which can automatically distinguish modes of activity, like walking, running, going up or down stairs, standing, sitting, and so on. And it is also important to develop an algorithm which can automatically identify the modes of transportation, like by elevator, escalator, car, bus, trains, and so on. If these algorithms are developed, and the positioning system can estimate one’s modes of activity and modes of transportation in real-time basis, the utility value of positional information will be more higher.
Next, it is necessary to combine this personal positioning system with map matching technology. It is expected that the combination is especially effective in buildings, underground shopping malls, and so on.
After discussing about these ideas, we will conduct detailed experiments to prove the accuracy and availability of this personal positioning system.
References
- Hideo KAWAI, Haruyoshi IWAMOTO and Kojii TORII, 1988. A Route Navigational Composite System of Pedometer and Azimuth Compass for Human. T.IEICE, J71- A-11, pp.2054-2062.
- Takayuki TAKAHASHI, Hirofumi OSAWA, Akihiro SUZUKI and Hikaru INOOKA, 1995. An Improved Algorithm and Implementation for the Ambulatory Measuring System of Human Gait. T.SICE, 32- 7, pp.1057-1064.
- Koichi SAGAWA, Atsushi INA, Takayuki TAKAHASHI, Tadashi ISHIHARA and Hikaru INOOKA, 1998. Estimation of Human Moving Behavior by Using Acceleration and Air Pressure. T.SICE, 35-2, pp.184-190.