Indoor pedestrian navigation based on double-IMU framework

XU Yuan, CHEN Xi-yuan, LI Qing-hua, TANG Jian

(1. School of Electrical Engineering, University of Jinan, Jinan 250022, China; 2. School of Instrument Science and Engineering, Southeast University, Nanjing 210096, China; 3. Key Laboratory of Micro-Inertial Instrument and Advanced Navigation Technology Ministry of Education, China, Nanjing 210096, China)

Indoor pedestrian navigation based on double-IMU framework

XU Yuan1, CHEN Xi-yuan2,3, LI Qing-hua2,3, TANG Jian2,3

(1. School of Electrical Engineering, University of Jinan, Jinan 250022, China; 2. School of Instrument Science and Engineering, Southeast University, Nanjing 210096, China; 3. Key Laboratory of Micro-Inertial Instrument and Advanced Navigation Technology Ministry of Education, China, Nanjing 210096, China)

In order to realize low-cost indoor pedestrian navigation, a framework with two inertial measurement units (IMU) is proposed. In this framework, one IMU is fixed on the foot, and the other one is fixed on the shoulder. When pedestrian is at rest in the process of walking, the velocity error and angular velocity error from observation are used by Kalman filter to estimate the calculation error of foot-mounted IMU. Meanwhile, the yaw error observation is fulfilled by achieving the difference between the yaw measured from shoulder-mounted IMU and the one measured from foot-mounted IMU. In this architecture, the two-IMU framework is arranged in closed loop mode. Experimental results show that the proposed method can provide navigation information of pedestrians, and the position’s mean error is reduced by about 14.93% compared with that by the methods in open loop mode.

indoor pedestrian navigation; inertial navigation system; Kalman filter; foot-mounted IMU.

Nowadays, the Pedestrian Navigation (PN) in Global Positioning Systems (GPS) denied areas has received grate attention over the past few decades[1-2], especially in indoor environment. Although there are many beacon-based solutions have been investigated (Such as ultrasound, radio or vision)[3-5], the beacon-free solutions are preferable since they do not depend on a pre-installed infrastructure[6], and there are many attempts have been proposed. For example, Zhang J L et al. proposed shoe-mounted personal navigation system using MEMS inertial technology[7], Ali A et al. give the MEMS-based PN for GPS-denied areas in [8]. As one of the most widely used framework of PN, the foot-mounted Inertial Measurement Unit (IMU)-based framework (proposed by Foxlin E in [9]) employs Zero-velocity update (ZUPT) to reduce the drift error. In order to provide morerestrictions for IMU, there are many attempts have been proposed. For instance, Feng W et al. and Xie B et al. employ building layout for the yaw drift reduction[10-11], however, it requires that we should know the geographical features in advance. Jiménez A R et al. employ Zero Angular Rate Update (ZARU) Heuristic Heading Reduction (HDR) on the basis of ZUPT in [6], however, the foot-mounted magnetic sensors is poor in yaw calculation for PN due to its strong vibration during normal walking.

In this work, a two low-cost IMU-based framework for indoor pedestrian navigation is proposed. In this mode, one IMU is fixed on the foot, and the other one is fixed on the shoulder. When the person is in a stance phase, the Kalman filter (KF) is working with the observation value of velocity error and angular velocity error, meanwhile, the proposed framework is able to provide the yaw error with the shoulder-mounted IMU. Moreover, the proposed method employs closed loop-based framework. The remainder of the paper is organized as follows: Section 2 gives the principle of the two low-cost IMU-based indoor pedestrian navigation. The tests and discussion are illustrated in Section 3. Finally, the conclusions are given.

1. Principle of the two low-cost IMU-based indoor pedestrian navigation

1.1 Two low-cost IMU-based framework

The two IMU-based indoor pedestrian navigation framework is shown in Fig.1. In this work, two low cost IMU is used, one is fixed on the foot (called foot-mounted IMU), and the other is fixed on the shoulder (called shoulder-mounted IMU). The foot-mounted IMU is used to provide the position, velocity, and the attitude information, and the shoulder -mounted IMU is used to provide the yaw. In this mode, the accelerometer and gyroscope data are use to detect the phase of the person, when the person is in a stance phase, the Kalman filter (KF) is working. Since the velocity and the angular velocity should be zero when the person is in a stance phase, the velocity and the attitude calculated by the foot-mounted IMU is input to KF as the observation value of velocity error and angular velocity error. Moreover, the yaw measured from the shoulder-mounted IMU should be the same as the yaw measured from the foot-mounted IMU, thus, the difference between yaw measured from the foot-mounted IMU and that measured from the shoulder-mounted IMU is used as the observation value of yaw error. The KF is used for INS error estimation. Then, the INS solution is corrected by the error estimation. The coordinate frames used in this work is also shown in Fig.1, including the body frame (so called b-frame) and the navigation frame (so called n-frame).

1.2 Two low-cost IMU-based framework

The KF employs 15-element vector contains the errors in attitude, velocity, and position, the biases for accelerometer and gyroscope in n-frame. The state equation can be obtained as in Eq. (1),

is the acceleration in east, north, and downward respectively, Tis sampling period, ωkis the Gaussian white noise with zero mean, and its covariance matrix is Qk. The measurement equation for KF is shown in Eq. (2),

Fig.1 Two-IMU-based indoor pedestrian navigation framework

2 Tests and discussion

2.1 Indoor test environment

The test platform used in this work consists of two 9 DOF IMU. One is fixed on shone, and the other one is fixed on the shoulder. The IMU employs ADXL203, ADXRS620, and HMC5983 as accelerometer, gyroscope, and magnetometer respectively. Meanwhile, an encoder and an IMU are used to provide the reference trajectory. Here, the IMU (consists of MPU6050 and HMC5883) is used to provide the yaw, and encode is used for walking velocity measurement of the person. In this work, the data refresh rate of the computer is 50 Hz. The prototype of the test platform is shown in Fig.2.

Fig.2 Prototype of the test platform

2.2 Performance in real indoor environments

In this section, the performance for the proposed navigation method will be discussed. In this work, the person walks along the reference path (show in Fig.3) from start point to end point, the path length is about 78m. The yaw measured from foot-mounted IMU and that value measured from shoulder-mounted IMU are shown in Fig. 3. It can be seen that from the figure, the yaw measured from foot-mounted IMU has strong vibration compared with that from shoulder-mounted IMU.

The closed loop method is used in this work, thus, we compare the performance between the open loop method and the proposed method. The trajectories of two strategies are shown in Fig.4. It can be seen that both the strategies are able to provide the trajectory of the person, and the proposed method is closer to the reference path. Fig.5 and Fig.6 show the east and the north position errors of two strategies. And Tab. 1 shows the comparison of two strategies in terms of position error, from thetable we can see the proposed method is more effective to reduce the position drift. The root-mean-square error (RMSE) of east position is 2.0852 m, and the RMSE of north position is 0.8236 m. Comparing with the open loop method, the proposed method reduces the mean RMSE of position by about 14.93%.

Fig.3 Yaw measured from foot-mounted IMU and shoulder-mounted IMU

Fig.4 Trajectories of the real test

Fig.5 East position error of two strategies

Fig.6 North position error of two strategies

Tab.1 Comparison on two strategies in terms of position errors

3 Conclusions

In this work, a two low-cost IMU-based framework for indoor pedestrian navigation is proposed. In this mode, one IMU is fixed on the foot, and the other one is fixed on the shoulder. When the person is in a stance phase, the KF is working, and the observation value of velocity error and angular velocity error are input to the KF, meanwhile, the proposed framework is able to provide the yaw error with the shoulder-mounted IMU. Moreover, the proposed method employs closed loop-based framework. The experimental results show that the proposed method is effective to reduce the mean error of the position by about 14.93% compared with the open loop-based method.

[1] Ojeda L, Borenstein J. Non-GPS navigation for security personnel and first responders[J]. Journal of Navigation, 2007, 60(3): 391-407.

[2] Zhang Z, Meng X. Use of an inertial/magnetic sensor module for pedestrian tracking during normal walking[J]. IEEE Transactions on Instrumentation and Measurement, 2015, 64(3): 776-783.

[3] Nguyen V H, Pyun J Y. Location detection and tracking of moving targets by a 2D IR-UWB radar system[J]. Sensors, 2015, 15(3): 6740-62.

[4] Xu Y, Chen X Y, Li Q H. Autonomous integrated navigation for indoor robots utilizing on-line iterated extended Rauch-Tung-Striebel smoothing[J]. Sensors, 2013, 13(12): 15937-15953.

[5] Girard G, Cooté S, Zlatanova S, et al. Indoor pedestrian navigation using foot-mounted IMU and portable ultrasound range sensors[J]. Sensors, 2011, 11(8): 7606-7624.

[6] Jiménez A R, Seco F, Prieto J C, et al. Indoor pedestrian navigation using an INS/EKF framework for yaw drift reduction and a foot-mounted IMU[J]. Institute of Electrical & Electronics Engineers, 2010: 135 - 143.

[7] 張金亮, 秦永元, 梅春波. 基于 MEMS慣性技術的鞋式個人導航系統[J]. 中國慣性技術學報, 2011, 19(3): 253-256. Zhang J L, Qin Y Y, Mei C B. Shoe-mounted personal navigation system based on MEMS inertial technology[J]. Journal of Chinese Inertial Technology, 2011, 19(3): 253-256.

[8] Ali A, El-Sheimy N. Low-cost MEMS-based pedestrian navigation technique for GPS-denied areas[J]. Journal of Sensors, 2013(2013): 1-10.

[9] Foxlin E. Pedestrian tracking with shoe-mounted inertial sensors[J]. IEEE Computer Graphics and Applications, 2005, 25(6): 38-46.

[10] Feng W, Zhao H, Zhao Q, et al. Integration of GPS and low cost INS for pedestrian navigation aided by building layout[J]. Chinese Journal of Aeronautics, 2013, 26(5): 1283-1289.

[11] 謝波, 江一夫, 嚴恭敏, 等. 個人導航融合建筑平面信息的粒子濾波方法[J]. 中國慣性技術學報, 2013, 21(1), 1-6. Xie B, Jiang Y F, Yan G M, et al. Novel particle filter approach for fusing building plane into pedestrian navigation[J]. Journal of Chinese Inertial Technology, 2013, 21(1): 1-6.

為了實現低成本的室內行人導航，提出了一種雙慣性測量單元（IMU）框架。在這種模式下，一個IMU固定于足部，另一個IMU固定于肩部。當行人在行走過程中處于靜止狀態時，卡爾曼濾波器利用測量得到的速度和角速度誤差對足部IMU的解算誤差進行預估，與此同時，通過對足部IMU和肩部IMU測量得到的航向角做差完成對航向角誤差的觀測。在此基礎上，雙IMU框架結構采用了閉環模式。實驗結果顯示，采用該方法能夠提供行人導航信息，平均位置誤差與采用開環模式的方法相比降低了14.93%左右。

室內行人導航；慣性導航系統；卡爾曼濾波；足部慣性測量單元

TN967.3

A

2015-07-12；

：2015-09-18

國家自然科學基金資助項目（50975049，41204025，51375087）；山東省自然科學基金項目（ZR2014FP010）；濟南大學博士基金項目（XBS1503）

徐元（1985—），男，講師，博士，從事組合導航研究。E-mail: xy_abric@126.com

1005-6734(2015)06-0714-04

10.13695/j.cnki.12-1222/o3.2015.06.003

一種基于雙IMU框架的室內個人導航方法

徐 元1，陳熙源2,3，李慶華2,3，唐 建2,3

（1. 濟南大學 自動化與電氣工程學院，濟南 250022；2. 東南大學 儀器科學與工程學院，南京 210096；3. 東南大學 微慣性儀表與先進導航技術教育部重點實驗室，南京 210096）