WO2019061949A1 - Motion-behavior-assisted indoor fusion positioning method and apparatus and storage medium - Google Patents

Motion-behavior-assisted indoor fusion positioning method and apparatus and storage medium Download PDF

Info

Publication number
WO2019061949A1
WO2019061949A1 PCT/CN2018/072114 CN2018072114W WO2019061949A1 WO 2019061949 A1 WO2019061949 A1 WO 2019061949A1 CN 2018072114 W CN2018072114 W CN 2018072114W WO 2019061949 A1 WO2019061949 A1 WO 2019061949A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
behavior
pedestrian
motion
indoor
Prior art date
Application number
PCT/CN2018/072114
Other languages
French (fr)
Chinese (zh)
Inventor
周宝定
李清泉
朱家松
涂伟
Original Assignee
深圳大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳大学 filed Critical 深圳大学
Publication of WO2019061949A1 publication Critical patent/WO2019061949A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/04—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by terrestrial means
    • G01C21/08—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by terrestrial means involving use of the magnetic field of the earth
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/20—Instruments for performing navigational calculations
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/029—Location-based management or tracking services
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/33—Services specially adapted for particular environments, situations or purposes for indoor environments, e.g. buildings
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W64/00—Locating users or terminals or network equipment for network management purposes, e.g. mobility management

Abstract

The present invention provides a motion-behavior-assisted indoor fusion positioning method and apparatus and storage medium, said method comprising: on the basis of motion behavior of a smart terminal in multi-use mode, obtaining path-of-motion data of a pedestrian and wireless signal data received by a smartphone; parameterizing said path-of-motion data and wireless signal data, and constructing a multi-source data fusion model and obtaining an equation of the probability of transition between the continuous positions of a pedestrian; according to said multi-source data fusion model and transition probability equation, estimating the position at which the pedestrian is currently located. The multi-source data fusion model constructed by the present invention combines multi-source observation data in motion behavior recognition technology, reducing the degree of reliance on single-mode data and improving the accuracy of smart-terminal-based indoor positioning.

Description

一种运动行为辅助的室内融合定位方法及装置、存储介质Indoor fusion positioning method and device assisted by exercise behavior, storage medium 技术领域Technical field
本发明涉及室内定位领域，尤其涉及一种运动行为辅助的室内融合定位方法及装置、存储介质。The invention relates to the field of indoor positioning, in particular to an indoor fusion positioning method and device and a storage medium assisted by an activity behavior.
背景技术Background technique
近年来，室内定位成为位置服务领域的一大研究热点。基于智能手机实现室内定位无需用户携带额外设备，有利于室内定位技术的推广和普及。In recent years, indoor positioning has become a research hotspot in the field of location services. The indoor positioning based on the smart phone does not require the user to carry additional equipment, which is beneficial to the promotion and popularization of the indoor positioning technology.
智能手机内置多种传感器获取的多源数据均可用于室内定位。然而，受价格成本的约束，目前智能手机内置传感器的精度均不高，依靠单一传感器的定位方法(Wifi定位、惯性定位、蓝牙定位等)精度较差，无法满足室内定位的需求。随着智能手机内置传感器功能的日益增强，通过智能手机多传感器采集的数据可以识别行人的行为，通过运动行为识别可以推算相对运动轨迹以及关联智能手机获取的时序多源观测数据。另外，行人在室内环境中的行为包含了丰富的情景信息，受室内建筑结构的约束，行人在室内特殊位置会呈现不同的运动状态，产生与位置相关的行为，例如在乘电梯时会出现超重和失重状态。Multi-source data acquired by a variety of sensors built into the smartphone can be used for indoor positioning. However, due to the price and cost constraints, the accuracy of the built-in sensors of smart phones is not high at present, and the positioning method (Wifi positioning, inertial positioning, Bluetooth positioning, etc.) relying on a single sensor is inferior in accuracy, and cannot meet the needs of indoor positioning. With the increasing function of the built-in sensor of the smart phone, the data collected by the multi-sensor of the smart phone can recognize the behavior of the pedestrian, and the relative motion trajectory and the multi-source observation data acquired by the associated smart phone can be estimated through the motion behavior recognition. In addition, the behavior of pedestrians in the indoor environment contains rich situational information. Due to the constraints of the indoor building structure, pedestrians will present different movement states in special indoor locations, resulting in position-related behaviors, such as overweight when riding an elevator. And weightlessness.
室内场景复杂多样，具有不同的拓扑结构，且可能布设不同的室内定位信号源(WiFi路由器等)，具有不同的地磁场分布；智能手机用户具有不同的身体特征(身高、步长等)；不同的智能手机使用模式下(手持、打电话、摆动、口袋等)，手机处于不同的姿态，使得手机坐标系与行人坐标系存在航向偏差；智能手机设备多种多样，受硬件条件的影响，基于智能手机接收到的WiFi与地磁信号强度与设备相关。以上因素要求基于智能手机的多源数据融合室内定位方法具有很强的自适应性，需要适应不同信号环境的室内场景、不同用户、不同使用模式以及不同型号的手机设备。The indoor scenes are complex and diverse, with different topologies, and may have different indoor positioning signal sources (WiFi routers, etc.) with different geomagnetic field distributions; smartphone users have different physical characteristics (height, step size, etc.); different In the smart phone usage mode (handheld, phone, swing, pocket, etc.), the mobile phone is in different postures, which makes the mobile phone coordinate system and the pedestrian coordinate system have a heading deviation; the smart phone device is diverse and affected by hardware conditions, based on The WiFi and geomagnetic signal strength received by the smartphone is related to the device. The above factors require that the multi-source data fusion indoor positioning method based on smart phones has strong adaptability, and needs to adapt to indoor scenes of different signal environments, different users, different usage modes, and different types of mobile phone devices.
因此，现有技术还有待于改进和发展。Therefore, the prior art has yet to be improved and developed.
发明内容Summary of the invention
鉴于上述现有技术的不足之处，本发明的目的在于提供一种运动行为辅助的室内融合定位方法及装置、存储介质，旨在解决现有室内定位方法定位不准确的 问题。In view of the above-mentioned deficiencies of the prior art, an object of the present invention is to provide an indoor behavior locating method and apparatus for assisting motion behavior, and a storage medium, aiming at solving the problem of inaccurate positioning of the existing indoor positioning method.
为了达到上述目的，本发明采取了以下技术方案：In order to achieve the above object, the present invention adopts the following technical solutions:
一种运动行为辅助的室内融合定位方法，其中，包括步骤：An exercise behavior-assisted indoor fusion positioning method, comprising the steps of:
A、基于智能终端在多使用模式下的运动行为获取行人的运动轨迹数据以及智能终端接收到的无线信号数据；A. Acquiring the motion track data of the pedestrian and the wireless signal data received by the intelligent terminal based on the motion behavior of the smart terminal in the multi-use mode;
B、对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式；B. Parametrically expressing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model and obtaining a transition probability formula of pedestrians between consecutive positions;
C、根据所述多源数据融合模型以及转移概率公式推算行人当前所处位置。C. Calculate the current location of the pedestrian according to the multi-source data fusion model and the transition probability formula.
所述运动行为辅助的室内融合定位方法，其中，所述无线信号数据包括Wifi信号数据和地磁信号数据。The motion behavior-assisted indoor fusion positioning method, wherein the wireless signal data includes Wifi signal data and geomagnetic signal data.
所述运动行为辅助的室内融合定位方法，其中，所述步骤B具体包括：The motion behavior-assisted indoor fusion positioning method, wherein the step B specifically includes:
B1、结合运动轨迹数据与室内地图模型计算得到行为识别的特征函数；B1, combined with the motion trajectory data and the indoor map model to calculate the feature function of the behavior recognition;
B2、计算Wifi信号数据与位置指纹数据库中各个位置信号强度之间的欧式距离，通过欧氏距离的倒数归一化得到Wifi定位的特征函数；B2. Calculating the Euclidean distance between the Wifi signal data and the signal strength of each position in the location fingerprint database, and obtaining the eigenfunction of the Wifi positioning by reciprocal normalization of the Euclidean distance;
B3、计算地磁信号数据与地磁数据库中各个位置信号强度之间的欧式距离，通过欧式距离的倒数归一化得到地磁定位的特征函数；B3. Calculating the Euclidean distance between the geomagnetic signal data and the signal intensity of each position in the geomagnetic database, and obtaining a characteristic function of the geomagnetic positioning by reciprocal normalization of the Euclidean distance;
B4、采用条件随机场模型融合所述行为识别特征函数、Wifi定位特征函数以及地磁定位特征函数得到多源数据融合模型公式，
Figure PCTCN2018072114-appb-000001
其中，S k表示k时刻行人的室内位置，
Figure PCTCN2018072114-appb-000002
表示k时刻的第n种多源观测数据，
Figure PCTCN2018072114-appb-000003
表示k时刻在S k位置观测到第n种观测数据的概率，λ i为特征函数
Figure PCTCN2018072114-appb-000004
的权值参数； B4. Using the conditional random field model to fuse the behavior recognition feature function, the Wifi positioning feature function, and the geomagnetic localization feature function to obtain a multi-source data fusion model formula,
Figure PCTCN2018072114-appb-000001
Where S k represents the indoor position of the pedestrian at time k,
Figure PCTCN2018072114-appb-000002
Indicates the nth multi-source observation data at time k,
Figure PCTCN2018072114-appb-000003
Indicates the probability that the nth observation data is observed at the S k position at time k , and λ i is the characteristic function
Figure PCTCN2018072114-appb-000004
Weight parameter
B5、根据运动轨迹数据获得行人在连续位置之间的转移概率
Figure PCTCN2018072114-appb-000005
其中，d表示观测距离，θ表示观测距离，σ d和σ θ分别为方向和距离估计的标准差。 B5. Obtaining the transition probability of pedestrians between consecutive positions according to the motion trajectory data
Figure PCTCN2018072114-appb-000005
Where d is the observed distance, θ is the observed distance, and σ d and σ θ are the standard deviations of the direction and distance estimates, respectively.
所述运动行为辅助的室内融合定位方法，其中，所述步骤C具体包括：The motion behavior-assisted indoor fusion positioning method, wherein the step C specifically includes:
根据所述多源数据融合模型公式以及转移概率公式获得推算行人当前所处位置的公式：P(S k)＝P(S k-1)·P(S k|S k-1)·P(S k|Z k)，其中，P(S k|Z k)表示基于观测数据经过多源数据融合模型公式计算出的当前状态的归一化概率值。 Obtaining a formula for estimating the current position of the pedestrian according to the multi-source data fusion model formula and the transition probability formula: P(S k )=P(S k-1 )·P(S k |S k-1 )·P( S k |Z k ), where P(S k |Z k ) represents a normalized probability value of the current state calculated based on the observed data through the multi-source data fusion model formula.
所述运动行为辅助的室内融合定位方法，其中，所述智能终端包括智能手机、智能平板、智能手环或智能手表。The motion behavior-assisted indoor fusion positioning method, wherein the smart terminal comprises a smart phone, a smart tablet, a smart bracelet or a smart watch.
一种运动行为辅助的室内融合定位装置，其中，包括：An activity-assisted indoor fusion positioning device, which comprises:
处理器，适于实现各指令；以及a processor adapted to implement each instruction;
存储设备，适于存储多条指令，所述指令适于由处理器加载并执行以下步骤：A storage device adapted to store a plurality of instructions adapted to be loaded by the processor and to perform the following steps:
基于智能终端在多使用模式下的运动行为获取行人的运动轨迹数据以及智能手机接收到的无线信号数据；Acquiring the motion track data of the pedestrian and the wireless signal data received by the smart phone based on the motion behavior of the smart terminal in the multi-use mode;
对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式；Parametrically expressing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model and obtaining a transition probability formula of pedestrians between consecutive positions;
根据所述多源数据融合模型以及转移概率公式推算行人当前所处位置。The current location of the pedestrian is estimated according to the multi-source data fusion model and the transition probability formula.
所述在线目标空间划分装置，其中，所述无线信号数据包括Wifi信号数据和地磁信号数据。The online target space dividing device, wherein the wireless signal data includes Wifi signal data and geomagnetic signal data.
所述在线目标空间划分装置，其中，对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式的步骤具体包括：The online target space dividing device, wherein the step of parameterizing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model, and obtaining a transition probability formula between pedestrians in consecutive positions comprises:
结合运动轨迹数据与室内地图模型计算得到行为识别的特征函数；The feature function of behavior recognition is calculated by combining the motion trajectory data with the indoor map model;
计算Wifi信号数据与位置指纹数据库中各个位置信号强度之间的欧式距离，通过欧氏距离的倒数归一化得到Wifi定位的特征函数；Calculating the Euclidean distance between the Wifi signal data and the signal intensity of each position in the location fingerprint database, and obtaining the eigenfunction of the Wifi positioning by reciprocal normalization of the Euclidean distance;
计算地磁信号数据与地磁数据库中各个位置信号强度之间的欧式距离，通过欧式距离的倒数归一化得到地磁定位的特征函数；Calculating the Euclidean distance between the geomagnetic signal data and the signal intensity of each position in the geomagnetic database, and obtaining the characteristic function of the geomagnetic positioning by reciprocal normalization of the Euclidean distance;
采用条件随机场模型融合所述行为识别特征函数、Wifi定位特征函数以及地磁定位特征函数得到多源数据融合模型公式
Figure PCTCN2018072114-appb-000006
其中，S k表示k时刻行人的室内位置，
Figure PCTCN2018072114-appb-000007
表示k时刻的第n种多源观测数据，
Figure PCTCN2018072114-appb-000008
表示k时刻在S k位置观测到第n种观测数据的概率，λ i为特征函数
Figure PCTCN2018072114-appb-000009
的权值参数； The multi-source data fusion model formula is obtained by using the conditional random field model to fuse the behavior recognition feature function, Wifi positioning feature function and geomagnetic localization feature function.
Figure PCTCN2018072114-appb-000006
Where S k represents the indoor position of the pedestrian at time k,
Figure PCTCN2018072114-appb-000007
Indicates the nth multi-source observation data at time k,
Figure PCTCN2018072114-appb-000008
Indicates the probability that the nth observation data is observed at the S k position at time k , and λ i is the characteristic function
Figure PCTCN2018072114-appb-000009
Weight parameter
根据运动轨迹数据获得行人在连续位置之间的转移概率
Figure PCTCN2018072114-appb-000010
其中，d表示观测距离，θ表示观测距离，σ d和σ θ分别为方向和距离估计的标准差。 Obtaining the transition probability of pedestrians between consecutive positions based on the motion trajectory data
Figure PCTCN2018072114-appb-000010
Where d is the observed distance, θ is the observed distance, and σ d and σ θ are the standard deviations of the direction and distance estimates, respectively.
所述在线目标空间划分装置，其中，根据所述多源数据融合模型公式以及转移概率公式获得推算行人当前所处位置的公式：P(S k)＝P(S k-1)·P(S k|S k-1)·P(S k|Z k)，其中，P(S k|Z k)表示基于观测数据经过多源数据融合模型公式计算出的当前状态的归一化概率值。 The online target space dividing device, wherein a formula for estimating a current position of a pedestrian is obtained according to the multi-source data fusion model formula and a transition probability formula: P(S k )=P(S k-1 )·P(S k |S k-1 )·P(S k |Z k ), where P(S k |Z k ) represents a normalized probability value of the current state calculated based on the observed data through the multi-source data fusion model formula.
一种存储介质，其中，存储有多条指令，所述指令适于由处理器加载并执行上述任一项所述运动行为辅助的室内融合定位方法的步骤。A storage medium, wherein a plurality of instructions are stored, the instructions being adapted to be loaded by a processor and to perform the steps of the indoor fusion positioning method assisted by the motion behavior of any of the above.
有益效果：本发明提供的一种运动行为辅助的室内融合定位方法，通过研究智能终端在多使用模式下的运动行为识别，获取行人的运动轨迹数据以及智能终端接收到的无线信号数据，基于所述运动轨迹数据以及无线信号数据构建面向室内定位的多源数据融合模型，所述模型在运动行为识别的技术上融合多源观测数据，降低了对单模数据的依赖程度，提高了基于智能终端室内定位的精度。Advantageous Effects: The present invention provides an indoor behavior locating method for assisting an activity behavior, and by acquiring motion behavior recognition of a smart terminal in a multi-use mode, acquiring trajectory data of a pedestrian and wireless signal data received by the intelligent terminal, based on The motion trajectory data and the wireless signal data construct a multi-source data fusion model for indoor positioning, which integrates multi-source observation data in the technique of motion behavior recognition, reduces the dependence on single-mode data, and improves the intelligent terminal-based The accuracy of indoor positioning.
附图说明DRAWINGS
图1为本发明一种运动型为辅助的室内融合定位方法较佳实施例的流程图。1 is a flow chart of a preferred embodiment of a motion-assisted indoor fusion positioning method according to the present invention.
图2为本发明一种面向室内定位的多源数据融合模型构建示意图。2 is a schematic diagram of a multi-source data fusion model for indoor positioning according to the present invention.
图3为本发明一种运动型为辅助的室内融合定位装置较佳实施例的结构框图。3 is a structural block diagram of a preferred embodiment of a motion-assisted indoor fusion positioning device according to the present invention.
具体实施方式Detailed ways
本发明提供一种运动行为辅助的室内融合定位方法及装置、存储介质，为使本发明的目的、技术方案及效果更加清楚、明确，以下参照附图并举实施例对本 发明进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本发明，并不用于限定本发明。The present invention provides a method and a device for locating a motion-assisted indoor fusion, and a storage medium. The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
请参阅图1，图1是本发明提供的一种运动行为辅助的室内融合定位方法较佳实施例的流程图。如图1所示，其中，包括以下步骤：Please refer to FIG. 1. FIG. 1 is a flow chart of a preferred embodiment of a motion behavior assisted indoor fusion positioning method provided by the present invention. As shown in Figure 1, the following steps are included:
S10、基于智能终端在多使用模式下的运动行为获取行人的运动轨迹数据以及智能终端接收到的无线信号数据；S10. Acquire motion trajectory data of the pedestrian and wireless signal data received by the intelligent terminal based on the motion behavior of the smart terminal in the multi-use mode;
S20、对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式；S20, parameterizing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model, and obtaining a transition probability formula of the pedestrian between successive positions;
S30、根据所述多源数据融合模型以及转移概率公式推算行人当前所处位置。S30. Calculate a current location of the pedestrian according to the multi-source data fusion model and the transition probability formula.
本实施中，首先，分析智能终端在不同使用模式下，不同行为产生的传感器信号，提取不同使用模式下各个行为的特征；其中，智能终端的使用模式包括手持模式、打电话模式、摆动模式以及口袋模式等，所述传感器包括加速度传感器、角速度传感器、气压传感器等；通过大量的样本训练得到分类器参数，最后基于行为特征，采用机器学习方法实现行为分类，位置行为用于行为地图匹配，行走行为用于推算运动轨迹。In this implementation, firstly, the sensor signals generated by different actions of the smart terminal in different usage modes are analyzed, and the characteristics of each behavior in different usage modes are extracted; wherein the usage modes of the smart terminal include a handheld mode, a call mode, a swing mode, and In the pocket mode, the sensor includes an acceleration sensor, an angular velocity sensor, a barometric pressure sensor, etc.; the classifier parameters are obtained through a large number of sample training, and finally, based on the behavior characteristics, a machine learning method is used to implement behavior classification, and the position behavior is used for behavior map matching, walking. Behavior is used to estimate the motion trajectory.
具体来说，运动轨迹数据是通过行人航位推算的方法获得的，行人航位推算有三部分组成：步子检测、步长估计和航向估计，其中步子检测和步长估计用于行走距离估算。行人在行走过程中产生的加速度信号具有周期性特征，本发明采用峰值检测算法实现步子检测。步长估计采用步频步长模型，同时考虑行人的身高因素：l t＝h(αf t+β)，其中l t为步长估计值，h为用户身高，f t为步频，(α,β)为未知参数，不同用户具有不同的步长参数，初始定位时使用经验值，之后通过模型参数自适应学习方法实时调整。 Specifically, the trajectory data is obtained by the method of pedestrian dead reckoning. The pedestrian dead reckoning has three parts: step detection, step estimation and heading estimation. Step detection and step estimation are used for walking distance estimation. The acceleration signal generated by the pedestrian during walking has a periodic characteristic, and the present invention uses the peak detection algorithm to implement the step detection. The step size estimation uses the step frequency step model, taking into account the pedestrian height factor: l t =h(αf t +β), where l t is the step size estimate, h is the user's height, and f t is the step frequency, (α , β) is an unknown parameter, different users have different step parameters, the empirical value is used in the initial positioning, and then adjusted in real time by the model parameter adaptive learning method.
进一步地，由于不同使用模式下，智能终端处于不同的姿态，使得智能终端坐标系与行人坐标系之间存在航向偏差，初始定位时使用经验值设置航向偏差，之后通过模型参数自适应学习方法，实时调整航向偏差。航向变化通过磁力计和陀螺仪的数据计算，其中磁力计输出航向角，陀螺仪输出角速度的变化值。为了提高航向估计的精确度，拟采用卡尔曼滤波方法融合磁力计和陀螺仪的数据。Further, since the smart terminal is in different postures in different usage modes, there is a heading deviation between the intelligent terminal coordinate system and the pedestrian coordinate system, and the heading deviation is set using the empirical value in the initial positioning, and then the model parameter adaptive learning method is adopted. Adjust the heading deviation in real time. The heading change is calculated by the data of the magnetometer and the gyroscope, wherein the magnetometer outputs the heading angle and the gyroscope outputs the angular velocity. In order to improve the accuracy of heading estimation, Kalman filtering method is proposed to fuse the data of magnetometer and gyroscope.
更进一步地，在本发明中，所述步骤S20、对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式，具体包括：Further, in the present invention, the step S20, parameterizing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model and obtaining a transition probability formula of the pedestrian between consecutive positions, specifically including :
S21、结合运动轨迹数据与室内地图模型计算得到行为识别的特征函数；S21, combining the motion trajectory data and the indoor map model to calculate a feature function of the behavior recognition;
S22、计算Wifi信号数据与位置指纹数据库中各个位置信号强度之间的欧式距离，通过欧氏距离的倒数归一化得到Wifi定位的特征函数；S22. Calculate an Euclidean distance between the Wifi signal data and the signal strength of each position in the location fingerprint database, and obtain a WLAN positioning characteristic function by reciprocal normalization of the Euclidean distance;
S23、计算地磁信号数据与地磁数据库中各个位置信号强度之间的欧式距离，通过欧式距离的倒数归一化得到地磁定位的特征函数；S23. Calculating a Euclidean distance between the geomagnetic signal data and the signal strength of each position in the geomagnetic database, and obtaining a characteristic function of the geomagnetic positioning by reciprocal normalization of the Euclidean distance;
S24、采用条件随机场模型融合所述行为识别特征函数、Wifi定位特征函数以及地磁定位特征函数得到多源数据融合模型公式，
Figure PCTCN2018072114-appb-000011
其中，S k表示k时刻行人的室内位置，
Figure PCTCN2018072114-appb-000012
表示k时刻的第n种多源观测数据，
Figure PCTCN2018072114-appb-000013
表示k时刻在S k位置观测到第n种观测数据的概率，λ i为特征函数
Figure PCTCN2018072114-appb-000014
的权值参数； S24. Using the conditional random field model to fuse the behavior recognition feature function, the Wifi positioning feature function, and the geomagnetic positioning feature function to obtain a multi-source data fusion model formula,
Figure PCTCN2018072114-appb-000011
Where S k represents the indoor position of the pedestrian at time k,
Figure PCTCN2018072114-appb-000012
Indicates the nth multi-source observation data at time k,
Figure PCTCN2018072114-appb-000013
Indicates the probability that the nth observation data is observed at the S k position at time k , and λ i is the characteristic function
Figure PCTCN2018072114-appb-000014
Weight parameter
S25、根据运动轨迹数据获得行人在连续位置之间的转移概率
Figure PCTCN2018072114-appb-000015
其中，d表示观测距离，θ表示观测距离，σ d和σ θ分别为方向和距离估计的标准差。 S25. Obtain a transition probability of pedestrians between consecutive positions according to motion trajectory data.
Figure PCTCN2018072114-appb-000015
Where d is the observed distance, θ is the observed distance, and σ d and σ θ are the standard deviations of the direction and distance estimates, respectively.
示例性地，所述基于多使用模式下的运动行为识别可以获得行人的运动轨迹数据，并且通过智能终端还可以接收用户室内定位的无线信号数据，所述无线信号数据包括Wifi信号数据和地磁信号数据。Exemplarily, the motion behavior recognition based on the multi-use mode can obtain the motion track data of the pedestrian, and the wireless terminal can also receive the wireless signal data of the indoor location of the user, where the wireless signal data includes the Wifi signal data and the geomagnetic signal. data.
具体来说，行人的运动行为可以描述成一个隐马尔科夫过程，采用隐马尔科夫因子图对观测因子进行参数化表达，构建各观测数据的特征函数。在本发明中，观测因子包括运动轨迹数据、Wifi定位和地磁定位。Specifically, the pedestrian's motor behavior can be described as a hidden Markov process. The hidden Markov factor graph is used to parameterize the observation factors and construct the characteristic function of each observation data. In the present invention, the observation factors include motion trajectory data, Wifi positioning, and geomagnetic positioning.
将室内地图进行建模，得到室内地图的“点线”模型，即室内地图模型，其中“点”为可能发生特殊行为的位置，例如，转角、电梯、楼梯等，“线”为连接“点”之间的边。如图2所示，结合运动轨迹数据与室内地图模型计算得到行 为识别的特征函数，表示为
Figure PCTCN2018072114-appb-000016
Wifi定位基于位置指纹法实现，即通过接收的Wifi信号数据估计行人处于当前位置的概率，首先计算Wifi信号强度与位置指纹数据库中各个位置信号强度之间的欧氏距离，然后通过欧氏距离的倒数归一化得到Wifi定位的特征函数，表示为
Figure PCTCN2018072114-appb-000017
类似于Wifi定位，地磁定位通过计算地磁信号数据与地磁数据库中各个位置信号强度之间的欧式距离，通过欧式距离的倒数归一化得到地磁定位的特征函数，表示为
Figure PCTCN2018072114-appb-000018
The indoor map is modeled to obtain the "dotted line" model of the indoor map, that is, the indoor map model, wherein the "point" is a position where special behavior may occur, for example, a corner, an elevator, a staircase, etc., and the "line" is a connection "point""Between the sides. As shown in Fig. 2, the feature function of behavior recognition is calculated by combining the motion trajectory data with the indoor map model, expressed as
Figure PCTCN2018072114-appb-000016
The Wifi positioning is implemented based on the location fingerprint method, that is, the probability of the pedestrian being in the current position is estimated by the received Wifi signal data, and the Euclidean distance between the Wifi signal strength and the signal intensity of each position in the location fingerprint database is first calculated, and then the Euclidean distance is passed. The reciprocal normalization gives the Wifi positioning feature function, expressed as
Figure PCTCN2018072114-appb-000017
Similar to Wifi positioning, the geomagnetic positioning obtains the characteristic function of geomagnetic positioning by calculating the Euclidean distance between the geomagnetic signal data and the signal intensity of each position in the geomagnetic database, and is represented by the reciprocal normalization of the Euclidean distance.
Figure PCTCN2018072114-appb-000018
在单模观测数据特征函数基础上，采用条件随机场模型融合所述行为识别特征函数、Wifi定位特征函数以及地磁定位特征函数得到最初多源数据融合模型公式
Figure PCTCN2018072114-appb-000019
然后根据贝叶斯原理P(Z k|S k)＝P(S k|Z k)·P(Z k)/P(S k)，得到最终的多源数据融合模型公式为
Figure PCTCN2018072114-appb-000020
其中，S 0,S 1,...,S k分别表示行人0,1,…,k时刻的室内位置；
Figure PCTCN2018072114-appb-000021
分别表示k时刻的n种多源观测数据；
Figure PCTCN2018072114-appb-000022
表示k时刻在S k位置观测到第n种观测数据的概率，通过该观测数据的特征函数计算得到。 Based on the characteristic function of single-mode observation data, the conditional random field model is used to fuse the behavior recognition feature function, Wifi positioning feature function and geomagnetic localization feature function to obtain the initial multi-source data fusion model formula.
Figure PCTCN2018072114-appb-000019
Then according to the Bayesian principle P(Z k |S k )=P(S k |Z k )·P(Z k )/P(S k ), the final multi-source data fusion model formula is obtained as
Figure PCTCN2018072114-appb-000020
Wherein, S 0 , S 1 , ..., S k respectively represent indoor positions at the time of pedestrian 0, 1, ..., k;
Figure PCTCN2018072114-appb-000021
Representing n kinds of multi-source observation data at time k;
Figure PCTCN2018072114-appb-000022
The probability of observing the nth observation data at the S k position at time k is calculated and calculated by the characteristic function of the observation data.
基于多源观测数据，根据所述多源数据融合模型公式计算行人处于每个位置的概率P(S k|Z k)，然后通过运动行为识别推算运动轨迹，得到行人在连续位置之间的转移概率P(S k|S k-1)，由于行人航位推算的观测结果包括距离观测和角度观测两部分，两者相互独立且可以通过正态高斯分布描述，所述状态转移概率公式为
Figure PCTCN2018072114-appb-000023
其中，d表示观测距离，θ表示观测距离，σ d和σ θ分别为方向和距离估计的标准差。 Based on the multi-source observation data, the probability P(S k |Z k ) of the pedestrian at each position is calculated according to the multi-source data fusion model formula, and then the motion trajectory is estimated by the motion behavior recognition to obtain the pedestrian's transfer between successive positions. Probability P(S k |S k-1 ), since the observation results of the pedestrian dead reckoning include distance observation and angle observation, the two are independent of each other and can be described by a normal Gaussian distribution, and the state transition probability formula is
Figure PCTCN2018072114-appb-000023
Where d is the observed distance, θ is the observed distance, and σ d and σ θ are the standard deviations of the direction and distance estimates, respectively.
进一步地，本发明所述步骤S30、根据所述多源数据融合模型以及转移概率公式推算行人当前所处位置，具体包括：Further, in the step S30 of the present invention, the current location of the pedestrian is estimated according to the multi-source data fusion model and the transition probability formula, and specifically includes:
室内定位的过程即在多源数据融合模型得到的特征函数和行人行为推算得到的状态转移概率基础上，推算行人当前所处的位置。具体来说，先根据观测数据估计行人的初始位置S 0，然后根据特征函数计算其概率P(S 0)，最后，根据状态转移概率公式推算行人的下一个位置，位置推理公式为：P(S k)＝P(S k-1)·P(S k|S k-1)·P(S k|Z k)，其中，P(S k|Z k)表示基于观测数据经过多源数据融合模型公式计算出的当前状态的归一化概率值。 The process of indoor positioning is based on the feature function obtained by the multi-source data fusion model and the state transition probability derived from the pedestrian behavior estimation, and the current position of the pedestrian is estimated. Specifically, the pedestrian initial position S 0 is estimated based on the observation data, and then the probability P(S 0 ) is calculated according to the feature function. Finally, the next position of the pedestrian is estimated according to the state transition probability formula, and the position reasoning formula is: P ( S k )=P(S k-1 )·P(S k |S k-1 )·P(S k |Z k ), where P(S k |Z k ) represents multi-source data passing through the observation data The normalized probability value of the current state calculated by the fusion model formula.
更进一步地，由于观测数据存在误差，通过单一时刻的观测数据无法得到精确的初始位置。位置初始化时，根据多组初始位置及其概率，之后基于多组初始位置推理行人轨迹，计算每条轨迹的概率，以后验概率最大的状态为当前的估计结果。在位置推理过程中，首先使用经验模型参数，之后基于定位过程中产生的运动轨迹及观测数据，采用自适应参数学习方法，得到实时动态的模型参数。Further, due to the error in the observed data, an accurate initial position cannot be obtained by the observation data at a single moment. When the position is initialized, the probability of each trajectory is calculated according to the multiple sets of initial positions and their probabilities, and then the trajectories of each trajectory are calculated based on the plurality of sets of initial positions, and the state with the largest posterior probability is the current estimation result. In the location reasoning process, the empirical model parameters are first used, and then based on the motion trajectories and observation data generated during the positioning process, adaptive parameter learning methods are used to obtain real-time dynamic model parameters.
具体来说，在基于单模感知结果得到特征函数时，由于各个感知结果都包含着多个观测模型参数，包括智能终端坐标系和行人坐标系的航向偏差、Wifi和地磁观测误差、步长估计误差，另外，所构建多源数据融合模型中的单个特征函数的权值参数λ i也是模型数据之一。这些模型参数很难通过人工进行调整，因为它们与智能终端使用模式、设备、用户以及环境均有关系，因此无法使用统一的模型参数，需要自适应调整。 Specifically, when the feature function is obtained based on the single-mode perceptual result, each of the perceptual results includes multiple observation model parameters, including heading deviation of the intelligent terminal coordinate system and the pedestrian coordinate system, Wifi and geomagnetic observation error, and step estimation. Error, in addition, the weight parameter λ i of a single feature function in the constructed multi-source data fusion model is also one of the model data. These model parameters are difficult to adjust manually because they are related to smart terminal usage patterns, devices, users, and the environment, so unified model parameters cannot be used and adaptive adjustments are required.
多源数据融合模型的参数自适应学习方法如下所示：首先基于室内地图建模得到的“点－线”模型和运动行为识别得到的位置行为和位置行为之间的相对轨迹，采用行为地图匹配的方法推算用户的历史轨迹，以此作为参数学习的参考轨迹。由于室内环境中有且仅有一条轨迹与智能终端观测的多源数据得到的轨迹相匹配，只有模型参数适应当前环境、用户、使用模式和设备时，该轨迹和室内匹配的概率最大。The parameter adaptive learning method of multi-source data fusion model is as follows: Firstly, based on the "point-line" model obtained by indoor map modeling and the relative trajectory between positional behavior and positional behavior obtained by motion behavior recognition, behavior map matching is adopted. The method estimates the user's historical trajectory as a reference trajectory for parameter learning. Since there is only one trajectory in the indoor environment that matches the trajectory obtained by the multi-source data observed by the intelligent terminal, the probability of matching the trajectory and the indoor is the greatest when the model parameters are adapted to the current environment, the user, the usage mode, and the device.
基于这样的前提，模型参数学习问题变成了数据优化问题，假设历史观测数据得到的轨迹为Z，通过行为地图匹配得到的轨迹为Y，模型参数为X *，则构建目标方程为：X *＝argmaxln(p(Y|Z))，优化问题的原理是只有当模型参数与真实值相同时，基于观测数据得到的匹配轨迹的概率最大。优化问题的求解采用 最大期望(Expectation Maximization,EM)算法实现。EM算法是一种启发式的迭代算法，可以用于实现对含有隐变量模型参数的极大似然估计，采用迭代逼近的方式求解模型内部参数。 Based on this premise, the model parameter learning problem becomes a data optimization problem. It is assumed that the trajectory obtained by historical observation data is Z, the trajectory obtained by behavior map matching is Y, and the model parameter is X * , then the target equation is constructed as: X * =argmaxln(p(Y|Z)), the principle of the optimization problem is that the probability of the matching trajectory based on the observed data is the largest when the model parameters are the same as the true values. The optimization problem is solved by the Expectation Maximization (EM) algorithm. The EM algorithm is a heuristic iterative algorithm, which can be used to implement the maximum likelihood estimation of the parameters with hidden variables, and iterative approximation is used to solve the internal parameters of the model.
较佳地，在本发明中，所述智能终端包括智能手机、智能平板、智能手环或智能手表等，但不限于此。Preferably, in the present invention, the smart terminal includes a smart phone, a smart tablet, a smart wristband or a smart watch, etc., but is not limited thereto.
基于上述方法，本发明还提供了一种运动行为辅助的室内融合定位装置，如图3所示，其中，包括处理器10，适于实现各指令；以及Based on the above method, the present invention also provides an indoor behavior locating device for assisting an exercise behavior, as shown in FIG. 3, wherein the processor 10 is adapted to implement each instruction;
存储设备20，适于存储多条指令，所述指令适于由处理器加载并执行以下步骤：The storage device 20 is adapted to store a plurality of instructions adapted to be loaded by the processor and to perform the following steps:
基于智能终端在多使用模式下的运动行为获取行人的运动轨迹数据以及智能手机接收到的无线信号数据；Acquiring the motion track data of the pedestrian and the wireless signal data received by the smart phone based on the motion behavior of the smart terminal in the multi-use mode;
对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式；Parametrically expressing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model and obtaining a transition probability formula of pedestrians between consecutive positions;
根据所述多源数据融合模型以及转移概率公式推算行人当前所处位置。The current location of the pedestrian is estimated according to the multi-source data fusion model and the transition probability formula.
所述在线目标空间划分装置，其中，所述无线信号数据包括Wifi信号数据和地磁信号数据。The online target space dividing device, wherein the wireless signal data includes Wifi signal data and geomagnetic signal data.
所述在线目标空间划分装置，其中，对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式的步骤具体包括：The online target space dividing device, wherein the step of parameterizing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model, and obtaining a transition probability formula between pedestrians in consecutive positions comprises:
结合运动轨迹数据与室内地图模型计算得到行为识别的特征函数；The feature function of behavior recognition is calculated by combining the motion trajectory data with the indoor map model;
计算Wifi信号数据与位置指纹数据库中各个位置信号强度之间的欧式距离，通过欧氏距离的倒数归一化得到Wifi定位的特征函数；Calculating the Euclidean distance between the Wifi signal data and the signal intensity of each position in the location fingerprint database, and obtaining the eigenfunction of the Wifi positioning by reciprocal normalization of the Euclidean distance;
计算地磁信号数据与地磁数据库中各个位置信号强度之间的欧式距离，通过欧式距离的倒数归一化得到地磁定位的特征函数；Calculating the Euclidean distance between the geomagnetic signal data and the signal intensity of each position in the geomagnetic database, and obtaining the characteristic function of the geomagnetic positioning by reciprocal normalization of the Euclidean distance;
采用条件随机场模型融合所述行为识别特征函数、Wifi定位特征函数以及地磁定位特征函数得到多源数据融合模型公式
Figure PCTCN2018072114-appb-000024
其中，S k表示k时刻行人的室内位置，
Figure PCTCN2018072114-appb-000025
表示k时刻的第n种多源观测数据，
Figure PCTCN2018072114-appb-000026
表示k时刻在S k位置观测到第n种观测数据的概率，λ i为特征函数
Figure PCTCN2018072114-appb-000027
的权值参数； The multi-source data fusion model formula is obtained by using the conditional random field model to fuse the behavior recognition feature function, Wifi positioning feature function and geomagnetic localization feature function.
Figure PCTCN2018072114-appb-000024
Where S k represents the indoor position of the pedestrian at time k,
Figure PCTCN2018072114-appb-000025
Indicates the nth multi-source observation data at time k,
Figure PCTCN2018072114-appb-000026
Indicates the probability that the nth observation data is observed at the S k position at time k , and λ i is the characteristic function
Figure PCTCN2018072114-appb-000027
Weight parameter
根据运动轨迹数据获得行人在连续位置之间的转移概率
Figure PCTCN2018072114-appb-000028
其中，d表示观测距离，θ表示观测距离，σ d和σ θ分别为方向和距离估计的标准差。 Obtaining the transition probability of pedestrians between consecutive positions based on the motion trajectory data
Figure PCTCN2018072114-appb-000028
Where d is the observed distance, θ is the observed distance, and σ d and σ θ are the standard deviations of the direction and distance estimates, respectively.
所述在线目标空间划分装置，其中，根据所述多源数据融合模型公式以及转移概率公式获得推算行人当前所处位置的公式：P(S k)＝P(S k-1)·P(S k|S k-1)·P(S k|Z k)表示基于观测数据经过多源数据融合模型公式计算出的当前状态的归一化概率值。 The online target space dividing device, wherein a formula for estimating a current position of a pedestrian is obtained according to the multi-source data fusion model formula and a transition probability formula: P(S k )=P(S k-1 )·P(S k |S k-1 )·P(S k |Z k ) represents a normalized probability value of the current state calculated based on the observation data through the multi-source data fusion model formula.
进一步地，本发明还提供一种存储介质，其中，存储有多条指令，所述指令适于由处理器加载并执行上述任一项所述运动行为辅助的室内融合定位方法的步骤。Further, the present invention also provides a storage medium in which a plurality of instructions are stored, the instructions being adapted to be loaded by a processor and performing the steps of the indoor behavior locating method assisted by the motion behavior of any of the above.
综上所述，本发明提供的一种运动行为辅助的室内融合定位方法，通过研究智能终端在多使用模式下的运动行为识别，获取行人的运动轨迹数据以及智能终端接收到的无线信号数据，基于所述运动轨迹数据以及无线信号数据构建面向室内定位的多源数据融合模型。所述模型在运动行为识别的技术上融合多源观测数据，降低了对单模数据的依赖程度，提高了基于智能终端室内定位的精度；同时，本发明结合行为地图匹配推算的历史运动轨迹实现了多源数据融合模型的自适应模型参数自适应学习，可以实现自适应模型参数调整，增加了室内算法对场景、用户、使用模式以及手机设备的适应性。In summary, the present invention provides a motion-assisted indoor fusion positioning method, which acquires the motion track data of the pedestrian and the wireless signal data received by the intelligent terminal by studying the motion behavior recognition of the intelligent terminal in the multi-use mode. A multi-source data fusion model for indoor positioning is constructed based on the motion trajectory data and wireless signal data. The model combines multi-source observation data in the technique of motion behavior recognition, reduces the dependence on single-mode data, and improves the accuracy of indoor positioning based on intelligent terminal; meanwhile, the present invention combines behavioral map matching to calculate historical motion trajectory Adaptive model parameter adaptive learning of multi-source data fusion model can realize adaptive model parameter adjustment and increase the adaptability of indoor algorithm to scene, user, usage mode and mobile device.
可以理解的是，对本领域普通技术人员来说，可以根据本发明的技术方案及本发明构思加以等同替换或改变，而所有这些改变或替换都应属于本发明所附的权利要求的保护范围。It is to be understood that those skilled in the art can make equivalent substitutions or changes to the present invention and the present invention. All such changes and substitutions are intended to fall within the scope of the appended claims.

Claims (10)

  1. 一种运动行为辅助的室内融合定位方法，其特征在于，包括步骤：An exercise behavior-assisted indoor fusion positioning method, comprising the steps of:
    A、基于智能终端在多使用模式下的运动行为获取行人的运动轨迹数据以及智能手机接收到的无线信号数据；A. Acquiring the motion track data of the pedestrian and the wireless signal data received by the smart phone based on the motion behavior of the smart terminal in the multi-use mode;
    B、对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式；B. Parametrically expressing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model and obtaining a transition probability formula of pedestrians between consecutive positions;
    C、根据所述多源数据融合模型以及转移概率公式推算行人当前所处位置。C. Calculate the current location of the pedestrian according to the multi-source data fusion model and the transition probability formula.
  2. 根据权利要求1所述运动行为辅助的室内融合定位方法，其特征在于，所述无线信号数据包括Wifi信号数据和地磁信号数据。The indoor behavior locating method according to claim 1, wherein the wireless signal data comprises Wifi signal data and geomagnetic signal data.
  3. 根据权利要求2所述运动行为辅助的室内融合定位方法，其特征在于，所述步骤B具体包括：The method of claim 2, wherein the step B includes:
    B1、结合运动轨迹数据与室内地图模型计算得到行为识别的特征函数；B1, combined with the motion trajectory data and the indoor map model to calculate the feature function of the behavior recognition;
    B2、计算Wifi信号数据与位置指纹数据库中各个位置信号强度之间的欧式距离，通过欧氏距离的倒数归一化得到Wifi定位的特征函数；B2. Calculating the Euclidean distance between the Wifi signal data and the signal strength of each position in the location fingerprint database, and obtaining the eigenfunction of the Wifi positioning by reciprocal normalization of the Euclidean distance;
    B3、计算地磁信号数据与地磁数据库中各个位置信号强度之间的欧式距离，通过欧式距离的倒数归一化得到地磁定位的特征函数；B3. Calculating the Euclidean distance between the geomagnetic signal data and the signal intensity of each position in the geomagnetic database, and obtaining a characteristic function of the geomagnetic positioning by reciprocal normalization of the Euclidean distance;
    B4、采用条件随机场模型融合所述行为识别特征函数、Wifi定位特征函数以及地磁定位特征函数得到多源数据融合模型公式
    Figure PCTCN2018072114-appb-100001
    其中，S k表示k时刻行人的室内位置，
    Figure PCTCN2018072114-appb-100002
    表示k时刻的第n种多源观测数据，
    Figure PCTCN2018072114-appb-100003
    表示k时刻在S k位置观测到第n种观测数据的概率，λ i为特征函数
    Figure PCTCN2018072114-appb-100004
    的权值参数；     B4. Using the conditional random field model to fuse the behavior recognition feature function, the Wifi positioning feature function and the geomagnetic localization feature function to obtain a multi-source data fusion model formula
    Figure PCTCN2018072114-appb-100001
    Where S k represents the indoor position of the pedestrian at time k,
    Figure PCTCN2018072114-appb-100002
    Indicates the nth multi-source observation data at time k,
    Figure PCTCN2018072114-appb-100003
    Indicates the probability that the nth observation data is observed at the S k position at time k , and λ i is the characteristic function
    Figure PCTCN2018072114-appb-100004
    Weight parameter
    B5、根据运动轨迹数据获得行人在连续位置之间的转移概率
    Figure PCTCN2018072114-appb-100005
    其中，d表示观测距离，θ表示观测距离，σ d和σ θ分别为方向和距离估计的标准差。     B5. Obtaining the transition probability of pedestrians between consecutive positions according to the motion trajectory data
    Figure PCTCN2018072114-appb-100005
    Where d is the observed distance, θ is the observed distance, and σ d and σ θ are the standard deviations of the direction and distance estimates, respectively.
  4. 根据权利要求3所述运动行为辅助的室内融合定位方法，其特征在于， 所述步骤C具体包括：The method according to claim 3, wherein the step C includes:
    根据所述多源数据融合模型公式以及转移概率公式获得推算行人当前所处位置的公式：P(S k)＝P(S k-1)·P(S k|S k-1)·P(S k|Z k)，其中，P(S k|Z k)表示基于观测数据经过多源数据融合模型公式计算出的当前状态的归一化概率值。 Obtaining a formula for estimating the current position of the pedestrian according to the multi-source data fusion model formula and the transition probability formula: P(S k )=P(S k-1 )·P(S k |S k-1 )·P( S k |Z k ), where P(S k |Z k ) represents a normalized probability value of the current state calculated based on the observed data through the multi-source data fusion model formula.
  5. 根据权利要求1所述运动行为辅助的室内融合定位方法，其特征在于，所述智能终端包括智能手机、智能平板、智能手环或智能手表。The indoor behavior locating method according to claim 1, wherein the smart terminal comprises a smart phone, a smart tablet, a smart bracelet or a smart watch.
  6. 一种运动行为辅助的室内融合定位装置，其特征在于，包括：An activity-assisted indoor fusion positioning device, comprising:
    处理器，适于实现各指令；以及a processor adapted to implement each instruction;
    存储设备，适于存储多条指令，所述指令适于由处理器加载并执行以下步骤：A storage device adapted to store a plurality of instructions adapted to be loaded by the processor and to perform the following steps:
    基于智能终端在多使用模式下的运动行为获取行人的运动轨迹数据以及智能手机接收到的无线信号数据；Acquiring the motion track data of the pedestrian and the wireless signal data received by the smart phone based on the motion behavior of the smart terminal in the multi-use mode;
    对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式；Parametrically expressing the motion trajectory data and the wireless signal data, constructing a multi-source data fusion model and obtaining a transition probability formula of pedestrians between consecutive positions;
    根据所述多源数据融合模型以及转移概率公式推算行人当前所处位置。The current location of the pedestrian is estimated according to the multi-source data fusion model and the transition probability formula.
  7. 根据权利要求6所述在线目标空间划分装置，其特征在于，所述无线信号数据包括Wifi信号数据和地磁信号数据。The online target space dividing apparatus according to claim 6, wherein said wireless signal data comprises Wifi signal data and geomagnetic signal data.
  8. 根据权利要求7所述在线目标空间划分装置，其特征在于，对所述运动轨迹数据以及无线信号数据进行参数化表达，构建多源数据融合模型并获得行人在连续位置之间的转移概率公式的步骤具体包括：The online target space dividing device according to claim 7, wherein the motion trajectory data and the wireless signal data are parameterized, the multi-source data fusion model is constructed, and a transition probability formula of the pedestrian between successive positions is obtained. The steps specifically include:
    结合运动轨迹数据与室内地图模型计算得到行为识别的特征函数；The feature function of behavior recognition is calculated by combining the motion trajectory data with the indoor map model;
    计算Wifi信号数据与位置指纹数据库中各个位置信号强度之间的欧式距离，通过欧氏距离的倒数归一化得到Wifi定位的特征函数；Calculating the Euclidean distance between the Wifi signal data and the signal intensity of each position in the location fingerprint database, and obtaining the eigenfunction of the Wifi positioning by reciprocal normalization of the Euclidean distance;
    计算地磁信号数据与地磁数据库中各个位置信号强度之间的欧式距离，通过欧式距离的倒数归一化得到地磁定位的特征函数；Calculating the Euclidean distance between the geomagnetic signal data and the signal intensity of each position in the geomagnetic database, and obtaining the characteristic function of the geomagnetic positioning by reciprocal normalization of the Euclidean distance;
    采用条件随机场模型融合所述行为识别特征函数、Wifi定位特征函数以及地磁定位特征函数得到多源数据融合模型公式
    Figure PCTCN2018072114-appb-100006
    其中，S k表示k时刻行人的室内位置，
    Figure PCTCN2018072114-appb-100007
    表示k时刻的第n种多源观测数据，
    Figure PCTCN2018072114-appb-100008
    表示k时刻在S k位置观测到第n种观测数据的概率，λ i为特征函数
    Figure PCTCN2018072114-appb-100009
    的权值参数；     The multi-source data fusion model formula is obtained by using the conditional random field model to fuse the behavior recognition feature function, Wifi positioning feature function and geomagnetic localization feature function.
    Figure PCTCN2018072114-appb-100006
    Where S k represents the indoor position of the pedestrian at time k,
    Figure PCTCN2018072114-appb-100007
    Indicates the nth multi-source observation data at time k,
    Figure PCTCN2018072114-appb-100008
    Indicates the probability that the nth observation data is observed at the S k position at time k , and λ i is the characteristic function
    Figure PCTCN2018072114-appb-100009
    Weight parameter
    根据运动轨迹数据获得行人在连续位置之间的转移概率
    Figure PCTCN2018072114-appb-100010
    其中，d表示观测距离，θ表示观测距离，σ d和σ θ分别为方向和距离估计的标准差。     Obtaining the transition probability of pedestrians between consecutive positions based on the motion trajectory data
    Figure PCTCN2018072114-appb-100010
    Where d is the observed distance, θ is the observed distance, and σ d and σ θ are the standard deviations of the direction and distance estimates, respectively.
  9. 根据权利要求8所述在线目标空间划分装置，其特征在于，根据所述多源数据融合模型公式以及转移概率公式获得推算行人当前所处位置的公式：P(S k)＝P(S k-1)·P(S k|S k-1)·P(S k|Z k)，其中，P(S k|Z k)表示基于观测数据经过多源数据融合模型公式计算出的当前状态的归一化概率值。 The online target space dividing apparatus according to claim 8, wherein the formula for estimating the current position of the pedestrian is obtained according to the multi-source data fusion model formula and the transition probability formula: P(S k )=P(S k- 1 )·P(S k |S k-1 )·P(S k |Z k ), where P(S k |Z k ) represents the current state calculated based on the observation data through the multi-source data fusion model formula Normalize the probability value.
  10. 一种存储介质，其特征在于，其中存储有多条指令，所述指令适于由处理器加载并执行权利要求1-5任一项所述运动行为辅助的室内融合定位方法的步骤。A storage medium characterized by storing therein a plurality of instructions adapted to be loaded by a processor and performing the steps of the indoor behavior locating method assisted by the motion behavior of any of claims 1-5.
PCT/CN2018/072114 2017-09-27 2018-01-10 Motion-behavior-assisted indoor fusion positioning method and apparatus and storage medium WO2019061949A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710891775.6 2017-09-27
CN201710891775.6A CN107635204B (en) 2017-09-27 2017-09-27 Indoor fusion positioning method and device assisted by exercise behaviors and storage medium

Publications (1)

Publication Number Publication Date
WO2019061949A1 true WO2019061949A1 (en) 2019-04-04

Family

ID=61104098

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/072114 WO2019061949A1 (en) 2017-09-27 2018-01-10 Motion-behavior-assisted indoor fusion positioning method and apparatus and storage medium

Country Status (2)

Country Link
CN (1) CN107635204B (en)
WO (1) WO2019061949A1 (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11009583B1 (en) 2019-12-23 2021-05-18 At&T Intellectual Property I, L.P. Refining and selecting timing data to increase accuracy of timing-based location for user equipment
CN113114850A (en) * 2021-03-18 2021-07-13 电子科技大学 Online fusion positioning method based on surveillance video and PDR

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108322889A (en) * 2018-02-01 2018-07-24 深圳市交投科技有限公司 A kind of method, storage medium and the intelligent terminal of multisource data fusion indoor positioning
CN108362289B (en) * 2018-02-08 2020-12-08 浙江大学城市学院 Mobile intelligent terminal PDR positioning method based on multi-sensor fusion
CN108426582B (en) * 2018-03-03 2021-07-30 北京工业大学 Indoor three-dimensional map matching method for pedestrians
CN109218983B (en) * 2018-06-28 2020-09-18 中国人民解放军国防科技大学 Positioning method and positioning system
CN109029429B (en) * 2018-09-04 2022-02-11 电子科技大学 WiFi and geomagnetic fingerprint based multi-classifier global dynamic fusion positioning method
CN111561921B (en) * 2019-02-14 2022-05-10 华为技术有限公司 Positioning method and device
CN109781094A (en) * 2018-12-24 2019-05-21 上海交通大学 Earth magnetism positioning system based on Recognition with Recurrent Neural Network
CN109945865B (en) * 2019-02-25 2023-02-28 天津大学 Indoor positioning method based on WiFi and geomagnetic fusion
CN110082716B (en) * 2019-04-29 2023-07-11 徐州医科大学 Indoor positioning system and positioning method for complex environment of hospital
CN110619662B (en) * 2019-05-23 2023-01-03 深圳大学 Monocular vision-based multi-pedestrian target space continuous positioning method and system
CN110322067A (en) * 2019-07-03 2019-10-11 中山大学 Location of mobile users prediction technique based on factor graph model
CN111025269B (en) * 2019-12-19 2022-11-15 哈尔滨工程大学 Underwater robot sensor installation deviation estimation method
US11743678B2 (en) * 2020-08-24 2023-08-29 The Hong Kong University Of Science And Technology Generic signal fusion framework for multi-modal localization
CN113358117B (en) * 2021-03-09 2023-05-26 北京工业大学 Visual inertial indoor positioning method using map
CN112985392B (en) * 2021-04-19 2021-07-30 中国人民解放军国防科技大学 Pedestrian inertial navigation method and device based on graph optimization framework

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105910601A (en) * 2016-05-31 2016-08-31 天津大学 Indoor geomagnetic positioning method based on hidden Markov model
CN106054125A (en) * 2016-05-05 2016-10-26 南京邮电大学 Fusion indoor positioning method based on linear-chain conditional random field
CN106123897A (en) * 2016-06-14 2016-11-16 中山大学 Indoor fusion and positioning method based on multiple features
US20170171718A1 (en) * 2015-12-09 2017-06-15 Research & Business Foundation Of Sungkyunkwan University Location estimation method for indoor device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170171718A1 (en) * 2015-12-09 2017-06-15 Research & Business Foundation Of Sungkyunkwan University Location estimation method for indoor device
CN106054125A (en) * 2016-05-05 2016-10-26 南京邮电大学 Fusion indoor positioning method based on linear-chain conditional random field
CN105910601A (en) * 2016-05-31 2016-08-31 天津大学 Indoor geomagnetic positioning method based on hidden Markov model
CN106123897A (en) * 2016-06-14 2016-11-16 中山大学 Indoor fusion and positioning method based on multiple features

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11009583B1 (en) 2019-12-23 2021-05-18 At&T Intellectual Property I, L.P. Refining and selecting timing data to increase accuracy of timing-based location for user equipment
CN113114850A (en) * 2021-03-18 2021-07-13 电子科技大学 Online fusion positioning method based on surveillance video and PDR

Also Published As

Publication number Publication date
CN107635204B (en) 2020-07-28
CN107635204A (en) 2018-01-26

Similar Documents

Publication Publication Date Title
WO2019061949A1 (en) Motion-behavior-assisted indoor fusion positioning method and apparatus and storage medium
CN107396321B (en) Unsupervised indoor positioning method based on mobile phone sensor and iBeacon
WO2017088196A1 (en) Fusion navigation device and method based on wireless fingerprints and mems sensor
CN108632761B (en) Indoor positioning method based on particle filter algorithm
Bae et al. Large-scale indoor positioning using geomagnetic field with deep neural networks
WO2012035707A1 (en) Map generation device, map generation method, method for moving mobile body, and robot device
CN110100150B (en) Effective indoor positioning using geomagnetic field
WO2021012879A1 (en) Trajectory reconstruction method and apparatus, computer device and storage medium
CN106441302B (en) Indoor positioning method in large-scale open area
CN111595344B (en) Multi-posture downlink pedestrian dead reckoning method based on map information assistance
Wang et al. Pedestrian heading estimation based on spatial transformer networks and hierarchical LSTM
Al-Homayani et al. Improved indoor geomagnetic field fingerprinting for smartwatch localization using deep learning
Wu et al. Indoor positioning based on walking-surveyed Wi-Fi fingerprint and corner reference trajectory-geomagnetic database
Wang et al. Deep neural network‐based Wi‐Fi/pedestrian dead reckoning indoor positioning system using adaptive robust factor graph model
CN106028450B (en) Indoor positioning method and equipment
Yang et al. Deepwipos: a deep learning-based wireless positioning framework to address fingerprint instability
Yu et al. Neural networks-based Wi-Fi/PDR indoor navigation fusion methods
Wei et al. MM-Loc: Cross-sensor indoor smartphone location tracking using multimodal deep neural networks
CN108322889A (en) A kind of method, storage medium and the intelligent terminal of multisource data fusion indoor positioning
US10551195B2 (en) Portable device with improved sensor position change detection
Wu et al. CWIWD-IPS: A Crowdsensing/Walk-surveying Inertial/Wi-Fi Data-driven Indoor Positioning System
CN113916223B (en) Positioning method and device, equipment and storage medium
Li et al. An indoor positioning approach using smartphone based on PDR and EKF
Xu et al. An indoor pedestrian navigation algorithm based on smartphone mode recognition
Huang et al. VariFi: Variational Inference for Indoor Pedestrian Localization and Tracking Using IMU and WiFi RSS

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18862702

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 25.09.2020)

122 Ep: pct application non-entry in european phase

Ref document number: 18862702

Country of ref document: EP

Kind code of ref document: A1