Intelligent device pointing interaction control method based on UWB track

Abstract

An intelligent device pointing interaction control method based on a UWB track belongs to the field of man-machine interaction, and comprises the following steps: measuring the relative distance between UWB base stations to obtain UWB distance measurement information, and constructing a UWB base station coordinate system; a user touches each intelligent device through a worn intelligent ring integrated with a UWB module, and records three-dimensional space coordinates of each intelligent device in a UWB base station coordinate system; collecting pointing action track points by adopting a fusion type positioning method combining a multi-dimensional scale transformation method and extended Kalman filtering; extracting a user pointing vector from the user pointing action trajectory by adopting a multi-trajectory point combined pointing vector extraction method; performing user pointing vector offset correction by adopting a human body pointing model; and a two-stage filtering algorithm based on a heuristic rule is adopted to realize target equipment identification. The privacy leakage risk is avoided, the average recognition accuracy is improved, the user pointing intention recognition result is more stable, and the method is suitable for application and popularization.

Description

Technical Field

[0001] This invention belongs to the field of human-computer interaction technology, specifically relating to a smart device pointing interaction control method based on UWB trajectory. Background Technology

[0002] With the widespread application of various smart devices in home, office, and industrial settings, users' demand for more natural and intuitive interaction methods is growing. Gesture interaction, due to its ease of learning and natural intuitiveness, is widely considered an ideal means of human-computer interaction. However, pure gestures are difficult to accurately locate numerous devices distributed in the environment, and cannot meet users' fine-grained control needs in complex scenarios. Therefore, combining gesture recognition with spatial pointing intention recognition can provide users with an efficient, convenient, and intuitive control method, significantly improving the interactive experience of smart devices.

[0003] As wearable sensing devices, smart rings possess the ability to recognize fine-grained gestures and are characterized by convenient wear and low power consumption, making them ideal wearable interaction terminals. However, in existing research on pointing intent recognition based on smart rings, how to quickly and accurately identify the user's true pointing intent remains a key challenge.

[0004] In existing research on pointing intention recognition based on smart rings, one type of method achieves pointing control through infrared communication, such as RingIoT (Darbar, Rajkumar & Choudhury, Mainak & Mullick, Vikalp. (2017). RingIoT: A Smart Ring Controlling Things in Physical Spaces. 10.13140 / RG.2.2.14245.76007.) and Magic Ring (Jing L, Cheng Z, Zhou Y, et al. Magic ring: A self-contained gesture input device on finger[C] / / Proceedings of the 12th International Conference on Mobile and Ubiquitous Multimedia. 2013:1-4.). However, its recognition range is limited, and it is easily affected by occlusion interference, making it difficult to use stably in complex environments. Another type of method uses visual recognition, such as the schemes proposed by CyclopsRing, IRIS, and others (CyclopsRing: Chan L, Chen Y L, Hsieh CH, et al. Cyclopsring: Enabling whole-hand and context-aware interactions through a fisheye ring [C] / / Proceedings of the 28th Annual ACM Symposium on User Interface Software & Technology. 2015: 549-556.) and (IRIS: Kim M, Glenn A, Veluri B, et al. IRIS: Wireless ring for vision-based smart home interaction [C] / / Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology. 2024: 1-16.), which use cameras to acquire user posture and environmental information to recognize target devices. Although such methods have high accuracy in static scenes, they are sensitive to lighting conditions, pose privacy risks, and consume a lot of power, thus limiting their application in everyday interactions.

[0005] With the development of ultra-wideband (UWB) technology, its high-precision ranging capability has been widely used in spatial positioning and target recognition tasks in interactive scenarios. Existing systems such as Minuet (Kang R, Guo A, Laput G, et al. Minuet: Multimodal interaction with an internet of things [C] / / Symposium on spatial user interaction. 2019: 1-10.) and WristQue (Mayton BD, Zhao N, Aldrich M, et al. WristQue: A personal sensor wristband [C] / / 2013 IEEE International Conference on Body Sensor Networks. IEEE, 2013: 1-6.) both use wristband-style wearable devices combined with ultra-wideband (UWB) positioning and inertial measurement units (IMUs) to achieve spatial pointing and interactive control.

[0006] Among them, the Minuet system enables users to control surrounding smart devices via pointing and voice by wearing a wristband that integrates a UWB module and an IMU sensor. Its positioning accuracy reaches approximately 0.3 meters, supporting natural interaction in multi-device environments. The WristQue system, proposed by MIT, utilizes a wearable device on the wrist that integrates posture sensing, UWB positioning, and infrared communication to enable users to interact with and control smart building environments (such as lighting and air conditioning). However, these systems share common problems in long-term use: IMU posture estimation is susceptible to accumulated errors (drift), causing the inferred pointing direction to gradually deviate from the actual target over time; simultaneously, auxiliary solutions relying on vision or infrared links are easily affected by environmental occlusion, changes in lighting conditions, and other factors, leading to decreased system recognition accuracy and insufficient stability.

[0007] Therefore, how to improve the spatial accuracy and system robustness of pointing intention recognition while maintaining the naturalness of wearing the device has become a key issue that urgently needs to be addressed in the field of pointing interaction of smart devices. Summary of the Invention

[0008] The purpose of this invention is to provide a smart device pointing interaction control method based on UWB trajectory. By improving the spatial accuracy and system robustness of pointing intention recognition, it can be widely applied to smart home, smart office, vehicle system, industrial control and entertainment interaction scenarios.

[0009] The technical solution adopted by this invention to solve the technical problem is as follows:

[0010] This invention provides a method for intelligent device pointing and interactive control based on UWB trajectory, which specifically includes the following steps:

[0011] Step 1: Measure the relative distance between each UWB base station to obtain UWB ranging information and construct a UWB base station coordinate system; The user touches each smart device with a smart ring that integrates a UWB module, records the three-dimensional spatial coordinates of each smart device in the UWB base station coordinate system, and completes the construction of the smart device map.

[0012] Step 2: Use a fusion positioning method that combines multidimensional scaling transformation and extended Kalman filtering to collect pointing action trajectory points and form the user pointing action trajectory;

[0013] Step 3: Extract the user pointing vector from the user pointing action trajectory using a multi-trajectory point combination pointing vector extraction method; use a human pointing model to correct the user pointing vector offset;

[0014] Step 4: Use a two-stage filtering algorithm based on heuristic rules to identify the target device.

[0015] Furthermore, in step one, the UWB ranging information is obtained by measuring using a two-way ranging method; the UWB base station coordinate system is constructed using a multi-dimensional scaling transformation algorithm.

[0016] Furthermore, in step two, the multi-dimensional scaling transformation method is used to perform initial coordinate estimation of the smart ring position based on the UWB ranging information at the current moment, using the UWB base station coordinates A = [a1, a2, ... a2]. N ,] T a i =[x i ,y i ,z i ] T The i-th UWB base station represents its three-dimensional spatial coordinates in the UWB base station coordinate system, and N is the total number of UWB base stations; the observation distance vector from the smart ring to each UWB base station is D = [d1, d2, ... dn]. N ,] T d i This represents the ranging observation value from the smart ring to the i-th UWB base station; the superscript T indicates the transpose operation of a matrix or vector; through dual-centering Calculate the inner product matrix, and then perform an analysis on the inner product matrix B = VΛV. T Perform eigenvalue decomposition, and take the first three largest positive eigenvalues ​​λ1, λ2, and λ3 to form a matrix Λ3 = diag(λ1, λ2, λ3). Then, the initial coordinates of the smart ring in the local coordinate system are represented as follows: J is the centered matrix, I is the identity matrix, Λ is the eigenvalue diagonal matrix, and V is the eigenvector matrix.

[0017] Furthermore, by calculating the UWB base station coordinates A' in the local coordinate system, a minimum mean square error rigid transformation matrix R is constructed using the actual UWB base station coordinates A. The mathematical expression for this process is as follows:

[0018] X init =R(X) init ),R≈A(A')

[0019] Among them, X init The initial position estimate of the smart ring is used as the initial state input for the EKF.

[0020] Furthermore, in step two, the extended Kalman filter is used to recursively estimate the dynamic motion state of the smart ring, model a nonlinear observation equation, and achieve linear approximation of the nonlinear observation equation through a first-order Taylor expansion. Then, the nonlinear observation equation is linearized by the Jacobian matrix, and the standard EKF prediction and update steps are executed to complete the dynamic tracking of the smart ring's position.

[0021] Furthermore, the calculation formula for the linear observation equation is as follows:

[0022] Z(m)=||P(k)-a(l)||+v(k,l)

[0023] Where Z(m) is the m-th ranging observation, P(k) is the three-dimensional spatial coordinate of the smart ring's position at time k, and a(l) = [x l ,y l ,z l ] T Let v(k,l) be the three-dimensional spatial coordinates of the l-th UWB base station, and v(k,l) be the Gaussian white noise term.

[0024] In the process of achieving linear approximation of the nonlinear observation equation through first-order Taylor expansion, the system state vector is defined as follows: Let K be the spatial coordinates of the smart ring at time k. The three-dimensional velocity components of the smart ring at time k. Let X be the three-dimensional acceleration component of the smart ring at time k. Under the assumption of constant acceleration, the system state transition equation is X. k =FX k-1 +w(k), X k-1 Let K be the state vector of the system at time k-1. Let F be the process noise and F be the state transition matrix.

[0025] Furthermore, in the process of linearizing the nonlinear observation equation using the Jacobian matrix, taking the i-th UWB base station as an example, its nonlinear observation equation with respect to the three-dimensional spatial coordinates, i.e., the position component p, of the smart ring position at time k is... (k) The Jacobian matrix is ​​defined as follows:

[0026]

[0027] Among them, z (k,i) p represents the ranging observation value obtained by the i-th UWB base station at time k. (k) To represent the three-dimensional spatial coordinate vector of the smart ring at time k, a (i) Let i be the fixed positions of the i UWB base stations in the three-dimensional space of the coordinate system.

[0028] Furthermore, in step three, the specific implementation process of the pointing vector extraction method combining multiple trajectory points is as follows:

[0029] The pointing trajectory points are standardized. From the latter half of the pointing trajectory, two trajectory points P1 and P2 that form the largest angle with the starting point are selected. Let the starting point of the pointing trajectory be P0. The pointing vector is calculated according to the following rules.

[0030] 1) If ∠P0P1P2 < 5°, it indicates that the overall linearity of the pointing motion trajectory is high, and the vector between the starting point and the ending point can be directly used as the pointing vector;

[0031] 2) If 5° < ∠P0P1P2 < 12°, then calculate the vectors respectively. and Pointer vector

[0032] 3) If ∠P0P1P2>12°, then there may be points with severe drift among the trajectory points that form the maximum included angle, so this combination should be excluded.

[0033] Furthermore, in step three, the mathematical expression for the human body pointing model is as follows:

[0034]

[0035] Where θ is the left and right offset angle caused by the human body, w is the user's shoulder width, and L is the user's arm length; based on the extracted pointing vector, the pointing vector is rotated and corrected in the opposite direction along the line perpendicular to the line connecting the human eye to the target device, so that the final pointing vector conforms to the actual line of sight.

[0036] Furthermore, in step four, the specific implementation process of the two-stage filtering algorithm based on heuristic rules is as follows:

[0037] (1) Based on the user's current location P u Spatial registration locations of smart devices P1, P2, ... P N ,calculate Retain The initial screening has been completed, among which, For user location P u Starting from point P, the device is located at position P. i The spatial vector that ends there;

[0038] (2) Next, calculate the angle between any two smart devices and the user's location, i.e., the starting position of the smart ring, and apply the following precision rules to filter out non-target devices:

[0039] a.when It assumes that the two smart devices are collinear in line of sight, and selects the device closest to the user. For user location P u Starting from point P, the device is located at position P. j The spatial vector is the endpoint.

[0040] b. When If so, then further determine whether it meets the requirements. If satisfied, filter out. Larger smart devices.

[0041] c. For Then filter it directly. Larger smart devices.

[0042] The beneficial effects of this invention are:

[0043] ① Compared with camera-based pointing intention recognition methods, the method proposed in this invention relies on UWB positioning technology to realize user pointing intention recognition, without the need to collect image data, thus avoiding the acquisition of sensitive information such as user face and home environment, and avoiding the risk of privacy leakage.

[0044] ② Compared with IMU-based pointing intention recognition methods, the method proposed in this invention makes full use of the characteristic of UWB positioning without cumulative error, making the user pointing intention recognition results more stable and effectively avoiding recognition drift problems caused by time accumulation.

[0045] ③ In the context of smart home and entertainment interaction, an experiment was conducted with 57 users. The method proposed in this invention achieved an average recognition accuracy of 94.86%, demonstrating good stability and practicality, and is suitable for widespread application. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of a smart device pointing and interactive control method based on UWB trajectory provided by the present invention.

[0047] Figure 2 The flowchart illustrates a UWB trajectory-based intelligent device pointing and interactive control method provided by this invention. Detailed Implementation

[0048] The present invention will be further described in detail below with reference to the accompanying drawings.

[0049] like Figure 1 As shown, this invention provides a smart device pointing interaction control method based on UWB trajectory, which mainly includes four parts: smart device map construction, pointing action trajectory point acquisition, user pointing vector extraction, and target device recognition algorithm. Through a smart ring worn by the user integrating a UWB module, combined with a preset UWB base station, the method accurately identifies the user's pointing intention when interacting with the smart device.

[0050] See Figure 2 As shown, the present invention provides a smart device pointing and interactive control method based on UWB trajectory, the specific implementation process of which is as follows:

[0051] Step 1: Building a map of smart devices;

[0052] In this invention, the user needs to wear a smart ring that integrates a UWB module.

[0053] To obtain the three-dimensional spatial coordinates of the smart ring's location, at least three UWB base stations need to be pre-deployed. The relative distance between each UWB base station is measured using the two-way ranging method (TWR) to obtain UWB ranging information, and a UWB base station coordinate system is constructed based on the multidimensional scaling transformation (MDS) algorithm.

[0054] After the UWB base station coordinate system is established, users can touch each smart device with a smart ring (which integrates a UWB module) to record the three-dimensional spatial coordinates of each smart device in the UWB base station coordinate system, thereby completing the construction of the smart device map.

[0055] Step 2: Collect the movement trajectory points;

[0056] When a user points to a smart device, their hand rises from their sides to the center of their line of sight. To achieve high-precision real-time acquisition of the user's hand trajectory during this process, this invention proposes a fusion localization method combining Multidimensional Scale Transform (MDS) and Extended Kalman Filter (EKF) to achieve coordinated accuracy in initial localization and trajectory tracking. The specific implementation process is as follows:

[0057] S2.1: Initial estimation of the smart ring's position based on MDS;

[0058] To achieve a reasonable initial state setting for EKF and thus achieve rapid convergence, this invention first uses the Multidimensional Scaling Transform (MDS) method to estimate the initial coordinates of the smart ring position based on the current UWB ranging information.

[0059] Specifically, using UWB base station coordinates A = [a1, a2, ... a N ,] T a i =[x i ,y i ,z i ] T Represents the three-dimensional spatial coordinates (x, y) of the i-th UWB base station in the UWB base station coordinate system. i ,y i ,z i N is the total number of UWB base stations; the observation distance vector from the smart ring to each UWB base station is D = [d1, d2, ... dn]. N ,] T d i This represents the ranging observation value from the smart ring to the i-th UWB base station; the superscript T indicates the transpose operation of a matrix or vector; through dual-centering Calculate the inner product matrix, and then perform an analysis on the inner product matrix B = VΛV. T Perform eigenvalue decomposition, and take the first three largest positive eigenvalues ​​λ1, λ2, and λ3 to form a matrix Λ3 = diag(λ1, λ2, λ3). Then, the initial coordinates of the smart ring in the local coordinate system are represented as follows: Where J is the centering matrix, I is the identity matrix, Λ is the eigenvalue diagonal matrix, V is the eigenvector matrix, and X is the eigenvector matrix. init Initialize the coordinates for the smart ring's position.

[0060] Due to the initial coordinates X of the smart ring's position init Within the local coordinate system obtained from the dimensionality reduction operation, a rigid transformation is needed to align it to the real UWB base station coordinate system. To this end, the UWB base station coordinates A' in the corresponding local coordinate system are calculated using the same method (step one: "UWB base stations measure relative distances using the two-way ranging method (TWR) to obtain UWB ranging information, and construct a UWB base station coordinate system based on the multidimensional scaling transformation (MDS) algorithm"). A minimum mean square error rigid transformation matrix R is then constructed using the real UWB base station coordinates A. The mathematical expression for this process is as follows:

[0061] X init =R(X) init ),A≈R(A')

[0062] Through the above calculations, the initial position estimate X of the smart ring in the actual UWB base station coordinate system is finally obtained. init , as the initial state input of EKF.

[0063] S2.2: Dynamic tracking of the position of a smart ring based on EKF;

[0064] Based on the MDS initialization, this invention uses the extended Kalman filter (EKF) to recursively estimate the dynamic motion state of the smart ring and directly models the nonlinear observation equation, thereby achieving full integration of ranging information and system state in the time and space domains.

[0065] Specifically, the calculation formula for the nonlinear observation equation is as follows:

[0066] Z(m)=||P(k)-a(l)||+v(k,l)

[0067] Where Z(m) is the ranging observation value, which is the m-th ranging observation value of "the observed distance from the smart ring to the UWB base station" in step 2.1, P(k) is the three-dimensional spatial coordinate of the smart ring position at time k, and a(l) = [x l ,y l ,z l ] T Let v(k,l) be the three-dimensional spatial coordinates of the l-th UWB base station, and v(k,l) be the Gaussian white noise term.

[0068] Unlike traditional trilateration methods that approximate nonlinear equations by subtracting two equations, this invention directly introduces the aforementioned nonlinear observation equations into the EKF framework and achieves linear approximation of the nonlinear observation equations through first-order Taylor expansion, thereby improving the accuracy and stability of multi-base station joint estimation.

[0069] Specifically, the system state vector is defined as follows: Let K be the spatial coordinates of the smart ring at time k. The three-dimensional velocity components of the smart ring at time k. Let X be the three-dimensional acceleration component of the smart ring at time k. Under the assumption of constant acceleration, the system state transition equation is X. k =FX k-1 +w(k), X k-1 Let K be the state vector of the system at time k-1. This is process noise.

[0070] F is the state transition matrix, and its specific calculation formula is as follows:

[0071]

[0072] Where, I∈R 3×3Let be the identity matrix, and Δt be the sampling time interval.

[0073] In this invention, the nonlinear observation equation is in a nonlinear form and needs to be linearized using the Jacobian matrix.

[0074] Specifically, taking the i-th UWB base station as an example, its nonlinear observation equation relates to the three-dimensional spatial coordinates, i.e., the position component p, of the smart ring position at time k. (k) The Jacobian matrix is ​​defined as follows:

[0075]

[0076] Among them, z (k，i) p represents the ranging observation value obtained by the i-th UWB base station at time k. (k) To represent the three-dimensional spatial coordinate vector of the smart ring at time k, a (i) Let i be the fixed positions of the i UWB base stations in the three-dimensional space of the coordinate system.

[0077] Based on the observation information of all available UWB base stations, a multidimensional observation vector and Jacobian matrix are constructed. By executing the standard EKF prediction and update steps, the dynamic tracking of the smart ring's position can be completed, thereby completing the collection of pointing action trajectory points and forming the user's pointing action trajectory.

[0078] Step 3: Extracting the user-pointed vector;

[0079] Extracting a high-precision pointing vector from the user's pointing motion trajectory is the core task of this invention. Traditional methods typically use the line connecting the start and end points of the pointing motion trajectory as the pointing vector. However, this method is sensitive to UWB positioning errors and user hand tremors, which can easily cause the pointing motion to deviate from the actual target device, resulting in poor robustness.

[0080] Statistical analysis and practical observation revealed that in scenarios where users point with their right hand, the arm typically rises from the right side of the body and tends to align with the line connecting the user's eyes to the target device. This pointing motion pattern causes a slight leftward deviation in the actual pointing trajectory, further affecting the accuracy of the user's pointing vector calculation.

[0081] To address the above problems, the present invention proposes the following improved method:

[0082] S3.1: A method for extracting pointing vectors by combining multiple trajectory points;

[0083] This invention first standardizes the pointing trajectory points, reducing the number of points to 30 to reduce computational load, smooth the pointing trajectory, and minimize the impact of outliers. After obtaining the standardized pointing trajectory, two trajectory points P1 and P2, forming the largest angle with the starting point, are selected from the latter half of the trajectory to assist in calculating the pointing vector. Let the starting point of the pointing trajectory be P0, and the pointing vector is calculated according to the following rules. Calculation:

[0084] 1) If ∠P0P1P2 < 5°, it indicates that the overall linearity of the pointing motion trajectory is high, and the vector between the starting point and the ending point can be directly used as the pointing vector;

[0085] 2) If 5° < ∠P0P1P2 < 12°, then calculate the vectors respectively. and Pointer vector

[0086] 3) If ∠P0P1P2>12°, then there may be points with severe drift among the trajectory points that form the maximum included angle, so this combination should be excluded.

[0087] S3.2: Pointing vector offset correction based on human pointing model;

[0088] The user pointing vector required by this invention should accurately reflect the spatial connection between the user's eye position and the target device. To make the pointing vector more accurately represent the user's pointing action, this invention constructs a human pointing model for pointing vector offset correction, which systematically corrects the original pointing vector, thereby more realistically restoring the user's actual pointing intention. Its specific implementation process is as follows:

[0089] The offset angle of the original pointing vector is related to the user's shoulder width and arm length, and can be approximated as follows:

[0090]

[0091] Where θ is the left and right offset angle caused by the human body, w is the user's shoulder width, and L is the user's arm length.

[0092] Statistical measurements revealed that for adult users (18-65 years old), the offset angle of the original pointing vector is generally between 13.8° and 16.2°. Based on whether the user uses their left or right hand, this invention performs a reverse rotation correction along a direction perpendicular to the line connecting the user's eye to the target device, making the final pointing vector more consistent with the actual line of sight, thereby improving the accuracy of target recognition.

[0093] Step 4: Target Device Identification Algorithm;

[0094] To balance the accuracy of target device identification with the user interaction experience, this invention designs an adaptive target device identification algorithm: when the smart devices are sparsely distributed or the user's pointing intention is clear, the most likely single target device can be directly output; while in scenarios where the smart devices are densely distributed or the user's pointing intention is uncertain, the two most likely candidate devices are output, and then gesture recognition is combined to allow the user to actively select the target device by swiping left and right.

[0095] To achieve the above objectives, this invention employs a two-stage filtering algorithm based on heuristic rules, the specific implementation process of which is as follows:

[0096] (1) Based on the user's current location P u Spatial registration locations of smart devices P1, P2, ... P N ,calculate Retain The initial screening has been completed, among which, For user location P u Starting from point P, the device is located at position P. i The spatial vector is the endpoint.

[0097] (2) Next, calculate the angle between any two smart devices and the user's position (i.e., the starting position of the smart ring), and apply three precision rules to filter out non-target devices, as follows:

[0098] a.when It assumes that the two smart devices are collinear in line of sight, and selects the device closest to the user. For user location P u Starting from point P, the device is located at position P. j The spatial vector is the endpoint.

[0099] b. When If so, then further determine whether it meets the requirements. If satisfied, filter out. Larger smart devices.

[0100] c. For Then filter it directly. Larger smart devices.

[0101] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent device pointing and interactive control based on UWB trajectory, characterized in that, Includes the following steps: Step 1: Measure the relative distance between each UWB base station to obtain UWB ranging information and construct a UWB base station coordinate system; The user touches each smart device with a smart ring that integrates a UWB module, records the three-dimensional spatial coordinates of each smart device in the UWB base station coordinate system, and completes the construction of the smart device map. Step 2: Use a fusion positioning method that combines multidimensional scaling transformation and extended Kalman filtering to collect pointing action trajectory points and form the user pointing action trajectory; Step 3: Extract the user pointing vector from the user pointing action trajectory using a multi-trajectory point combination pointing vector extraction method; use a human pointing model to correct the user pointing vector offset; Step 4: Use a two-stage filtering algorithm based on heuristic rules to identify the target device.

2. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 1, characterized in that, In step one, the UWB ranging information is obtained by measuring using a two-way ranging method; the UWB base station coordinate system is constructed using a multi-dimensional scaling transformation algorithm.

3. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 1, characterized in that, In step two, the multi-dimensional scaling transformation method is used to estimate the initial coordinates of the smart ring position based on the UWB ranging information at the current moment, using the UWB base station coordinates A = [a1, a2, ... a2]. N ,] T a i =[x i ,y i ,z i ] T The i-th UWB base station represents its three-dimensional spatial coordinates in the UWB base station coordinate system, and N is the total number of UWB base stations; the observation distance vector from the smart ring to each UWB base station is D = [d1, d2, ... dn]. N ,] T d i This represents the ranging observation value from the smart ring to the i-th UWB base station; the superscript T indicates the transpose operation of a matrix or vector; through dual-centering Calculate the inner product matrix, and then perform an analysis on the inner product matrix B = VΛV. T Perform eigenvalue decomposition, and take the first three largest positive eigenvalues ​​λ1, λ2, and λ3 to form a matrix Λ3 = diag(λ1, λ2, λ3). Then, the initial coordinates of the smart ring in the local coordinate system are represented as follows: J is the centered matrix, I is the identity matrix, Λ is the eigenvalue diagonal matrix, and V is the eigenvector matrix.

4. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 3, characterized in that, By calculating the UWB base station coordinates A in the local coordinate system ’ A minimum mean square error rigid transformation matrix R is constructed using the actual UWB base station coordinates A. The mathematical expression for this process is as follows: X init =R(X init ),A≈R(A’) Among them, X init The initial position estimate of the smart ring is used as the initial state input for the EKF.

5. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 1, characterized in that, In step two, the extended Kalman filter is used to recursively estimate the dynamic motion state of the smart ring, model a nonlinear observation equation, and achieve linear approximation of the nonlinear observation equation through a first-order Taylor expansion. Then, the nonlinear observation equation is linearized by the Jacobian matrix, and the standard EKF prediction and update steps are executed to complete the dynamic tracking of the smart ring's position.

6. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 5, characterized in that, The calculation formula for the linear observation equation is as follows: Z(m)=||P(k)-a(l)||+v(k,l) Where z(m) is the m-th ranging observation, P(k) is the three-dimensional spatial coordinate of the smart ring position at time k, and a(l) = [x l ,y l ,z l ] T Let v(k,l) be the three-dimensional spatial coordinates of the l-th UWB base station, and v(k,l) be the Gaussian white noise term. In the process of achieving linear approximation of the nonlinear observation equation through first-order Taylor expansion, the system state vector is defined as follows: Let K be the spatial coordinates of the smart ring at time k. The three-dimensional velocity components of the smart ring at time k. Let X be the three-dimensional acceleration component of the smart ring at time k. Under the assumption of constant acceleration, the system state transition equation is X. k =FX k-1 +w(k), X k-1 Let K be the state vector of the system at time k-1. Let F be the process noise and F be the state transition matrix.

7. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 5, characterized in that, In the process of linearizing the nonlinear observation equation using the Jacobian matrix, taking the i-th UWB base station as an example, its nonlinear observation equation with respect to the three-dimensional spatial coordinates, i.e., the position component p, of the smart ring position at time k is... (k) The Jacobian matrix is ​​defined as follows: Among them, z (k,i) p represents the ranging observation value obtained by the i-th UWB base station at time k. (k) To represent the three-dimensional spatial coordinate vector of the smart ring at time k, a (i) Let i be the fixed positions of the i UWB base stations in the three-dimensional space of the coordinate system.

8. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 1, characterized in that, In step three, the specific implementation process of the multi-trajectory point combined pointing vector extraction method is as follows: The pointing trajectory points are standardized. From the latter half of the pointing trajectory, two trajectory points P1 and P2 that form the largest angle with the starting point are selected. Let the starting point of the pointing trajectory be P0. The pointing vector is calculated according to the following rules. 1) If ∠P0P1P2 < 5°, it indicates that the overall linearity of the pointing motion trajectory is high, and the vector between the starting point and the ending point can be directly used as the pointing vector; 2) If 5° < ∠P0P1P2 < 12°, then calculate the vectors respectively. and Pointer vector 3) If ∠P0P1P2>12°, then there may be points with severe drift among the trajectory points that form the maximum included angle, so this combination should be excluded.

9. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 1, characterized in that, In step three, the mathematical expression for the human body pointing model is as follows: Where θ is the left and right offset angle caused by the human body, w is the user's shoulder width, and L is the user's arm length; based on the extracted pointing vector, the pointing vector is rotated and corrected in the opposite direction along the line perpendicular to the line connecting the human eye to the target device, so that the final pointing vector conforms to the actual line of sight.

10. The intelligent device pointing and interactive control method based on UWB trajectory according to claim 1, characterized in that, In step four, the specific implementation process of the heuristic rule-based two-stage filtering algorithm is as follows: (1) Based on the user's current location P u Spatial registration locations P1, P2, ... P of smart devices N ,calculate Retain The initial screening has been completed, among which, For user location P u Starting from point P, the device is located at position P. i The spatial vector that ends there; (2) Next, calculate the angle between any two smart devices and the user's location, i.e., the starting position of the smart ring, and apply the following precision rules to filter out non-target devices: a.When It assumes that the two smart devices are collinear in line of sight, and selects the device closest to the user. For user location P u Starting from point P, the device is located at position P. j The spatial vector is the endpoint. b. When If so, then further determine whether it meets the requirements. If satisfied, filter out. Larger smart devices. c. For Then filter it directly. Larger smart devices.

Images