CN-122018706-A - Non-inductive man-machine interaction method and system integrating UWB and space sound field

CN122018706ACN 122018706 ACN122018706 ACN 122018706ACN-122018706-A

Abstract

The invention discloses a non-inductive man-machine interaction method and system fusing UWB and space sound field, belonging to intelligent control technique field, the invention takes the original distance measurement value between the current UWB positioning label and each UWB positioning base station of the user as input through the trained neural network model, outputs the distance measurement error correction value corresponding to the current UWB positioning label and each UWB positioning base station of the user, according to the ranging error correction value, a corrected ranging value between the current UWB positioning tag of the user and each UWB positioning base station is obtained, after an arrival time difference equation set is constructed by using the corrected ranging value, first coordinate data of the user is obtained through calculation by a chan algorithm, accuracy of UWB positioning is improved, and therefore interactive instruction recognition efficiency and accuracy are improved.

Inventors

Zheng Kuixiong
CHEN SHIBIN
HUANG PINFENG
LI JINGHONG
LIU ZHIHONG

Assignees

广东粤景润科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260416

Claims (10)

1. The non-inductive man-machine interaction method for fusing UWB and space sound field is characterized by comprising the following steps: S1, establishing a UWB positioning coordinate system and an acoustic field coordinate system based on an interaction space, and performing space joint calibration on the UWB positioning coordinate system and the acoustic field coordinate system to obtain transformation parameters between the UWB positioning coordinate system and the acoustic field coordinate system, wherein the transformation parameters comprise a rotation matrix R1 and a translation vector T1; s2, in an interaction space, a user wears equipment with UWB positioning tags on hands, and based on a UWB positioning coordinate system and the UWB positioning tags of the user, first coordinate data of the user are obtained through real-time calculation, specifically: s21, using a test original ranging value sequence from the tested UWB tag to each UWB positioning base station and a corresponding test ranging error correction value as training samples to train a neural network model, taking an original ranging value between the current UWB positioning tag of a user and each UWB positioning base station as input, and outputting a corresponding ranging error correction value between the current UWB positioning tag of the user and each UWB positioning base station; s22, selecting one UWB positioning base station as a reference base station, and constructing an arrival time difference equation set by using the corrected ranging value: ; The method comprises the steps that at the time t, a corrected ranging value is formed between a UWB positioning tag of a user and an ith UWB positioning base station; At time t, a corrected ranging value between the UWB positioning tag of the user and the reference base station; The position coordinates of the UWB positioning tag of the user in a UWB positioning coordinate system at the time t; The position coordinate of the ith UWB positioning base station in a UWB positioning coordinate system is obtained; position coordinates of a reference base station in a UWB positioning coordinate system; S23, solving the arrival time difference equation set through an improved chan algorithm to obtain first coordinate data of the user, namely, the position coordinates of the UWB positioning tag of the user in a UWB positioning coordinate system; S3, acquiring a first coordinate data sequence in the latest time period according to the step S2, acquiring a hand motion track of a user in the latest time period according to the first coordinate data sequence, and carrying out interactive instruction recognition and matching according to the hand motion track; And S4, controlling corresponding equipment to complete interaction instruction actions according to the matched interaction instructions, simultaneously converting the current first coordinate data of the user into a sound field coordinate system according to the rotation matrix R1 and the translation vector T1 to obtain corresponding second coordinate data, and controlling each loudspeaker to conduct sound directional projection to the second coordinate data through a phased array beam forming technology to broadcast interaction feedback voice.
2. The non-inductive man-machine interaction method fusing UWB and space sound fields according to claim 1, wherein the UWB positioning coordinate system is formed by disposing at least 4 UWB positioning base stations in the interaction space, and one UWB positioning base station is used as a coordinate origin of the UWB positioning coordinate system; The sound field coordinate system is formed by disposing a plurality of speakers in an interaction space, and taking one speaker as a coordinate origin of the sound field coordinate system.
3. The method for non-inductive man-machine interaction of merging UWB and spatial sound field according to claim 1, wherein the spatial joint calibration is performed on the UWB positioning coordinate system and the acoustic field coordinate system to obtain transformation parameters between the UWB positioning coordinate system and the acoustic field coordinate system, specifically: Setting a calibration target in an interaction space, wherein a reference UWB positioning label and a reference microphone are arranged on the calibration target, sequentially moving the calibration target to a plurality of preset positions in the interaction space, recording a plurality of label coordinates P1 calculated by UWB and a plurality of sound source receiving coordinates P2 which are transmitted by a sound field generator, received and reversely calculated by the reference microphone in the process, wherein the label coordinates and the sound source receiving coordinates corresponding to the same preset position are regarded as a coordinate pair, and solving a rotation matrix R1 and a translation vector T1 between a UWB positioning coordinate system and a sound field coordinate system by a least square method, so that P2=R1×P1+T1 holds for all coordinate pairs.
4. The non-inductive human-computer interaction method for fusing UWB and space sound field according to claim 1, wherein the test ranging error correction value is obtained through advanced test, specifically: setting a test UWB label to continuously move in an interaction space, adopting a bilateral two-way ranging protocol in the process, collecting test original flight time between the test UWB label and each UWB positioning base station at a fixed frequency, and converting the test original flight time into a corresponding test original ranging value: ; representing the test raw ranging value between the test UWB tag and the i-th UWB positioning base station at time t, c being the speed of light, The method comprises the steps that at the time t, the original flight time of a test between a UWB tag and an ith UWB positioning base station is shown; Acquiring real geometric distances between the test UWB labels corresponding to the test original ranging values and the corresponding UWB positioning base stations, and acquiring corresponding test ranging error correction values according to the original ranging values and the corresponding real geometric distances: ; Indicating the test range error correction value of the test UWB tag and the ith UWB positioning base station at time t, To test the true geometrical distance between the UWB tag and the i-th UWB positioning base station at time t.
5. The non-inductive man-machine interaction method fusing UWB and space sound field according to claim 1, wherein the interactive instruction recognition matching is performed according to the hand motion track, specifically: acquiring a preset interaction instruction model library, wherein the interaction instruction model library records a track motion sequence and a track final imaging graph corresponding to each interaction instruction; Projecting each coordinate point in a first coordinate data sequence corresponding to the hand motion track onto the same plane, and smoothly connecting the projection points according to the sequence of the first coordinate data acquisition to obtain a first graph; And screening out a corresponding matched track motion sequence from an interactive instruction model library according to the formation sequence of the hand motion track, performing similarity analysis on a track final imaging graph corresponding to the matched track motion sequence and the first graph, and matching a corresponding interactive instruction with the hand motion track if the similarity is larger than a preset similarity threshold.
6. The method of claim 1, wherein the phased array beam forming technology is used to control each speaker to perform sound directional projection to the second coordinate data, specifically: obtaining the linear distance from each loudspeaker to the current user according to the coordinate position of each loudspeaker in the sound field coordinate system and the second coordinate data; calculating the required emission phase of each loudspeaker according to the linear distance: ; the transmit phase to be applied for the j-th speaker, f is the carrier frequency of the sound wave emitted by the speaker, cy is the speed of sound, For the straight line distance of the jth speaker to the current user, The initial reference phase is preset; according to the required transmitting phase of each loudspeaker, each loudspeaker simultaneously transmits the sound wave after phase adjustment.
7. An unmanned man-machine interaction system for fusing UWB and space sound field, which is applied to the unmanned man-machine interaction method for fusing UWB and space sound field as set forth in any one of claims 1 to 6, and is characterized by comprising: The calibration module is used for establishing a UWB positioning coordinate system and an acoustic field coordinate system based on the interaction space, and carrying out space joint calibration on the UWB positioning coordinate system and the acoustic field coordinate system to obtain transformation parameters between the UWB positioning coordinate system and the acoustic field coordinate system, wherein the transformation parameters comprise a rotation matrix R1 and a translation vector T1; The resolving module is used for wearing equipment provided with UWB positioning labels on the hands of the users in the interaction space, and resolving and acquiring first coordinate data of the users in real time based on the UWB positioning coordinate system and the UWB positioning labels of the users; The instruction matching module is used for acquiring a first coordinate data sequence in the latest time period, acquiring a hand motion track of a user in the latest time period according to the first coordinate data sequence, and carrying out interactive instruction identification matching according to the hand motion track; And the execution broadcasting module is used for controlling corresponding equipment to complete interaction instruction actions according to the matched interaction instructions, simultaneously converting the current first coordinate data of the user into a sound field coordinate system according to the rotation matrix R1 and the translation vector T1 to obtain corresponding second coordinate data, and controlling each loudspeaker to carry out sound directional projection on the second coordinate data through a phased array beam forming technology to broadcast interaction feedback voice.
8. The non-inductive man-machine interaction system fusing UWB and spatial sound fields according to claim 7, wherein the UWB positioning coordinate system is specifically configured by disposing at least 4 UWB positioning base stations in the interaction space, and using one of the UWB positioning base stations as the origin of coordinates of the UWB positioning coordinate system; The sound field coordinate system is formed by disposing a plurality of speakers in an interaction space, and taking one speaker as a coordinate origin of the sound field coordinate system.
9. The non-inductive man-machine interaction system fusing UWB and space sound fields according to claim 7, wherein the spatial joint calibration is performed on a UWB positioning coordinate system and an acoustic field coordinate system to obtain transformation parameters between the UWB positioning coordinate system and the acoustic field coordinate system, specifically: Setting a calibration target in an interaction space, wherein a reference UWB positioning label and a reference microphone are arranged on the calibration target, sequentially moving the calibration target to a plurality of preset positions in the interaction space, recording a plurality of label coordinates P1 calculated by UWB and a plurality of sound source receiving coordinates P2 which are transmitted by a sound field generator, received and reversely calculated by the reference microphone in the process, wherein the label coordinates and the sound source receiving coordinates corresponding to the same preset position are regarded as a coordinate pair, and solving a rotation matrix R1 and a translation vector T1 between a UWB positioning coordinate system and a sound field coordinate system by a least square method, so that P2=R1×P1+T1 holds for all coordinate pairs.
10. The system of claim 7, wherein the real-time resolving obtains first coordinate data of the user, specifically: The method comprises the steps of using a test original ranging value sequence from a tested UWB tag to each UWB positioning base station and a corresponding test ranging error correction value as training samples to train a neural network model, using an original ranging value between a current UWB positioning tag of a user and each UWB positioning base station as input, and outputting a ranging error correction value corresponding to the current UWB positioning tag of the user and each UWB positioning base station; Selecting one UWB positioning base station as a reference base station, and constructing an arrival time difference equation set by using the corrected ranging value: ; The method comprises the steps that at the time t, a corrected ranging value is formed between a UWB positioning tag of a user and an ith UWB positioning base station; At time t, a corrected ranging value between the UWB positioning tag of the user and the reference base station; The position coordinates of the UWB positioning tag of the user in a UWB positioning coordinate system at the time t; The position coordinate of the ith UWB positioning base station in a UWB positioning coordinate system is obtained; position coordinates of a reference base station in a UWB positioning coordinate system; And solving the arrival time difference equation set through an improved chan algorithm to obtain first coordinate data of the user, namely the position coordinates of the UWB positioning tag of the user in a UWB positioning coordinate system.

Description

Non-inductive man-machine interaction method and system integrating UWB and space sound field Technical Field The invention relates to the technical field of intelligent control, in particular to a non-inductive man-machine interaction method and system integrating UWB and a space sound field. Background In traditional man-machine interaction, a user needs to learn the language of a machine, such as learning to knock a keyboard, clicking an icon, sliding a screen, even remembering complex instructions, the machine can understand what the user needs to do, and with the development of technology, most of the machine equipment also realizes non-inductive man-machine interaction, the machine actively senses and understands gesture actions of people and the like, and then executes corresponding instructions to automatically provide services; however, in the existing non-inductive man-machine interaction, most of the non-inductive man-machine interaction needs a camera device to identify the user action, the efficiency is low, and in the interaction space, the situation that the user action cannot be identified due to the fact that the camera device cannot shoot is often shielded, so that a non-inductive man-machine interaction method through an Ultra Wide Band (UWB) technology is also generated, but in the existing non-inductive interaction based on UWB, the position positioning is inaccurate when the distance measurement error is present, and the identification of an interaction instruction is affected. Disclosure of Invention The invention provides a non-inductive man-machine interaction method for fusing UWB and space sound field, which solves the technical problems existing in the prior art and comprises the following steps: S1, establishing a UWB positioning coordinate system and an acoustic field coordinate system based on an interaction space, and performing space joint calibration on the UWB positioning coordinate system and the acoustic field coordinate system to obtain transformation parameters between the UWB positioning coordinate system and the acoustic field coordinate system, wherein the transformation parameters comprise a rotation matrix R1 and a translation vector T1; s2, in an interaction space, a user wears equipment with UWB positioning tags on hands, and based on a UWB positioning coordinate system and the UWB positioning tags of the user, first coordinate data of the user are obtained through real-time calculation, specifically: s21, using a test original ranging value sequence from the tested UWB tag to each UWB positioning base station and a corresponding test ranging error correction value as training samples to train a neural network model, taking an original ranging value between the current UWB positioning tag of a user and each UWB positioning base station as input, and outputting a corresponding ranging error correction value between the current UWB positioning tag of the user and each UWB positioning base station; s22, selecting one UWB positioning base station as a reference base station, and constructing an arrival time difference equation set by using the corrected ranging value: ; The method comprises the steps that at the time t, a corrected ranging value is formed between a UWB positioning tag of a user and an ith UWB positioning base station; At time t, a corrected ranging value between the UWB positioning tag of the user and the reference base station; to be at time t, the position coordinates of the UWB positioning tag of the user in the UWB positioning coordinate system The position coordinate of the ith UWB positioning base station in a UWB positioning coordinate system is obtained; position coordinates of a reference base station in a UWB positioning coordinate system; S23, solving the arrival time difference equation set through an improved chan algorithm to obtain first coordinate data of the user, namely, the position coordinates of the UWB positioning tag of the user in a UWB positioning coordinate system; S3, acquiring a first coordinate data sequence in the latest time period according to the step S2, acquiring a hand motion track of a user in the latest time period according to the first coordinate data sequence, and carrying out interactive instruction recognition and matching according to the hand motion track; And S4, controlling corresponding equipment to complete interaction instruction actions according to the matched interaction instructions, simultaneously converting the current first coordinate data of the user into a sound field coordinate system according to the rotation matrix R1 and the translation vector T1 to obtain corresponding second coordinate data, and controlling each loudspeaker to conduct sound directional projection to the second coordinate data through a phased array beam forming technology to broadcast interaction feedback voice. Further, the UWB positioning coordinate system is formed by disposing at least 4 UWB positioning base stat