CN-121978677-A - Upper door endowment service supervision system based on millimeter wave radar and multi-source sensing fusion

CN121978677ACN 121978677 ACN121978677 ACN 121978677ACN-121978677-A

Abstract

A home-in-care service supervision system based on the integration of millimeter wave radar and multi-source sensing comprises a millimeter wave radar, a microphone, a gyroscope, a satellite positioning module, a communication module, a loudspeaker, two processors, a battery, a memory card, a power key and a service key, wherein the home-in-care service supervision system realizes the home-in verification and service quality assessment of a worker, the home-in verification of the worker is completed through satellite positioning, the identification, positioning, tracking and liveness assessment of the worker and the old are completed through the combination of the millimeter wave radar monitoring algorithm and the microphone and the gyroscope, and finally the worker service quality assessment is formed.

Inventors

GU CHANGZHAN
LIU JINTAO
LI YUCHEN
LI DAIJIANG
YUAN XIANG

Assignees

上海希卡立科技有限公司
福寿康智慧医疗养老服务(上海)有限公司

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (4)

1. The upper door care service supervision system based on the fusion of millimeter wave radar and multi-source sensing realizes the upper door care verification and service quality assessment of a worker, and is characterized by comprising a millimeter wave radar, a microphone, a gyroscope, a satellite positioning module, a communication module, a loudspeaker, two processors, a battery, a memory card, a power key and a service key, wherein the upper door care verification of the worker is completed through satellite positioning, the identification, positioning, tracking and liveness assessment of the worker and the old are completed through a millimeter wave radar supervision algorithm, the service quality assessment of the worker is finally formed, the service scene switching judgment is performed through the gyroscope, the scene recording is completed through the microphone, and the supplement to the service quality assessment is formed.
2. The upper door endowment service supervision system based on the fusion of millimeter wave radar and multi-source sensing according to claim 1 is characterized in that a millimeter wave radar supervision algorithm comprises two parts of a point cloud processing and target generation algorithm and a service quality evaluation algorithm.
3. The system for supervising the top-door pension service based on the fusion of millimeter wave radar and multi-source sensing as set forth in claim 2, wherein the point cloud processing and target generation algorithm comprises defining a t-th frame point cloud, analyzing the t-th frame point cloud to obtain a point set Wherein For the nth point of the t-th frame, The historical target set is marked as the total point number of the t frame Wherein For the mth historical target of the t-th frame, The number of historical targets of the t frame is the number of historical targets of each historical target At least comprise a position Initial position Frame number of continuous matching Residual life cycle Target state 0 Means unacknowledged, 1 means acknowledged, and the trace ; 1) After the point cloud is input, analyzing the point cloud according to a point cloud packaging protocol, and obtaining a single-frame spherical coordinate system point cloud set: wherein For the distance of the radar to the nth point, For the azimuth angle, A Cartesian coordinate system point cloud set is obtained through coordinate transformation: ; 2) Boundary noise point filtering, namely, performing boundary noise filtering on point cloud, and eliminating points which do not meet the height constraint: rejecting, wherein d is the installation height, and only the point cloud in the core height range is reserved; 3) DBSCAN clustering, namely filtering out point cloud in plane Performing DBSCAN clustering on the obtained product to obtain Current target clusters and noise point sets: , wherein Representing the kth cluster and, The number of points in the cluster; Is a noise point set, namely a noise cluster; 4) Hungarian matching and noise matching: ① Extracting the position information of the current cluster and the historical target, namely extracting the position information of each current target cluster Calculating the average value of points in the cluster as the cluster position: the historical target position is Wherein ; ② Hungarian matching to obtain matched pair and unmatched set And (3) with Constructing a cost matrix, and completing matching by adopting a Hungary algorithm to obtain a matching pair set: , Representing historical targets With the current cluster Matching, co-blending And a plurality of matching pairs, wherein there is no matching history target set: There is no matching current target set: ; ④ Noise matching, noise density continuation, namely, all unmatched current target clusters Is added to the noise point set For each history object without matching In its position Radius of circumference Neighborhood statistical noise points of (1) Wherein Typical radius of the cluster of points, if: wherein Is the threshold value of the number of noise points, and the remaining life cycle is calculated Increase in Frames, and upper bound constraints: wherein Is the maximum life cycle upper limit; 5) Kalman filtering update, matching position update and track write, for each matching pair Taking the current cluster position as observation: And combining the historical target state of the previous frame, and updating the historical target position through Kalman filtering: And writing the new position to the target track: ; 6) Updating a target: ① Generating new historical target from unmatched current clusters, wherein each unmatched current target cluster Generating a new history object And initializing, letting : , Wherein the method comprises the steps of Initiating a life cycle for the candidate target and initializing a track ; ② Matched historical objective property update for each matched historical objective : , If it is And is also provided with Then: And sets its life cycle: wherein The number of matching frames required for the target validation, A life cycle for the identified target; ③ Unmatched historical objective Property update and removal for each unmatched historical objective : , If (if) Then from The history object is removed.
4. The system for supervising the upper door pension service based on the integration of millimeter wave radar and multi-source sensing as set forth in claim 2, wherein the quality of service evaluation algorithm is specifically defined to be at most one worker-protecting main target at any time With an elderly secondary target For each history object The following evaluation fields are added, namely a protection mark Sign for elder Candidate elderly count Near-worker count Anchor point for old people Anchor radius ; 1) The carer identifies updates, i.e. master target selects slave Selecting a worker-protecting main target And update : ① Construction of worker candidate set construction of candidate set The elements thereof satisfy: , And has "mobility" displacement constraints: wherein Is a displacement constraint threshold and satisfies a stability constraint: Or (b) Wherein Is the minimum continuous matching frame number requirement, and simultaneously requires Namely, the old people cannot be used as a worker; ② Triggering the reselection of the worker if any one of the following conditions is met, namely a) no worker is currently protected B) the current protector is removed in the present frame c) the current protector continuously mismatch such that its life cycle Wherein In order to protect the worker for the maximum life cycle, A maximum mismatch threshold for triggering reselection; ③ And (3) selecting a worker protection main target selection rule, namely selecting when reselection is triggered: wherein Selecting post-maintenance work protecting mark as weight: wherein Is the object of Is used for the protection mark of the tool, Representing maintenance worker trajectories for all: ; 2) Candidate elderly people identification, i.e. low moving range targets, identifying "candidate elderly people" for targets other than careers, and updating , ; ① Window length setting, if a worker exists, using a short window If the protector is not present, a long window is used ; ② Calculation of moving Range, taking target Recently, it has been proposed to The frame track point set is insufficient Frames do not participate in candidates: , Calculating a moving range: ,, ③ Candidate determination and count update if: wherein A movement range threshold value for the candidate decision, A survival threshold for candidate target counts, then: Otherwise: ; ④ Near the worker count, starting the worker protection, if the worker exists, calculating the distance between the worker and the target: , If it is Wherein For a distance threshold where the candidate target is close to the career, then: Otherwise: ; 3) Old people identification update, namely secondary target selection, namely selecting old people secondary targets from candidate old people And update ; ① The old man reselection triggering condition is that if any one of the following conditions is met, triggering reselection a) no old man exists currently B) the current elderly is removed or their life cycle C) the current elderly continuously mismatch such that the life cycle Wherein Is the maximum life cycle of the old people, The threshold value of the maximum mismatch of the aged for triggering reselection; ② The old man candidate set and the selection rule are that the old man candidate set is constructed: wherein Is candidate confirmation threshold value, if the worker is present, the worker is preferentially selected The largest ones: , If the worker is not present, the worker is preferentially selected Maximum, sub-selection The smallest ones: selecting and updating the old man mark: wherein Is the object of Is a sign of the old; ③ The old man anchor point establishment, namely when the old man target stably reaches a threshold value, setting an anchor point: wherein the anchor point radius is The method is used for stabilizing constraint of a follow-up leaning scene; 4) The worker protection activity assessment, namely the worker protection activity ACT is a parameter between 0 and 1, and the calculation method comprises the following steps: ; 5) Quality of service evaluation, evaluation window In, to protect the worker's track Calculating service index and outputting service quality: ① The movable range index: ; ② Average speed index: ; ③ Duty cycle of care pad , A distance threshold for effective service; ④ Tracking reliability penalty terms with caregivers' mismatched frame duty cycle or within a window A frame duty cycle below the threshold represents: ; ⑤ Quality of service scoring: wherein 、、、 The weights of the four indexes are respectively.

Description

Upper door endowment service supervision system based on millimeter wave radar and multi-source sensing fusion Technical Field The invention relates to the technology of the field of health care, in particular to an upper door care service supervision system integrating multi-source sensing fusion of millimeter wave radar, satellite positioning (Global navigation SATELLITE SYSTEM, GNSS), a microphone and a gyroscope. Background Currently, in the field of health care, the supervision of the home care services faces challenges, especially in the aspects of the caretaker's home verification and quality of service evaluation, and effective technical products are still lacking. Disclosure of Invention The invention aims at the difficult problem of supervision of the upper door pension service, and provides an upper door pension service supervision system (hereinafter referred to as supervision system) integrated with four sensors of millimeter wave radar, satellite positioning, microphone and gyroscope, so as to realize upper door verification and service quality assessment of a worker. The invention is realized by the following technical scheme: the invention is a self-powered portable system with built-in rechargeable batteries, the overall size is about 10cm multiplied by 2cm, and the self-powered portable system is carried by a caretaker to the home of the old. Firstly, aiming at the problem of checking the entrance of the worker, the monitoring system integrates a satellite positioning module, so that the position of the worker during the service of the worker can be positioned, and the entrance of the worker is ensured. Secondly, aiming at the evaluation of the service quality of the worker, the supervision system integrates three sensing modules of a millimeter wave radar, a microphone and a gyroscope. The millimeter wave radar is a core of the invention and mainly comprises millimeter wave radar hardware and millimeter wave radar algorithm, wherein the millimeter wave radar hardware generates millimeter wave Lei Dadian cloud data, the millimeter wave radar algorithm analyzes millimeter wave Lei Dadian cloud, clustering, target extraction, matching, kalman tracking, worker and old man identification and worker behavior identification are performed, and the service quality of the worker is evaluated through the movement track of the worker, the position relation between the worker and the old man and the activity of the worker. The microphone has recording and uploading functions, is used for recording the sound in the service process, and is an important supplement for confirming the service quality of the worker. The gyroscope is used for assisting the millimeter wave radar in service scene switching identification. Besides the three sensing modules, the monitoring system also integrates a communication module, and can upload monitoring data to the cloud end so as to realize unified management of multi-equipment information. Technical effects The invention integrally solves the problem of supervision of the nursing service of the nursing work upper door, solves the problem of checking the nursing work upper door through the satellite positioning module, and solves the problem of evaluating the service quality of the nursing work through the combination of the millimeter wave radar, the microphone and the gyroscope. Drawings Fig. 1 is a block diagram of a top-door pension service supervision system. Fig. 2 is a circuit diagram of the upper door pension service supervision system. Fig. 3 is a flow chart of the upper door pension service supervision system work. Fig. 4 is a schematic diagram of a millimeter wave radar circuit according to an embodiment, in which LAN refers to a Low-Noise Amplifier (Low-Noise Amplifier), PGA refers to a Programmable gain Amplifier (Programmable GAIN AMPLIFIER), PA refers to a Power Amplifier (Power Amplifier), ADC refers to an Analog-to-Digital Converter (Analog-to-Analog) converter, and FMCW Waveform Generator refers to a chirped continuous wave waveform generator. Fig. 5 is a flowchart of a millimeter wave radar algorithm. Detailed Description The upper door endowment service supervision system comprises a millimeter wave radar, a microphone, a gyroscope, a satellite positioning module, a communication module, a loudspeaker, two processors, a battery, a memory card, a power key and a service key, wherein the upper door endowment service supervision system is shown in fig. 1 and 2, the specific work flow of the supervision system is shown in fig. 3, the upper door of a worker is protected, after the power key is pressed, equipment is started, the worker is prompted to aim the equipment at a service scene through the loudspeaker, after the service key is pressed by the worker, the system starts millimeter wave radar supervision, recording, satellite positioning and timing functions, in the service process, the gyroscope judges that the equipment moves, the scene is switched,