Intelligent safety patrol robot for real-time positioning and risk early warning

Abstract

The application provides an intelligent safety patrol robot for real-time positioning and risk early warning, comprising a task scheduling module for determining a current patrol task and a target point task according to a patrol task issued by a background management terminal; a positioning module for acquiring current position information; a path planning module for generating a moving path according to the target point and the current position information; a moving module for moving according to the moving path; a collecting module for adapting to patrol task requirements according to the current patrol task, acquiring patrol collection data and fixed-point collection data; a recording module for storing the fixed-point collection data and generating a patrol record containing a time stamp and position information and uploading the record to the background management terminal; a voice and video interaction module for voice prompting or voice alarming. The application builds a robot for building safety patrol, which can replace property personnel to complete routine water, electricity, gas, security camera, fire-fighting equipment and building external patrol and other inspection tasks.

Description

Technical Field

[0001] This invention relates to the field of security patrol robot technology, and more specifically, to an intelligent security patrol robot with real-time positioning and risk warning. Background Technology

[0002] With the acceleration of urbanization, the scale of various residential communities, commercial complexes, and industrial parks is constantly expanding. Traditional security patrol methods mainly rely on security personnel conducting scheduled patrols at fixed times and locations. This approach suffers from high labor costs, limited patrol frequency, and difficulty in ensuring patrol quality at night and in inclement weather. Furthermore, human factors can easily lead to blind spots or missed inspections, making it difficult to promptly detect and address sudden safety hazards such as fires, burst water pipes, and gas leaks. In recent years, with the rapid development of robotics technology, various patrol robots have been gradually applied to the security field. However, existing patrol robots cannot adapt to the diverse patrol tasks within buildings, including those related to water, electricity, gas, and fire safety. Summary of the Invention

[0003] To address the shortcomings of the existing technology, the present invention aims to provide an intelligent security patrol robot with real-time positioning and risk warning capabilities, thereby overcoming the deficiencies in the prior art.

[0004] To achieve the above objectives, this invention provides an intelligent security patrol robot with real-time positioning and risk warning, comprising a task scheduling module, a positioning module, a path planning module, a movement module, a data acquisition module, a recording module, a communication module, and a voice and video interaction module. The task scheduling module establishes a connection with a backend management terminal via the communication module, enabling it to receive patrol tasks issued by the backend management terminal. The task scheduling module determines the current patrol task and its target location based on the patrol task. The positioning module is electrically and signal-connected to the task scheduling module, allowing the task scheduling module to send a first trigger signal to the positioning module after determining the target location. Upon receiving the first trigger signal, the positioning module... Upon triggering a signal, positioning is initiated and current location information is acquired. The path planning module is electrically and signal-connected to both the task scheduling module and the positioning module, enabling it to generate a movement path from the current location to the target location based on the target location provided by the task scheduling module and the current location information provided by the positioning module. The movement module is electrically and signal-connected to both the path planning module and the positioning module, allowing it to move according to the movement path provided by the path planning module. During movement, it performs path correction based on the real-time location information acquired by the positioning module and stops moving upon reaching the target location. The data acquisition module is electrically connected to the movement module, the task scheduling module, and the communication module. The system connects to a signal source, enabling the data acquisition module to adapt to the patrol task requirements based on the current patrol task sent by the task scheduling module. During robot movement, it initiates real-time along-path data acquisition to obtain data along the route. After reaching the target point and stopping, the movement module sends a second trigger signal to the acquisition module. Upon receiving the second trigger signal, the acquisition module initiates a fixed-point data acquisition action to obtain fixed-point data at the target point. The acquisition module processes and identifies both the along-path patrol data and the fixed-point data. Furthermore, when either the along-path patrol data or the fixed-point data is abnormal, the acquisition module reports to the backend management terminal via the communication module. Records are also provided. The recording module is electrically and signal-connected to the path planning module and the data acquisition module, respectively, so that the recording module stores the movement path generated by the path planning module, and under normal circumstances stores the fixed-point data acquired by the data acquisition module at the target location, and generates a patrol record containing timestamps and location information; the recording module also establishes a connection with the back-end management terminal through the communication module, so that the recording module uploads the generated patrol record to the back-end management terminal; the voice and video interaction module is electrically and signal-connected to the data acquisition module and the communication module, respectively, so that the data acquisition module identifies the patrol data along the route and the fixed-point data acquisition, and triggers the voice and video interaction module to provide voice prompts or voice alarms when an abnormality is identified.

[0005] Through the above technical solution, and by the organic coordination of the task scheduling module, positioning module, path planning module, movement module, data acquisition module, recording module, communication module, and voice and video interaction module, the robot achieves fully automated inspection from task reception, autonomous movement, along-path and fixed-point data acquisition to anomaly reporting. This constructs a robot for building security inspection, capable of replacing property management personnel in routine inspections such as water, electricity, gas, security cameras, fire equipment, and external building inspections. Specifically, the task scheduling module connects to the backend management terminal via the communication module, remotely receiving inspection tasks and accurately matching target locations without human intervention, thus automating the issuance and execution of inspection tasks. The positioning module obtains the robot's location information in real time, providing data support for path planning and movement correction, avoiding inspection errors caused by positioning deviations. The path planning module generates the optimal path based on the real-time location and target location, and the movement module moves autonomously along the path and makes real-time corrections, ensuring the robot accurately reaches the inspection points, completely replacing manual labor in repetitive and high-intensity inspection tasks. The data acquisition module adapts to patrol task requirements, balancing patrols along the route and fixed-point data collection at target locations, ensuring no blind spots in the patrol. It also intelligently processes and identifies anomalies in the collected data, distinguishing between normal patrol data and hazard data to avoid invalid data transmission and improve hazard identification efficiency. The recording module stores the patrol route and fixed-point data in real time, generating patrol records with timestamps and location information and uploading them to the backend, enabling traceability and verification of the entire patrol process. If any collected data is abnormal, it immediately reports to the backend via the communication module, simultaneously triggering alarms through the voice and video interaction modules, providing real-time hazard warnings, remote synchronous awareness, and on-site audio-visual alerts, thus improving the response speed to safety hazards.

[0006] As a further explanation of the intelligent security patrol robot with real-time positioning and risk warning described in this invention, preferably, the task scheduling module includes a control circuit board and a first memory disposed inside the robot, and a touch screen disposed on the robot; the control circuit board is configured with a task acquisition unit, a task queue management unit, a task execution control unit, and a status monitoring unit; wherein, the task acquisition unit is electrically and signal-connected to the communication module and the first memory respectively, so that the task acquisition unit receives patrol tasks issued from the background management terminal through the communication module and stores them in the first memory, the patrol tasks include at least one of water use patrol tasks, electricity use patrol tasks, gas patrol tasks, security camera patrol tasks, fire equipment patrol tasks, and building exterior patrol tasks, each patrol task also includes coordinate information of various target points, patrol frequency, and patrol standard parameters; the task queue management unit is electrically and signal-connected to the task acquisition unit. The system is connected to the task queue management unit, which receives patrol tasks sent by the task acquisition unit, sorts them according to their priority and preset time, and generates a task queue. The task execution control unit is electrically and signal-connected to the task queue management unit, the first memory, the path planning module, and the positioning module, respectively, so that the task execution control unit can obtain the currently pending task from the task queue management unit, obtain the coordinate information of the corresponding target point from the first memory according to the currently pending task and send it to the path planning module, and send a first trigger signal to the positioning module. The status monitoring unit is electrically and signal-connected to the task execution control unit and the touch screen, so that the status monitoring unit can display the patrol task on the touch screen according to the currently pending task of the task execution control unit, and control the task execution control unit to prioritize the execution of the patrol task set by the property personnel according to the patrol task set by the property personnel obtained from the touch screen.

[0007] Through the above technical solution, the task scheduling module is specifically set up as a task acquisition unit, a task queue management unit, a task execution control unit, and a status monitoring unit, realizing the unified reception, priority sorting, and dynamic execution of various types of patrol tasks. The task acquisition unit can receive diverse patrol tasks such as water, electricity, gas, and fire protection tasks issued from the backend. The task queue management unit sorts tasks according to priority and preset time to avoid multiple task conflicts, ensuring that emergency patrol tasks are executed first, and improving the logic and efficiency of patrol task execution. The task execution control unit accurately extracts the coordinates of target points and simultaneously triggers the positioning module to start, ensuring seamless connection between patrol tasks and positioning and route planning, eliminating task execution deviations. At the same time, the touch screen can intuitively display patrol tasks, facilitating on-site viewing and manual setting of emergency tasks by property personnel, achieving a combination of remote scheduling and on-site control, and improving the flexibility of equipment operation. The status monitoring unit tracks the progress of task execution throughout the process and provides real-time feedback on equipment operating status, enabling backend and on-site personnel to promptly grasp the patrol dynamics and quickly intervene in and handle abnormal situations.

[0008] As a further explanation of the intelligent security patrol robot with real-time positioning and risk warning described in this invention, preferably, the positioning module includes a positioning processor and a Beidou / GPS dual-mode positioning chip disposed inside the robot, a lidar and a vision sensor fixed on the top of the robot, an inertial measurement unit mounted on the robot chassis, and an encoder disposed on the robot's drive wheels; the positioning processor is configured with a task semantic parsing unit, a positioning requirement generation unit, a positioning scene recognition unit, a positioning information generation unit, and a positioning information output unit; wherein, the task semantic parsing unit is electrically and signal-connected to the task scheduling module and the Beidou / GPS dual-mode positioning chip respectively, so as to enable task semantic parsing The unit receives a first trigger signal with a target point from the task scheduling module, extracts the regional attributes of the target point, and generates a first enable signal based on the regional attributes, which is then sent to the BeiDou / GPS dual-mode positioning chip. This enables the BeiDou / GPS dual-mode positioning chip to start and acquire the robot's initial geographic location coordinates. The positioning requirement generation unit is electrically and signal-connected to the task semantic parsing unit, so that the positioning requirement generation unit receives the regional attributes extracted by the task semantic parsing unit to generate positioning strategy parameters. The positioning scene cognition unit is electrically and signal-connected to the positioning requirement generation unit, the LiDAR, the vision sensor, the inertial measurement unit, and the encoder, respectively, to enable the positioning... The scene recognition unit acquires point cloud data collected by the LiDAR, image data collected by the vision sensor, inertial navigation data collected by the inertial measurement unit, and odometer data collected by the encoder in real time during robot movement, as well as the positioning strategy parameters generated by the positioning requirement generation unit. It then performs feature extraction and scene recognition on the point cloud data, image data, inertial navigation data, and odometer data to generate a current scene type identifier. The positioning information generation unit is electrically and signal-connected to the positioning requirement generation unit, the scene recognition unit, the BeiDou / GPS dual-mode positioning chip, the LiDAR, the vision sensor, the inertial measurement unit, and the encoder, enabling the positioning information generation unit to receive... The initial geographic location coordinates from the BeiDou / GPS dual-mode positioning chip are combined with inertial navigation data collected by the inertial measurement unit and mileage data collected by the encoder to calculate the trajectory and generate the current location information, which is then sent to the path planning module. During the robot's movement, the fusion algorithm weights are dynamically adjusted based on the positioning strategy parameters generated by the positioning requirement generation unit and the current scene type identifier generated by the positioning scene cognition unit. Adaptive fusion calculation is then performed using real-time positioning data from the BeiDou / GPS dual-mode positioning chip, point cloud data collected by the LiDAR, image data collected by the visual sensor, inertial navigation data collected by the inertial measurement unit, and mileage data collected by the encoder to generate real-time location information.The positioning information output unit is electrically and signal-connected to the positioning information generation unit, the path planning module, and the motion module, respectively. This allows the positioning information output unit to receive real-time position information generated by the positioning information generation unit and send it to the path planning module for path correction, while simultaneously sending it to the motion module to control robot movement. Furthermore, when the robot reaches the target location, the positioning information output unit sends a position confirmation signal to the motion module to stop the robot's movement.

[0009] Through the above technical solution, the positioning module is configured to integrate BeiDou / GPS dual-mode positioning with multi-source fusion of LiDAR, visual sensors, inertial measurement units, and encoders. Combined with task semantic parsing and scene cognition, the indoor and outdoor positioning weights are dynamically adjusted based on task attributes and the current scene, achieving high-precision positioning with seamless switching between indoor and outdoor environments. The positioning information is transmitted to the path planning module and the motion module in real time, enabling dynamic path correction during movement and ensuring that the robot travels along the optimal path. Upon reaching the target location, a precise positioning signal is sent to control the robot to stop moving, ensuring the accuracy of fixed-point data collection.

[0010] As a further explanation of the intelligent security patrol robot with real-time positioning and risk warning described in this invention, preferably, the acquisition module includes an acquisition processor installed inside the robot, a visible light camera and an infrared thermal imager fixed to the top of the robot, a water immersion sensor installed at the bottom of the robot, a sound acquisition unit and a multi-gas sensor array installed at the front of the robot, and an ultrasonic sensor array installed on the side of the robot; the acquisition processor is configured with a patrol task matching unit, multiple patrol acquisition units, and a risk warning processing unit; wherein, the patrol task matching unit is electrically and signal-connected to the task scheduling module and the multiple patrol acquisition units, so that the patrol task matching unit sends a start command to the corresponding target patrol acquisition unit according to the current patrol task sent by the task scheduling module; each patrol acquisition unit is divided according to the patrol task type. The robot is electrically and signal-connected to at least one of the following: a visible light camera, an infrared thermal imager, a water immersion sensor, a sound collector, a multi-gas sensor array, and an ultrasonic sensor array. This allows the robot to acquire patrol data and fixed-point collection data along the route and upon reaching the target location, based on the risk type of the target object. The risk warning processing unit is electrically and signal-connected to each patrol and collection unit, the voice and video interaction module, and the communication module. This allows the risk warning processing unit to determine the warning level based on the data collected by each patrol and collection unit, and then provide voice prompts or alarms through the voice and video interaction module, and report to the back-end management terminal through the communication module.

[0011] Through the aforementioned technical solution, and by configuring various sensors such as visible light cameras, infrared thermal imagers, water immersion sensors, sound collectors, and multi-gas sensor arrays, the robot possesses multimodal perception capabilities, covering a variety of safety hazards including overheating from electricity, water leaks, gas leaks, abnormal noises, and obstructions, achieving comprehensive safety monitoring. The patrol task matching unit activates the corresponding data collection unit based on the current task type, such as setting up data collection units for electricity patrol, water patrol, gas patrol, security camera patrol, fire equipment patrol, and building exterior patrol. This enables on-demand access to sensing resources, avoids ineffective sensor operation, reduces equipment power consumption, and improves the relevance of collected data. The risk warning processing unit determines the warning level based on the collected data and triggers differentiated alarm responses, while simultaneously uploading the data to the backend, achieving accurate hazard warnings and tiered handling, avoiding false alarms and missed alarms, and improving the reliability of warnings.

[0012] As a further explanation of the intelligent security patrol robot with real-time positioning and risk warning described in this invention, preferably, the data acquisition module also includes a fire detection unit. The fire detection unit is electrically and signal-connected to a visible light camera, an infrared thermal imager, and a multi-gas sensor array mounted on the robot, respectively, so that fire determination can be made based on image data acquired by the visible light camera, temperature data acquired by the infrared thermal imager, and smoke concentration data acquired by the multi-gas sensor array during the robot's movement. When the fire detection unit determines a fire, a fire response action is executed: the task scheduling module interrupts the current patrol task, the path planning module re-plans the path based on the fire location provided by the positioning module, and the movement module controls the robot to move to a preset safe distance from the fire scene. Simultaneously, voice and video communication are conducted. The interaction module activates a fire alarm voice and broadcasts the location of nearby fire-fighting equipment. It also sends a fire warning containing on-site video and location information to the back-end management terminal via the communication module, and stops the alarm upon receiving confirmation from the back-end management terminal. When the fire detection unit determines a suspected fire, the task scheduling module suspends the current patrol task, the path planning module generates a confirmed path to the smoke source, the movement module controls the robot to move to the smoke source, and the fire detection unit performs a secondary determination using image data collected by a visible light camera and temperature data collected by an infrared thermal imager. If the secondary determination confirms a fire, the fire response action is executed. If the secondary determination rules out a fire, the task scheduling module resumes the original patrol task, and the path planning module replans a path to the original target location to continue the patrol task.

[0013] The above technical solution integrates multi-dimensional data such as images, temperature, and smoke concentration to determine fires, avoiding misjudgments from single sensors and improving the accuracy of fire identification. For suspected fires, the robot automatically proceeds to verify and re-evaluate, preventing false alarms from interfering with normal patrols and avoiding missed reports that could lead to safety accidents. Upon confirmation of a fire, the robot immediately interrupts its routine patrols, quickly reaches the safe area on-site, activates audible and visual alarms, broadcasts the location of fire-fighting equipment, and simultaneously uploads on-site video and location information. This enables rapid early warning, on-site alerts, and remote reporting of fire hazards, buying valuable time for personnel evacuation and firefighting operations.

[0014] As a further explanation of the intelligent security patrol robot with real-time positioning and risk warning described in this invention, preferably, the acquisition module also includes a water leakage detection unit. The water leakage detection unit is electrically and signal-connected to a visible light camera, a water immersion sensor, and a sound collector mounted on the robot, respectively, so that water leakage can be determined using image data acquired by the visible light camera, humidity data acquired by the water immersion sensor, and sound data of flowing water acquired by the sound collector during the robot's movement. When the water leakage detection unit determines a leak, a water leakage response action is executed: the task scheduling module interrupts the current patrol task, the path planning module re-plans the path based on the location of the leak point provided by the positioning module, and the movement module controls the robot to move to a preset distance from the leak site. Simultaneously, voice and video communication are integrated. The interaction module activates a water leakage voice alarm and sends a water leakage warning containing on-site video and location information to the back-end management terminal via the communication module. The alarm stops only after receiving confirmation from the back-end management terminal. When the water leakage detection unit determines that a water leakage is suspected, the task scheduling module suspends the current patrol task, the path planning module generates a confirmed path to the suspected water leakage point, the movement module controls the robot to move to the suspected water leakage point, and the water leakage detection unit performs a secondary judgment by collecting on-site image data through a visible light camera and collecting humidity data through a water immersion sensor. If the secondary judgment confirms a water leakage, the water leakage response action is executed. If the secondary judgment rules out a water leakage, the task scheduling module resumes the original patrol task, and the path planning module replans a path to the original target point to continue the patrol task.

[0015] By combining image, humidity, and water flow sound data, the above technical solution accurately identifies potential leaks such as pipe ruptures and seepage. It automatically performs secondary verification on suspected leak points, distinguishing between normal water accumulation and leaks, thus reducing false alarms. Upon confirmation of a leak, the robot immediately initiates an emergency procedure, interrupting routine patrols, docking at a safe location, issuing an alarm, and uploading hazard information. This facilitates rapid on-site response by maintenance personnel, preventing the leak from escalating and causing property damage.

[0016] As a further explanation of the intelligent security patrol robot with real-time positioning and risk warning described in this invention, preferably, the communication module is configured with a homeowner-side interaction unit, so that the task scheduling module sends an in-home patrol appointment notification and patrol time information to the corresponding homeowner terminal through the communication module, and after receiving the confirmation instruction from the homeowner terminal, the path planning module plans a dedicated patrol route to the target location based on the real-time location of the positioning module.

[0017] Through the above technical solution, the communication module adds a homeowner-side interaction unit to realize a humanized appointment management system for home visits. The robot can send appointment notifications and time slot information to the homeowner's terminal. After the homeowner confirms, it plans a dedicated home visit route, taking into account both security patrols and homeowner privacy, thereby improving residents' acceptance.

[0018] As a further explanation of the intelligent security patrol robot with real-time positioning and risk warning described in this invention, preferably, the recording module includes a second memory and a data cleaning unit. The data cleaning unit is electrically connected to the communication module and the second memory, respectively, so that after the recording module uploads the patrol records to the back-end management terminal through the communication module, it receives the upload confirmation instruction, data verification instruction, or deletion instruction from the back-end management terminal. Then, the data cleaning unit performs the corresponding operation according to the back-end feedback instruction, including automatically cleaning up the patrol records that have passed verification after a preset time, marking and temporarily storing the patrol records that have failed verification, and immediately deleting the patrol records that have received the deletion instruction.

[0019] Through the above technical solution, the recording module adds a data cleaning unit. This unit verifies, retains, and cleans the inspection records according to backend instructions. Verified records are automatically cleaned periodically to free up storage space; abnormal records are marked and temporarily stored for later review and verification; and records are immediately deleted upon receiving a deletion command, achieving refined data management. After uploading to the backend, a confirmation command is received to ensure the complete upload of inspection records, preventing data loss. Simultaneously, it achieves synchronized management of local and backend data, improving data security and traceability.

[0020] The beneficial effects of this invention are as follows: Through the organic coordination of a task scheduling module, a positioning module, a path planning module, a movement module, a data acquisition module, a recording module, a communication module, and a voice and video interaction module, this invention achieves fully automated inspection of the robot from task reception, autonomous movement, along-path and fixed-point data acquisition to anomaly reporting. This constructs a robot for building security inspections, capable of replacing property management personnel in routine inspections such as water and electricity, gas, security cameras, fire equipment, and external building inspections. Specifically, the task scheduling module connects to the backend management terminal via the communication module, remotely receiving inspection tasks and accurately matching target locations without human intervention, thus automating the issuance and execution of inspection tasks. The positioning module obtains the robot's location information in real time, providing data support for path planning and movement correction, avoiding inspection errors caused by positioning deviations. The path planning module generates the optimal path based on the real-time location and target location, and the movement module moves autonomously along the path and corrects in real time, ensuring the robot accurately reaches the inspection points, completely replacing manual labor in repetitive and high-intensity inspection tasks. The data acquisition module adapts to patrol task requirements, balancing patrols along the route and fixed-point data collection at target locations, ensuring no blind spots in the patrol. It also intelligently processes and identifies anomalies in the collected data, distinguishing between normal patrol data and hazard data to avoid invalid data transmission and improve hazard identification efficiency. The recording module stores the patrol route and fixed-point data in real time, generating patrol records with timestamps and location information and uploading them to the backend, enabling traceability and verification of the entire patrol process. If any collected data is abnormal, it immediately reports to the backend via the communication module, simultaneously triggering alarms through the voice and video interaction modules, providing real-time hazard warnings, remote synchronous awareness, and on-site audio-visual alerts, thus improving the response speed to safety hazards. Attached Figure Description

[0021] Figure 1 This is a block diagram of the overall structure of the intelligent security patrol robot with real-time positioning and risk warning of the present invention; Figure 2 This is a structural block diagram of the task scheduling module of the present invention; Figure 3 This is a structural block diagram of the positioning module of the present invention; Figure 4 This is a structural block diagram of the acquisition module of the present invention. Detailed Implementation

[0022] To further understand the structure, features, and other objectives of the present invention, a detailed description is provided below with reference to the accompanying drawings. The embodiments illustrated in these drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0023] This invention discloses an intelligent security patrol robot with real-time positioning and risk warning capabilities, primarily used for building security patrols, such as those involving water, electricity, gas, security cameras, fire-fighting equipment, and external building inspections. The intelligent security patrol robot can adopt a conventional wheeled mobile body design, similar to a food delivery robot, or other existing robot body designs on the market. The internal body houses the main control circuit board and core hardware modules. Positioning and data acquisition sensors are mounted on the top, movement and positioning auxiliary components are located on the chassis, interaction and detection sensors are located at the front, and obstacle avoidance sensors are located on the sides. All modules communicate via an internal bus. The specific module configuration is as follows: Figure 1 As shown, the intelligent security patrol robot mainly includes a task scheduling module 1, a positioning module 2, a path planning module 3, a movement module 4, a data acquisition module 5, a recording module 6, a communication module 7, and a voice and video interaction module 8.

[0024] In this embodiment, the task scheduling module 1 establishes a connection with the backend management terminal 9 through the communication module 7. The communication module 7 is used to communicate with the backend management terminal 9, receiving patrol tasks issued by the backend management terminal 9 and then sending them to the task scheduling module 1. The task scheduling module 1 can determine the current patrol task and its target location based on the patrol task. The positioning module 2 is electrically and signal-connected to the task scheduling module 1. After determining the target location, the task scheduling module 1 sends a first trigger signal to the positioning module 2. Upon receiving the first trigger signal, the positioning module 2 starts positioning and obtains the current location information. The path planning module 3 is electrically and signal-connected to both the task scheduling module 1 and the positioning module 2. The path planning module 3 generates a movement path from the current location to the target location based on the target location issued by the task scheduling module 1 and the current location information provided by the positioning module 2. The movement module 4 is electrically and signal-connected to both the path planning module 3 and the positioning module 2. The movement module 4 moves according to the movement path issued by the path planning module 3. During the movement, it performs path correction based on the real-time location information obtained by the positioning module 2 and stops moving when it reaches the target location. In this embodiment, the task scheduling module 1 connects to the backend management terminal 9 through the communication module 7, which can remotely receive inspection tasks and accurately match target locations without manual on-site intervention, realizing the automated issuance and execution of inspection tasks; the positioning module 2 obtains the robot's position information in real time, providing data support for path planning and movement correction, avoiding inspection errors caused by positioning deviations; the path planning module 3 generates the optimal path by combining the real-time position and target location, and the movement module 4 moves autonomously along the path and makes real-time corrections, ensuring that the robot accurately arrives at the inspection location, completely replacing manual labor to complete repetitive and high-intensity inspection tasks.

[0025] In this embodiment, the acquisition module 5 is electrically and signal-connected to the movement module 4, the task scheduling module 1, and the communication module 7, respectively. The acquisition module 5 adapts to the patrol task requirements based on the current patrol task sent by the task scheduling module 1, and initiates a real-time patrol acquisition action along the route during robot movement to obtain patrol data along the route. After the movement module 4 reaches the target point and stops moving, it sends a second trigger signal to the acquisition module 5. Upon receiving this second trigger signal, the acquisition module 5 initiates a fixed-point task acquisition action to obtain fixed-point acquisition data for the target point. The acquisition module 5 processes and identifies the patrol data along the route and the fixed-point acquisition data. If the patrol data along the route or the fixed-point acquisition data is abnormal, the acquisition module 5 reports directly to the backend management terminal 9 via the communication module 7. The recording module 6 is electrically and signal-connected to the path planning module 3 and the acquisition module 5, respectively. The recording module 6 stores the movement path generated by the path planning module 3, and under normal circumstances, stores the fixed-point acquisition data of the acquisition module 5 at the target point, and generates a patrol record containing timestamps and location information. The recording module 6 also establishes a connection with the back-end management terminal 9 through the communication module 7, enabling the recording module 6 to upload the generated inspection records to the back-end management terminal 9. The voice and video interaction module 8 is electrically and signal-connected to the acquisition module 5 and the communication module 7, respectively, enabling the acquisition module 5 to identify patrol data along the route and fixed-point acquisition data. When an anomaly is detected, the acquisition module 5 triggers the voice and video interaction module 8 to provide voice prompts or alarms. The voice and video interaction module 8 establishes a connection with the back-end management terminal 9 through the communication module 7, allowing the robot to activate video and voice, allowing the back-end management terminal 9 to view the robot's moving perspective image or communicate with other personnel at the robot's location. The acquisition module 5 in this implementation can adapt to the needs of inspection tasks, balancing patrol along the route during movement with fixed-point acquisition at target locations, achieving inspection without blind spots; simultaneously, it intelligently processes and identifies anomalies in the acquired data, distinguishing between normal inspection data and potential hazard data, avoiding invalid data transmission, and improving the efficiency of hazard identification. The recording module 6 stores the patrol route and fixed-point data in real time, generates patrol records with timestamps and location information, and uploads them to the backend, enabling the entire patrol process to be traceable and verifiable. Once the collected data is abnormal, it immediately reports to the backend through the communication module 7, and at the same time triggers the alarm of the voice and video interaction module 8, realizing real-time early warning of hidden dangers, remote synchronous awareness, and on-site audio and visual prompts, thereby improving the response speed to safety hazards.

[0026] To achieve unified reception, prioritization, and dynamic execution of various types of patrol tasks, in some embodiments, such as Figure 2As shown, the task scheduling module 1 preferably includes a control circuit board 11 and a first memory 12 disposed inside the robot, and a touch screen 13 disposed on the robot. The control circuit board 11 is configured with a task acquisition unit 111, a task queue management unit 112, a task execution control unit 113, and a status monitoring unit 114. The control circuit board 11 preferably uses a GD32 F4 / F5 series MCU from GigaDevice Semiconductor Co., Ltd. as the core processor (such as the GD32 F405, which has a built-in 168MHz Cortex-M4 core, 512KB Flash, 192KB SRAM, and is equipped with 6 UART, 3 SPI, 3 I2C, Ethernet and USB communication interfaces). The task acquisition unit 111 realizes data reception and storage control through the MCU's UART, SPI, Ethernet or USB communication peripherals. The task queue management unit 112 realizes task sorting and queue management through the MCU's timers, SRAM and interrupt resources. The task execution control unit 113 realizes task scheduling and peripheral triggering through the MCU's general I / O, timers, communication peripherals and address bus. The status monitoring unit 114 realizes status display and on-site interactive control through the MCU's display driver interface (such as SPI / MIPI), touch detection interface, timers and SRAM.

[0027] In this embodiment, the task acquisition unit 111 is electrically and signal-connected to the communication module 7 and the first memory 12, respectively, so that the task acquisition unit 111 receives the inspection tasks issued by the background management terminal 9 through the communication module 7 and stores them in the first memory 12. The task acquisition unit 111 can receive diversified inspection tasks issued by the background, such as water use, electricity use, gas use, and fire protection. That is, the inspection tasks include at least one of water use inspection tasks, electricity use inspection tasks, gas use inspection tasks, security camera inspection tasks, fire protection equipment inspection tasks, and building exterior inspection tasks. Each inspection task also includes the coordinate information of various target points, inspection frequency, and inspection standard parameters.

[0028] In this embodiment, the task queue management unit 112 is electrically and signal connected to the task acquisition unit 111 so that the task queue management unit 112 receives the patrol tasks sent by the task acquisition unit 111, sorts them according to the priority and preset time of the patrol tasks and generates a task queue, avoids multi-task conflicts, ensures that emergency patrol tasks are executed first, and improves the logic and efficiency of patrol task execution.

[0029] In this embodiment, the task execution control unit 113 is electrically and signal-connected to the task queue management unit 112, the first memory 12, the path planning module 3, and the positioning module 2, respectively. This enables the task execution control unit 113 to obtain the current task to be executed from the task queue management unit 112, obtain the coordinate information of the corresponding target point from the first memory 12 according to the current task to be executed and send it to the path planning module 3, and send a first trigger signal to the positioning module 2 to synchronously trigger the positioning module 2 to start, ensuring seamless connection between the patrol task and the positioning and path planning links, and eliminating task execution deviations.

[0030] In this embodiment, the status monitoring unit 114 is electrically and signal-connected to the task execution control unit 113 and the touch screen 13, respectively. This allows the status monitoring unit 114 to display patrol tasks on the touch screen 13 based on the current tasks to be executed by the task execution control unit 113. The touch screen provides a clear view of the patrol tasks, facilitating on-site viewing by property management personnel and allowing them to manually set emergency tasks. This combines remote dispatching with on-site management, enhancing the flexibility of equipment operation. Furthermore, the status monitoring unit 114 controls the task execution control unit 113 to prioritize the execution of patrol tasks set by property management personnel, based on the patrol tasks obtained from the touch screen 13. The status monitoring unit 114 tracks the task execution progress throughout the process and provides real-time feedback on the equipment's operating status, enabling back-end and on-site personnel to promptly grasp the patrol dynamics and quickly intervene in any abnormal situations.

[0031] To ensure high-precision positioning with seamless switching between indoor and outdoor environments, in some embodiments, such as Figure 3 As shown, the positioning module 2 preferably includes a positioning processor 21 and a Beidou / GPS dual-mode positioning chip 22 (such as the GK9501 from Guoke Microelectronics) installed inside the robot, a lidar 23 (such as the RPLIDAR S3 from SLAMTEC) and a vision sensor 24 (such as the V3 Pro AI vision sensor from McMahon) fixed on the top of the robot, an inertial measurement unit 25 (such as the MUS600 inertial measurement unit from Beijing Micro-Yuan Times Technology Co., Ltd.) mounted on the robot chassis, and an encoder 26 (such as the absolute rotary encoder from Leadshine Intelligent) installed on the robot's drive wheels. The positioning processor 21 is equipped with a task semantic parsing unit 211, a positioning requirement generation unit 212, a positioning scene recognition unit 213, a positioning information generation unit 214, and a positioning information output unit 215.

[0032] In this embodiment, the task semantic parsing unit 211 is electrically and signal-connected to the task scheduling module 1 and the Beidou / GPS dual-mode positioning chip 22, respectively, so that the task semantic parsing unit 211 receives a first trigger signal with a target location sent from the task scheduling module 1, extracts the regional attributes (outdoor or indoor) of the target location, and generates a first enable signal based on the regional attributes and sends it to the Beidou / GPS dual-mode positioning chip 22, thereby enabling the Beidou / GPS dual-mode positioning chip 22 to start and obtain the robot's initial geographical location coordinates.

[0033] In this embodiment, the positioning requirement generation unit 212 is electrically and signal-connected to the task semantic parsing unit 211, so that the positioning requirement generation unit 212 receives the region attributes extracted by the task semantic parsing unit 211 to generate positioning strategy parameters. These positioning strategy parameters include initial confidence values ​​for each sensor and sampling frequency configurations.

[0034] In this embodiment, the positioning scene cognition unit 213 is electrically and signal connected to the positioning requirement generation unit 212, the lidar 23, the vision sensor 24, the inertial measurement unit 25, and the encoder 26, respectively, so as to enable the positioning scene cognition unit 213 to acquire point cloud data collected by the lidar 23, image data collected by the vision sensor 24, inertial navigation data collected by the inertial measurement unit 25, and mileage data collected by the encoder 26 in real time during the robot's movement. The positioning scene cognition unit 213 performs feature extraction and scene recognition on the point cloud data, the image data, the inertial navigation data, and the mileage data according to the positioning strategy parameters generated by the positioning requirement generation unit 212 to generate a current scene type identifier (outdoor or indoor identifier).

[0035] In this embodiment, the positioning information generation unit 214 is electrically and signal-connected to the positioning requirement generation unit 212, the positioning scene cognition unit 213, the Beidou / GPS dual-mode positioning chip 22, the lidar 23, the vision sensor 24, the inertial measurement unit 25, and the encoder 26, respectively. This enables the positioning information generation unit 214 to receive the initial geographic location coordinates from the Beidou / GPS dual-mode positioning chip 22, and combine them with the inertial navigation data collected by the inertial measurement unit 25 and the mileage data collected by the encoder 26 to perform trajectory estimation, generate the current position information, and send it to the path planning module 3. Trajectory estimation is a prior art technique, based on a recursive positioning method combining inertial navigation and odometer with an inertial measurement unit (IMU) and a wheel encoder. It is a conventional technique in this field. Its basic principle is: starting from the initial position, the attitude is obtained by integrating the acceleration / angular velocity, and combined with the wheel mileage increment, the continuous position of the robot is recursively calculated to achieve high-frequency continuous positioning output in scenarios without GNSS or laser matching. During robot movement, the positioning information generation unit 214 dynamically adjusts the fusion algorithm weights based on the positioning strategy parameters generated by the positioning requirement generation unit 212 and the current scene type identifier generated by the positioning scene cognition unit 213. It then performs adaptive fusion calculations to generate real-time location information by combining real-time positioning data from the BeiDou / GPS dual-mode positioning chip 22, point cloud data collected by the LiDAR 23, image data collected by the vision sensor 24, inertial navigation data collected by the inertial measurement unit 25, and odometer data collected by the encoder 26. In other words, through multi-source fusion of BeiDou / GPS dual-mode positioning with LiDAR, vision sensor, inertial measurement unit, and encoder, combined with task semantic parsing and scene cognition, and by dynamically adjusting indoor and outdoor positioning weights based on task attributes and the current scene, high-precision positioning with seamless indoor / outdoor switching is achieved. Adaptive fusion calculation is an existing technology; specifically, it can employ any of the commonly used multi-source data fusion algorithms such as Kalman filter (KF), extended Kalman filter (EKF), unscented Kalman filter (UKF), or particle filter. This embodiment does not improve the fusion algorithm itself, but only reuses the existing mature fusion framework. Based on the positioning scene type and the real-time confidence level of each sensor, it dynamically adjusts the weight coefficients of each sensor data in the fusion solution, making the positioning result more consistent with the current indoor / outdoor / transitional scene, achieving high-precision, continuous, and stable positioning output. Dynamically adjusting the fusion algorithm weights means that the positioning information generation unit 214 adjusts the weight coefficients of BeiDou / GPS, LiDAR, visual sensor, IMU, and encoder in the fusion solution according to preset rules based on the scene type identifier (outdoor / indoor / transitional area) output by the positioning scene cognition unit 213 and the real-time signal quality of each sensor.For example, in outdoor scenarios (with good GNSS signal), the weights are set as follows: BeiDou / GPS weight: 0.7~0.9, LiDAR weight: 0.1~0.2, visual sensor weight: 0.05~0.1, and IMU+encoder weight: 0.05 (for smoothing only). In indoor scenarios (without GNSS signal), the weights are set as follows: BeiDou / GPS weight: 0, LiDAR weight: 0.6~0.7, visual sensor weight: 0.2~0.25, and IMU+encoder weight: 0.05~0.15. In transitional scenarios (with weak / unstable GNSS signal), the weights are set as follows: BeiDou / GPS weight: linearly decreases from 0.7 to 0 (or vice versa), LiDAR weight: linearly increases from 0.1 to 0.7 (or vice versa), and the weights of other sensors are smoothly transitioned proportionally to avoid positioning jumps.

[0036] In this embodiment, the positioning information output unit 215 is electrically and signal-connected to the positioning information generation unit 214, the path planning module 3, and the movement module 4, respectively. This enables the positioning information output unit 215 to receive real-time location information generated by the positioning information generation unit 214 in real time. The positioning information output unit 215 then sends this real-time location information to the path planning module 3 for path correction, ensuring the robot travels along the optimal path. Simultaneously, the positioning information output unit 215 sends this real-time location information to the movement module 4 to control the robot's movement. When the robot reaches the target point, the positioning information output unit 215 sends a position confirmation signal to the movement module 4 to stop the robot's movement, ensuring the accuracy of the fixed-point data acquisition.

[0037] In this embodiment, the positioning processor 21 is preferably implemented using the IB3-771 industrial motherboard from Shenzhen Jiehe Technology Development Co., Ltd. This motherboard is equipped with an RK3588 processor and a 6TOPS NPU, and features multiple UART / RS485, Ethernet, GPIO, and CAN interfaces. It can simultaneously connect to the Beidou / GPS dual-mode positioning chip 22, LiDAR 23, visual sensor 24, inertial measurement unit 25, encoder 26, and task scheduling module 1, meeting the requirements for parallel acquisition, real-time processing, and scene recognition of multi-source positioning data. Specifically, the task semantic parsing unit 211 receives task signals via the UART / RS485 interface and controls GNSS startup via the GPIO port; the positioning requirement generation unit 212 stores strategy parameters in memory and distributes configurations via the internal bus; the positioning scene recognition unit 21 processes multi-sensor data through the NPU to complete scene recognition; the positioning information generation unit 21 performs trajectory calculation and adaptive fusion calculation through the FPU and NPU; and the positioning information output unit 215 outputs location information and arrival signals via Ethernet / UART / GPIO ports. The aforementioned units work together using the hardware resources of the IB3-771 motherboard to achieve high-precision positioning with seamless switching between indoor and outdoor environments.

[0038] To ensure the robot possesses multimodal perception capabilities, covering various safety hazards such as overheating, water leaks, gas leaks, abnormal noises, and obstacle obstructions, and achieving comprehensive safety monitoring, in some embodiments, such as Figure 4 As shown, the acquisition module 5 includes an acquisition processor 51 installed inside the robot, a visible light camera 52 (such as Hikvision MV-CA013-20GC) and an infrared thermal imager 53 (such as FLIR Lepton 3.5 infrared thermal imaging module) fixed to the top of the robot, a water immersion sensor 54 (such as the water immersion sensor from Shandong Renke Measurement and Control Technology Co., Ltd.) installed on the bottom of the robot, a sound acquisition device 55 (such as Hikvision DS-KAU70HG-S) and a multi-gas sensor array 56 (such as Hanwei Technology GS-A4, an industrial-grade four-in-one gas sensor that can simultaneously monitor CO, combustible gas, oxygen, and VOSc) installed on the front of the robot, and an ultrasonic sensor array 57 (such as the ultrasonic array probe from Ainttech Technology (Shanghai) Co., Ltd.) installed on the side of the robot. The acquisition processor 51 is equipped with a patrol task matching unit 511, multiple patrol acquisition units 512, and a risk warning processing unit 513.

[0039] In this embodiment, the patrol task matching unit 511 is electrically and signal-connected to the task scheduling module 1 and multiple patrol acquisition units 512, respectively, so as to enable the patrol task matching unit 511 to send a start command to the corresponding target patrol acquisition unit according to the current patrol task sent by the task scheduling module 1.

[0040] In this embodiment, each inspection and data collection unit 512 is electrically and signal connected to at least one of the following: a visible light camera 52, an infrared thermal imager 53, a water immersion sensor 54, a sound collector 55, a multi-gas sensor array 56, and an ultrasonic sensor array 57, depending on the inspection task type. This enables the robot to acquire patrol data along the route and fixed-point data during its movement and after reaching the target location, based on the risk type of the target object. Depending on the inspection task type, the inspection and data collection unit 512 can be configured as a water usage inspection and data collection unit, an electricity usage inspection and data collection unit, a gas usage inspection and data collection unit, a security camera inspection and data collection unit, a fire equipment inspection and data collection unit, and a building exterior inspection and data collection unit. For example, water usage inspection and data collection units are used for water usage inspections in buildings. Their targets include, but are not limited to, water tanks, water pumps, water meters, and water pipe wells. Corresponding risk types include whether the target's appearance is intact, its surface temperature, water meter readings, and the sound of flowing and dripping water in the pipes. Electricity inspection and data collection units are used for electricity usage inspections in buildings. Their targets include, but are not limited to, distribution boxes / cabinets, meter rooms / boxes, meters, power lines, charging piles, public area sockets / switches, elevators / water pumps, etc. Corresponding risk types include whether the target's appearance is intact, its surface temperature, water meter readings, and the sound of electric arcs in the power lines. Gas inspection and data collection units are used for gas pipeline inspections in buildings. Their targets include, but are not limited to, whether gas pipelines are leaking, valves, meters, fire doors, and alarms. Corresponding risk types include whether the target's appearance is intact, its surface temperature, and gas consumption. Security camera inspection and data collection units are set to target, but are not limited to, entrance / exit cameras, elevator car cameras, and corridor / hallway cameras. The risks associated with cameras such as those in underground parking garages, perimeter / wall cameras, high-altitude littering cameras, and access control systems include whether the target object's appearance is intact, whether the camera image is unclear, or whether there is obstruction. The targets for the fire equipment inspection and data collection unit include, but are not limited to, fire extinguishers, fire hydrants, sprinkler heads, detectors, and evacuation indicator lights. The risks associated with these targets include whether the target object's appearance is intact, whether it is damaged, and whether it is within its valid service life. The targets for the building exterior inspection and data collection unit include, but are not limited to, vehicles, manhole covers, ground surfaces, lighting, exterior walls, landscaping, and garbage. The risks associated with these targets include whether there are illegally parked vehicles, whether manhole covers are intact, whether the ground is damaged, whether lighting is damaged, whether there are loose objects on exterior walls, whether landscaping needs maintenance, and whether garbage has been promptly cleaned. These targets and their risk types correspond to those of routine property inspections. As the content of the robot's fixed-point data collection, this enables on-demand access to sensing resources, avoids ineffective sensor operation, reduces equipment power consumption, and improves the relevance of the collected data.

[0041] In this embodiment, the risk warning processing unit 513 is electrically and signal-connected to each patrol and data collection unit 512, the voice and video interaction module 8, and the communication module 7, respectively. This allows the risk warning processing unit 513 to determine the warning level based on the data collected by each patrol and data collection unit 512, and then provide voice prompts or alarms through the voice and video interaction module 8, and report to the backend management terminal 9 through the communication module 7. This achieves accurate warning and graded handling of potential hazards, avoiding false alarms and missed alarms, and improving the reliability of warnings. The risk warning processing unit 513 uses existing conventional threshold comparison and grading methods to determine the warning level. Specifically, it pre-stores corresponding Level 1 (emergency), Level 2 (severe), and Level 3 (general) warning thresholds in the data collection processor 51 according to different hazard types such as overheating, water leakage, gas leakage, abnormal noise, and obstruction. The risk warning processing unit 513 reads the collected data uploaded by each patrol and data collection unit 512 in real time, compares it with the preset thresholds, automatically determines the warning level based on the degree of exceedance, and executes graded alarms and reports. For example, in the event of overheating, a Level 3 warning (general) is issued when the equipment temperature is >50℃ and ≤60℃; a Level 2 warning (severe) is issued when the equipment temperature is >60℃ and ≤80℃; and a Level 1 warning (emergency) is issued when the equipment temperature is >80℃.

[0042] In this embodiment, the data acquisition processor 51 is preferably implemented using Advantech's UNO-2484G embedded fanless industrial computer, equipped with an Intel Celeron N4505 processor, 8GB DDR4 memory, and 32GB eMMC storage. It features multiple MIPI-CSI, USB 3.2, RS485, GPIO, and CAN interfaces, and can simultaneously connect to a visible light camera 52, an infrared thermal imager 53, a water immersion sensor 54, a sound collector 55, a multi-gas sensor array 56, an ultrasonic sensor array 57, and a task scheduling module 1, meeting the requirements for parallel acquisition, real-time processing, and risk warning of multimodal safety hazard data. The patrol task matching unit 51 receives task instructions via the RS485 / Ethernet interface and schedules the patrol acquisition unit 512 via GPIO and the internal bus. The patrol acquisition unit 512, relying on multiple sensor interfaces, drives the corresponding sensors to complete data acquisition according to the task type. The risk warning processing unit 51 analyzes the data and provides graded warnings through the processor and extended NPU, and reports to the backend via the communication module.

[0043] In some embodiments, the acquisition module 5 further includes a fire detection unit, which is electrically and signal-connected to the visible light camera 52, the infrared thermal imager 53, and the multi-gas sensor array 56 installed on the robot, respectively, so as to realize fire determination by image data acquired by the visible light camera 52, temperature data acquired by the infrared thermal imager 53, and smoke concentration data acquired by the multi-gas sensor array 56 during the robot's movement. It can also be used as a patrol acquisition unit 512 to acquire patrol data along the way.

[0044] When the fire detection unit 514 determines that a fire has occurred, it executes the fire response actions: the task scheduling module 1 interrupts the current patrol task, the path planning module 3 replans the path according to the fire location provided by the positioning module 2, the movement module 4 controls the robot to move to a preset safe distance from the fire scene, and at the same time the voice and video interaction module 8 activates the fire voice alarm and broadcasts the location of nearby fire-fighting equipment, and sends a fire warning containing on-site video and location information to the back-end management terminal 9 through the communication module 7, until the alarm stops after receiving the confirmation instruction from the back-end management terminal 9.

[0045] When the fire detection unit 514 determines that a fire is suspected, the task scheduling module 1 suspends the current patrol task, the path planning module 3 generates a confirmed path to the smoke source point, the movement module 4 controls the robot to move to the smoke source point, and the fire detection unit 514 performs a secondary determination using image data collected by the visible light camera 52 and temperature data collected by the infrared thermal imager 53; if the secondary determination confirms a fire, the fire response action is executed; if the secondary determination rules out a fire, the task scheduling module 1 resumes the original patrol task, and the path planning module 3 replans the path to the original target point to continue the patrol task.

[0046] In some embodiments, the acquisition module 5 may further include a leakage detection unit 515, which is electrically and signal-connected to the visible light camera 52, the water immersion sensor 54 and the sound collector 55 installed on the robot, so as to determine leakage by using image data acquired by the visible light camera 52, humidity data acquired by the water immersion sensor 54 and water sound data acquired by the sound collector 55 during the robot's movement. It also serves as a patrol acquisition unit 512 to acquire patrol data along the way.

[0047] When the leak detection unit 515 determines that there is a leak, it executes the leak response action: the task scheduling module 1 interrupts the current inspection task, the path planning module 3 replans the path according to the location of the leak point provided by the positioning module 2, the movement module 4 controls the robot to move to the preset distance of the leak site, and at the same time the voice and video interaction module 8 starts the leak voice alarm and sends a leak warning containing on-site video and location information to the back-end management terminal 9 through the communication module 7 until the alarm stops after receiving the confirmation instruction from the back-end management terminal 9.

[0048] When the leak detection unit 515 determines that there is a suspected leak, the task scheduling module 1 suspends the current inspection task, the path planning module 3 generates a confirmed path to the suspected leak point, and the movement module 4 controls the robot to move to the suspected leak point. The leak detection unit 515 collects on-site image data through the visible light camera 52 and collects humidity data through the water immersion sensor 54 for secondary judgment. If the secondary judgment confirms that there is a leak, the leak response action is executed. If the secondary judgment eliminates the leak, the task scheduling module 1 resumes the original inspection task, and the path planning module 3 replans the path to the original target point to continue the inspection task.

[0049] In some embodiments, particularly for situations requiring regular in-home meter checks (water, electricity, or gas), the communication module 7 is preferably equipped with a homeowner-side interaction unit. This allows the task scheduling module 1 to send an in-home inspection appointment notification and inspection time information to the corresponding homeowner terminal via the communication module 7. After receiving confirmation from the homeowner terminal, the path planning module 3, based on the real-time location of the positioning module 2, plans a dedicated inspection route to the target location, achieving humanized appointment management for in-home inspections while considering both security and homeowner privacy, thus increasing resident acceptance. When the robot comes to check the water, electricity, or gas meters, it can initiate video and voice calls with the backend management terminal 9 via the voice and video interaction module 8 to help complete the in-home inspection task. Since robots belonging to the building are more readily known and accepted by homeowners, it avoids safety concerns caused by unfamiliarity with property management personnel or gas company staff. The communication module 7 preferably uses Quectel's 5G RG500Q, a 5G / WiFi dual-mode industrial-grade module.

[0050] In some embodiments, the recording module 6 includes a second memory and a data cleaning unit. The data cleaning unit is electrically and signal-connected to the communication module 7 and the second memory, respectively. This allows the recording module 6 to upload inspection records to the backend management terminal 9 via the communication module 7, and then receive upload confirmation instructions, data verification instructions, or deletion instructions from the backend management terminal 9. The data cleaning unit then performs corresponding operations based on the backend feedback instructions, including automatically cleaning up inspection records that have passed verification after a preset time, marking and temporarily storing inspection records with abnormal verification, and immediately deleting inspection records upon receiving a deletion instruction. The data cleaning unit verifies, retains, and cleans inspection records according to backend instructions. Records that pass verification are automatically cleaned periodically to free up storage space; abnormal records are marked and temporarily stored for later review and verification; and records are immediately deleted upon receiving a deletion instruction, achieving refined data management. Receiving confirmation instructions after uploading to the backend ensures the complete upload of inspection records, avoids data loss, and simultaneously achieves synchronized management of local and backend data, improving data security and traceability.

[0051] In this embodiment, the intelligent security patrol robot is in standby mode after being fully charged, such as every morning. Property management personnel send the daily patrol task list to the robot through the back-end management terminal 9, such as conducting a fixed-point inspection of the fire hydrants in the central garden of the community at 9:00 am (fire equipment inspection task), inspecting the electrical boxes in the public areas of each floor of Building 3 at 10:00 am (electricity inspection task), monitoring for water leaks in the underground garage of Building 5 at 2:00 pm (water inspection task), and conducting in-home gas safety inspections according to the residents' appointments (gas inspection task).

[0052] The robot's communication module 7 listens for backend commands in real time and transmits the received task list to the task scheduling module 1. The task acquisition unit 111 within the task scheduling module 1 stores these tasks in the first memory 12. The task queue management unit 112 sorts the tasks according to their preset time (9:00 AM, 10:00 AM, 2:00 PM) and priority (household gas safety inspection has the highest priority), generating a task queue. The task execution control unit 113 retrieves the currently pending task (fire hydrant inspection at 9:00 AM) from the queue, retrieves the target coordinates of the fire hydrant inspection point (located in the southeast corner of the central garden) from the first memory 12, and sends a first trigger signal to the positioning module 2.

[0053] Meanwhile, the status monitoring unit 114 displays the current task information on the touch screen 13 on the robot body: "Task to be performed: Fire hydrant inspection, target location: Southeast corner of the central garden", which is convenient for on-site property staff to view. If property staff temporarily need the robot to perform other tasks first, they can set it directly on the touch screen 13, and the status monitoring unit 114 will control the task execution control unit 113 to respond first.

[0054] During the high-precision positioning and path planning performed during task execution, after the positioning module 2 receives the first trigger signal, the task semantic parsing unit 221 extracts the regional attributes of the target point (such as "open outdoor area"), and then generates the first enable signal to activate the Beidou / GPS dual-mode positioning chip 22 to obtain the robot's initial geographical coordinates (for example, the robot is currently located at the entrance of the property service center). The positioning requirement generation form 212 generates positioning strategy parameters based on the "open outdoor area" attribute, instructing the subsequent fusion algorithm to appropriately reduce the weight of indoor sensors. Subsequently, the path planning module 3 receives the target point coordinates from the task scheduling module 1 and the current location information from the positioning module, and generates the optimal movement path from the property service center to the southeast corner of the central garden. This path needs to pass through an indoor corridor (where GPS signal is weak) and an open outdoor area. The movement module 4 then begins moving according to the path. During movement, the positioning scene cognition unit 213 of the positioning module 2 acquires point cloud data collected by the lidar 23 (for obstacle avoidance and mapping), image data collected by the vision sensor 24 (for identifying corridor environmental features), inertial navigation data collected by the inertial measurement unit 25 (for measuring acceleration and angular velocity), and odometer data collected by the encoder 26 (for recording wheel revolutions) in real time. The positioning information generation unit 214 combines this data and dynamically adjusts the fusion algorithm weights according to the positioning strategy parameters of the positioning requirement generation unit 212 and the current scene type identifier (such as "indoor corridor" or "outdoor open area") generated by the positioning scene cognition unit 213. For example, in an indoor corridor, the GPS weight is reduced and the weights of lidar and vision are increased; in an outdoor open area, the GPS weight is restored. Through adaptive fusion calculation, real-time position information with centimeter-level accuracy is generated and sent to the motion module for path correction to ensure that the robot moves accurately along the predetermined path without deviation or collision. When the robot reaches the target point (southeast corner of the central garden), the positioning information output unit 215 sends a position confirmation signal to the motion module 4, and the robot stops moving.

[0055] During the task execution, both fixed-point and route-based data collection are performed. As the robot moves to the fire hydrant, the inspection task matching unit 511 of the data collection module 5 sends a start command to the corresponding inspection data collection unit 512 (linked to the visible light camera 52 and the infrared thermal imager 53) based on the current inspection task (fire equipment inspection) issued by the task scheduling module 1. The robot initiates the route-based inspection data collection action in real time during its movement. The visible light camera 52 continuously captures images of the environment along the route, and the infrared thermal imager 53 monitors for any abnormal temperature points along the route.

[0056] After the robot stops moving, the movement module 4 sends a second trigger signal to the acquisition module 5. The acquisition module 5 then initiates the fixed-point acquisition action. The visible light camera 52 takes pictures of the fire hydrant, and the infrared thermal imager 53 detects whether the surface temperature of the fire hydrant is abnormal (to determine if there is a potential electrical fire hazard). The acquired fixed-point acquisition data (fire hydrant image and temperature data) is transmitted to the acquisition processor 51 for identification and processing. If the data is normal, the recording module 6 stores the data and generates an inspection record containing a timestamp (e.g., 2026-XX-XX 09:05), location information (southeast corner of the central garden), and the acquired content. This record is then uploaded to the backend management terminal 9 for archiving via the communication module 7.

[0057] If the data acquisition module 5 detects that the fire hydrant is blocked by debris or the pressure gauge reading is abnormal, the risk warning processing unit 513 determines it as a general abnormality and triggers the voice and video interaction module 8 to give a voice prompt: "Please note that there is debris blocking the fire hydrant." At the same time, it reports the abnormal information to the backend through the communication module 7. The report can be repeated at intervals, and the backend will notify the property staff to handle it. The abnormal information will stop being reported to the backend after receiving feedback from the backend management terminal 9 that the backend has been aware of or has handled the issue.

[0058] Assuming the robot completes its fire hydrant inspection and is en route to Building 3 (patrolling along its movement path), the fire detection unit of acquisition module 5 begins operation. Visible light camera 52 acquires real-time images of the area ahead, while multi-gas sensor array 56 continuously monitors the smoke concentration in the air.

[0059] Scenario 1: Directly identified as a fire When the robot reaches the entrance of a building, the visible light camera 52 captures flames and thick smoke coming from a window of a first-floor resident's apartment. Based on a preset fire identification model (e.g., flame features in the image + smoke concentration exceeding a threshold), the fire detection unit directly determines it to be a "fire." At this point, the task scheduling module 1 immediately interrupts its current patrol task towards Building 3. The path planning module 3, based on the fire location (the resident's coordinates) provided by the positioning module 2, replans the shortest path. The movement module 4 controls the robot to quickly move to a preset safe distance from the fire scene (e.g., 10 meters from the fire point, ensuring its own safety while facilitating monitoring). Simultaneously, the voice and video interaction module 8 activates a high-volume fire alarm (e.g., "Emergency! Fire detected, please evacuate immediately!") and sends a fire warning containing real-time video, fire location, and fire intensity to the backend management terminal 9 via the communication module 7. On-duty personnel can view the scene on the backend management terminal 9, confirm the fire, dial 119, and send a confirmation command to the robot. Upon receiving the confirmation command, the robot stops alarming but continues uploading video until firefighters arrive. In addition, the communication module 7 can also send activation commands to nearby pre-set fire-fighting linkage equipment (such as sound and light alarms and fire exit gates in the unit building) to open evacuation routes; the voice and video interaction module 8 broadcasts evacuation instructions (such as "Please evacuate in an orderly manner through the safety passage").

[0060] Scenario 2: Determined to be a suspected fire In another scenario, during a robot patrol, the multi-gas sensor array 56 detects a slight exceedance of the smoke concentration in the air, but no open flame is found in the visible light camera image 52, and visibility is normal. The fire detection unit determines this as a "suspected fire." At this point, the task scheduling module 1 suspends the current patrol task. The path planning module 3 generates a confirmed path to the smoke source point based on the direction of the highest smoke concentration. The movement module 4 controls the robot to move along the confirmed path, gradually approaching the suspected area. Upon reaching the vicinity of the suspected point, the acquisition module 5 initiates a secondary judgment, and the infrared thermal imager 53 focuses on scanning the area to detect any abnormal temperature points; the visible light camera 52 zooms in to capture high-definition images for identification. If the infrared thermal imager 53 detects an abnormally high temperature in a distribution box, and the image shows charred marks, the secondary judgment confirms a "fire" (early stage of an electrical fire), then the aforementioned fire response actions (interruption, alarm, reporting, and linkage) are immediately executed. If, upon inspection, the excessive smoke concentration is found to be due to nearby residents burning dry leaves (the burning point is visible in the image, and the temperature is normal), and a secondary assessment rules out a fire, then task scheduling module 1 resumes the original patrol task, and path planning module 3 replans the path to the original target location (electrical box in Building 3). The robot then continues the suspended electrical inspection task. Throughout the process, the robot records the suspected investigation process (including movement trajectory and secondary assessment evidence) and uploads it to the backend for traceability.

[0061] Similarly, when the robot performs a leak inspection task in the underground parking garage of Building 5 at 2:00 PM, the leak detection unit of the data acquisition module 5 is activated. The water immersion sensor monitors the ground humidity, and the sound acquisition unit 55 listens for the sound of running water.

[0062] Scenario 1: Determined to be a water leak As the robot moves through the garage aisle, the water immersion sensor 54 suddenly detects a sharp increase in ground humidity, reaching the leakage threshold. Simultaneously, the sound collector 55 captures a distinct hissing sound of running water. The leakage detection unit determines it to be a leak. The task scheduling module 1 interrupts the current leak inspection task (in fact, a leak has been detected, and the original task objective can be adjusted to address the leak). The path planning module 3 plans a path to the site based on the leak point coordinates provided by the positioning module 2. Once the robot moves near the leak point, the voice and video interaction module 8 activates a leak voice alarm (e.g., "Attention, a leak has been detected!") and sends a leak warning containing on-site video and location information to the backend via the communication module 7. After confirmation by the backend staff, the robot can be instructed to continue monitoring via system commands, while simultaneously notifying maintenance personnel to handle the situation. The robot continues to alarm and upload data until it receives confirmation from the backend and then stops alarming. Furthermore, if there are automatic valve control devices in the garage area, the communication module 7 can send a shut-off command to them, attempting to cut off the water supply.

[0063] Scenario 2: Suspected leak If the water immersion sensor 54 detects a slight increase in local humidity, but it does not reach the leakage threshold, and the sound collector 55 does not detect any obvious running water sound, the leakage detection unit determines it as "suspected leakage." This could be due to a vehicle bringing in rainwater, or a small amount of standing water on the ground. The robot pauses its current task, and the path planning module 3 generates a confirmed path to the humidity anomaly point. After moving to that point, the visible light camera 52 captures an image of the ground, and the water immersion sensor 54 collects accurate humidity data again. If the image shows only a small amount of water stains on the ground (possibly dripping from an air conditioner or brought in by a vehicle), and the humidity does not continue to rise, the second determination rules out leakage, and the robot resumes its original patrol task, continuing to patrol along the original path. If the second determination confirms a continuous leakage caused by a burst water pipe, the above-mentioned leakage response actions are executed.

[0064] For scheduled in-home gas safety inspections, the homeowner interaction unit of communication module 7 first sends an in-home inspection appointment notification to the corresponding homeowner's mobile app, informing them of the expected inspection time (e.g., 3:00 PM) and the inspection content (gas pipeline inspection). After the homeowner confirms, they send a confirmation command. Upon receiving confirmation, the robot's path planning module 3 plans a dedicated inspection route to the homeowner's door based on the real-time location from the positioning module. Upon arrival at the homeowner's door, the robot prompts the homeowner to open the door via voice. After entering the home, the data acquisition module 5 activates the multi-gas sensor array 56 to detect the natural gas concentration, while the visible light camera 52 captures the gas meter reading. The recording module 6 stores the collected data and generates an inspection record containing a timestamp, location (homeowner's address), and detection results. If a gas pipeline leak or repair is detected, the voice and video interaction module 8 issues a voice prompt to the homeowner and asks if they need to contact property management personnel. When the homeowner needs to do so, they can communicate with backend personnel via video or voice and wait for professional personnel to handle the situation. After the inspection is completed, the recording module 6 uploads the inspection record to the backend management terminal 9 via the communication module 7. The backend system verifies the data. If the data is normal, it returns an upload confirmation command. The data cleaning unit inside the recording module 6 automatically cleans up the verified records after 30 days according to the command. If the backend finds that a record has abnormal data (e.g., a gas meter reading is unclear), it returns a data verification command, and the data cleaning unit marks the record as "abnormal and temporarily stored" for manual review. If the backend determines that a record does not need to be saved, it directly sends a deletion command, and the data cleaning unit immediately deletes the record, freeing up storage space.

[0065] Through the workflow of the intelligent security patrol robot in this embodiment, it can be seen that the present invention can efficiently and accurately complete various security patrol tasks in the community, and form an effective emergency response closed loop in emergency situations. It can replace property staff to complete routine inspection tasks such as water and electricity, gas, security cameras, fire-fighting equipment, and building exterior inspections.

[0066] It should be stated that the above-described invention content and specific embodiments are intended to demonstrate the practical application of the technical solution provided by this invention and should not be construed as limiting the scope of protection of this invention. Those skilled in the art can make various modifications, equivalent substitutions, or improvements within the spirit and principles of this invention. The scope of protection of this invention is defined by the appended claims.

Claims

1. An intelligent security patrol robot with real-time positioning and risk warning, characterized in that, It includes a task scheduling module (1), a positioning module (2), a path planning module (3), a movement module (4), a data acquisition module (5), a recording module (6), a communication module (7), and a voice and video interaction module (8); among which, The task scheduling module (1) establishes a connection with the background management terminal (9) through the communication module (7) so as to receive the patrol task issued by the background management terminal (9) through the communication module (7). The task scheduling module (1) determines the current patrol task and its target location according to the patrol task. The positioning module (2) is electrically and signal connected to the task scheduling module (1) so that the task scheduling module (1) sends a first trigger signal to the positioning module (2) after determining the target location. When the positioning module (2) receives the first trigger signal, it starts positioning and obtains the current location information. The path planning module (3) is electrically and signal connected to the task scheduling module (1) and the positioning module (2) respectively, so that the path planning module (3) generates a movement path from the current position to the target position according to the target position issued by the task scheduling module (1) and the current position information provided by the positioning module (2); The mobile module (4) is electrically and signal connected to the path planning module (3) and the positioning module (2) respectively, so that the mobile module (4) moves according to the mobile path issued by the path planning module (3), performs path correction based on the real-time location information obtained by the positioning module (2) during the movement, and stops moving when it reaches the target point. The acquisition module (5) is electrically and signal connected to the mobile module (4), the task scheduling module (1), and the communication module (7) respectively, so that the acquisition module (5) adapts to the current patrol task requirements sent by the task scheduling module (1) and starts the patrol acquisition action along the way in real time during the robot's movement to obtain the patrol data along the way during the movement; after the mobile module (4) moves to the target point and stops moving, it sends a second trigger signal to the acquisition module (5). When the acquisition module (5) receives the second trigger signal, it starts the fixed-point task acquisition action to obtain the fixed-point acquisition data of the target point. The acquisition module (5) processes and identifies the patrol data along the way and the fixed-point acquisition data respectively; and when the patrol data along the way and the fixed-point acquisition data are abnormal, the acquisition module (5) reports to the background management terminal (9) through the communication module (7). The recording module (6) is electrically and signal connected to the path planning module (3) and the acquisition module (5) respectively, so that the recording module (6) stores the movement path generated by the path planning module (3), and under normal circumstances stores the fixed-point acquisition data of the acquisition module (5) at the target point, and generates an inspection record containing timestamps and location information; the recording module (6) also establishes a connection with the background management terminal (9) through the communication module (7), so that the recording module (6) uploads the generated inspection record to the background management terminal (9); The voice and video interaction module (8) is electrically and signal connected to the acquisition module (5) and the communication module (7) respectively, so that the acquisition module (5) can identify the patrol data along the way and the fixed-point acquisition data, and trigger the voice and video interaction module (8) to provide voice prompts or voice alarms when the identification is abnormal.

2. The intelligent security patrol robot with real-time positioning and risk warning as described in claim 1, characterized in that, The task scheduling module (1) includes a control circuit board (11) and a first memory (12) installed inside the robot, and a touch screen (13) installed on the robot; the control circuit board (11) is equipped with a task acquisition unit (111), a task queue management unit (112), a task execution control unit (113), and a status monitoring unit (114); wherein, The task acquisition unit (111) is electrically and signal connected to the communication module (7) and the first memory (12) respectively, so that the task acquisition unit (111) receives the patrol task issued by the background management terminal (9) through the communication module (7) and stores it in the first memory (12). The patrol task includes at least one of water patrol task, electricity patrol task, gas patrol task, security camera patrol task, fire equipment patrol task and building exterior patrol task. Each patrol task also includes the coordinate information of various target points, patrol frequency and patrol standard parameters. The task queue management unit (112) is electrically and signal connected to the task acquisition unit (111) so that the task queue management unit (112) receives the inspection tasks sent by the task acquisition unit (111), sorts them according to the priority and preset time of the inspection tasks, and generates a task queue. The task execution control unit (113) is electrically and signal connected to the task queue management unit (112), the first memory (12), the path planning module (3), and the positioning module (2), respectively, so that the task execution control unit (113) obtains the current task to be executed from the task queue management unit (112), obtains the coordinate information of the corresponding target point from the first memory (12) according to the current task to be executed and sends it to the path planning module (3), and sends a first trigger signal to the positioning module (2); The status monitoring unit (114) is electrically and signal connected to the task execution control unit (113) and the touch screen (13) respectively, so that the status monitoring unit (114) displays the patrol task on the touch screen (13) according to the current task to be executed by the task execution control unit (113), and the status monitoring unit (114) controls the task execution control unit (113) to prioritize the execution of the patrol task set by the property personnel according to the patrol task set by the property personnel obtained by the touch screen (13).

3. The intelligent security patrol robot with real-time positioning and risk warning as described in claim 1, characterized in that, The positioning module (2) includes a positioning processor (21) and a Beidou / GPS dual-mode positioning chip (22) installed inside the robot, a lidar (23) and a vision sensor (24) fixed on the top of the robot, an inertial measurement unit (25) installed on the robot chassis, and an encoder (26) installed on the robot's drive wheels; the positioning processor (21) is equipped with a task semantic parsing unit (211), a positioning requirement generation unit (212), a positioning scene cognition unit (213), a positioning information generation unit (214), and a positioning information output unit (215); among which, The task semantic parsing unit (211) is electrically and signal connected to the task scheduling module (1) and the Beidou / GPS dual-mode positioning chip (22) respectively, so that the task semantic parsing unit (211) receives the first trigger signal with the target point sent from the task scheduling module (1), extracts the regional attributes of the target point, and generates a first enable signal according to the regional attributes and sends it to the Beidou / GPS dual-mode positioning chip (22), thereby enabling the Beidou / GPS dual-mode positioning chip (22) to start and obtain the robot's initial geographical location coordinates; The positioning requirement generation unit (212) is electrically and signalally connected to the task semantic parsing unit (211) so that the positioning requirement generation unit (212) receives the regional attributes extracted by the task semantic parsing unit (211) to generate positioning strategy parameters; The positioning scene cognition unit (213) is electrically and signal connected to the positioning requirement generation unit (212), the lidar (23), the vision sensor (24), the inertial measurement unit (25), and the encoder (26) respectively, so that the positioning scene cognition unit (213) can acquire point cloud data collected by lidar (23), image data collected by vision sensor (24), inertial navigation data collected by inertial measurement unit (25), and mileage data collected by encoder (26) in real time during robot movement, and perform feature extraction and scene recognition on the point cloud data, the image data, the inertial navigation data, and the mileage data according to the positioning strategy parameters generated by the positioning requirement generation unit (212) to generate the current scene type identifier; The positioning information generation unit (214) is electrically and signal connected to the positioning demand generation unit (212), the positioning scene cognition unit (213), the Beidou / GPS dual-mode positioning chip (22), the lidar (23), the vision sensor (24), the inertial measurement unit (25), and the encoder (26), respectively, so that the positioning information generation unit (214) receives the initial geographical location coordinates from the Beidou / GPS dual-mode positioning chip (22), and combines the inertial navigation data collected by the inertial measurement unit (25) and the mileage data collected by the encoder (26) to perform trajectory estimation and generate the current position. The positioning information is sent to the path planning module (3); and during the robot's movement, the fusion algorithm weights are dynamically adjusted based on the positioning strategy parameters generated by the positioning requirement generation unit (212) and the current scene type identifier generated by the positioning scene cognition unit (213). The fusion algorithm is then combined with the real-time positioning data of the Beidou / GPS dual-mode positioning chip (22), the point cloud data collected by the lidar (23), the image data collected by the visual sensor (24), the inertial navigation data collected by the inertial measurement unit (25), and the mileage data collected by the encoder (26) to perform adaptive fusion calculation to generate real-time position information. The positioning information output unit (215) is electrically and signal connected to the positioning information generation unit (214), the path planning module (3), and the movement module (4) respectively, so that the positioning information output unit (215) receives the real-time position information generated by the positioning information generation unit (214) in real time and sends it to the path planning module (3) for path correction, and sends it to the movement module (4) to control the robot's movement. When the robot moves to the target point, the positioning information output unit (215) sends a position confirmation signal to the movement module (4) to control the robot to stop moving.

4. The intelligent security patrol robot with real-time positioning and risk warning as described in claim 1, characterized in that, The data acquisition module (5) includes a data acquisition processor (51) installed inside the robot, a visible light camera (52) and an infrared thermal imager (53) fixed on the top of the robot, a water immersion sensor (54) installed on the bottom of the robot, a sound acquisition unit (55) and a multi-gas sensor array (56) installed on the front of the robot, and an ultrasonic sensor array (57) installed on the side of the robot; the data acquisition processor (51) is equipped with a patrol task matching unit (511), multiple patrol acquisition units (512), and a risk warning processing unit (513); among which, The patrol task matching unit (511) is electrically and signal connected to the task scheduling module (1) and multiple patrol acquisition units (512) respectively, so that the patrol task matching unit (511) sends a start command to the corresponding target patrol acquisition unit according to the current patrol task sent by the task scheduling module (1); Each inspection and data collection unit (512) is electrically and signal connected to at least one of the visible light camera (52), infrared thermal imager (53), water immersion sensor (54), sound collector (55), multi-gas sensor array (56), and ultrasonic sensor array (57) according to the inspection task type, so that the robot can acquire fixed-point data through at least one of the visible light camera (52), infrared thermal imager (53), water immersion sensor (54), sound collector (55), multi-gas sensor array (56), and ultrasonic sensor array (57) according to the risk type of the target object during the movement and after reaching the target point; The risk warning processing unit (513) is electrically and signal connected to each patrol collection unit (512), the voice and video interaction module (8), and the communication module (7) respectively, so that the risk warning processing unit (513) can determine the warning level based on the data collected by each patrol collection unit (512), and then provide voice prompts or voice alarms through the voice and video interaction module (8), and report to the background management terminal (9) through the communication module (7).

5. The intelligent security patrol robot with real-time positioning and risk warning as described in claim 1, characterized in that, The acquisition module (5) also includes a fire detection unit, which is electrically and signal-connected to a visible light camera (52), an infrared thermal imager (53), and a multi-gas sensor array (56) mounted on the robot, respectively, so as to determine the fire based on the image data acquired by the visible light camera (52), the temperature data acquired by the infrared thermal imager (53), and the smoke concentration data acquired by the multi-gas sensor array (56) during the robot's movement; wherein, When the fire detection unit determines that a fire is occurring, it executes a fire response action: the task scheduling module (1) interrupts the current patrol task, the path planning module (3) replans the path according to the fire location provided by the positioning module (2), the movement module (4) controls the robot to move to a preset safe distance from the fire scene, and at the same time the voice and video interaction module (8) starts the fire voice alarm and broadcasts the location of nearby fire-fighting equipment, and sends a fire warning containing on-site video and location information to the back-end management terminal (9) through the communication module (7), until the alarm stops after receiving the confirmation instruction from the back-end management terminal (9); When the fire detection unit determines that a fire is suspected, the task scheduling module (1) suspends the current patrol task, the path planning module (3) generates a confirmed path to the smoke source point, and the movement module (4) controls the robot to move to the smoke source point. The fire detection unit makes a secondary determination using image data collected by the visible light camera (52) and temperature data collected by the infrared thermal imager (53). If the secondary determination confirms a fire, the fire response action is executed. If the secondary determination excludes a fire, the task scheduling module (1) resumes the original patrol task, and the path planning module (3) replans the path to the original target point to continue the patrol task.

6. The intelligent security patrol robot with real-time positioning and risk warning as described in claim 1, characterized in that, The acquisition module (5) also includes a leakage detection unit, which is electrically and signal-connected to the visible light camera (52), water immersion sensor (54), and sound collector (55) installed on the robot, respectively, so as to determine leakage based on the image data acquired by the visible light camera (52), the humidity data acquired by the water immersion sensor (54), and the sound data of flowing water acquired by the sound collector (55) during the robot's movement; wherein, When the leakage detection unit determines that there is a leak, it executes a leakage response action: the task scheduling module (1) interrupts the current inspection task, the path planning module (3) replans the path according to the location of the leak point provided by the positioning module (2), the movement module (4) controls the robot to move to the preset distance of the leak site, and at the same time the voice and video interaction module (8) starts the leakage voice alarm and sends a leakage warning containing on-site video and location information to the back-end management terminal (9) through the communication module (7) until the alarm stops after receiving the confirmation instruction from the back-end management terminal (9); When the leakage detection unit determines that there is a suspected leakage, the task scheduling module (1) suspends the current inspection task, the path planning module (3) generates a confirmed path to the suspected leakage point, and the movement module (4) controls the robot to move to the suspected leakage point. The leakage detection unit collects on-site image data through the visible light camera (52) and collects humidity data through the water immersion sensor (54) for secondary judgment. If the secondary judgment confirms that there is a leakage, the leakage response action is executed. If the secondary judgment eliminates the leakage, the task scheduling module (1) resumes the original inspection task, and the path planning module (3) replans the path to the original target point to continue the inspection task.

7. The intelligent security patrol robot with real-time positioning and risk warning as described in claim 1, characterized in that, The communication module (7) is equipped with a homeowner interaction unit so that the task scheduling module (1) can send the home visit appointment notice and the patrol time information to the corresponding homeowner terminal through the communication module (7). After receiving the confirmation instruction from the homeowner terminal, the path planning module (3) plans a dedicated patrol route to the home visit target location based on the real-time location of the positioning module (2).

8. The intelligent security patrol robot with real-time positioning and risk warning as described in claim 1, characterized in that, The recording module (6) includes a second memory and a data cleaning unit. The data cleaning unit is electrically connected to the communication module (7) and the second memory, respectively, so that after the recording module (6) uploads the patrol record to the background management terminal (9) through the communication module (7), it receives the upload confirmation instruction, data verification instruction or deletion instruction from the background management terminal (9). Then, the data cleaning unit performs the corresponding operation according to the background feedback instruction, including automatically cleaning up the patrol record that has passed the verification after a preset time, marking the patrol record that has failed the verification as temporarily stored, and immediately deleting the patrol record that has received the deletion instruction.

Images