Security cleaning robot coordination system and method based on state time-varying feedback

Abstract

The application discloses a security cleaning robot cooperative system and method based on state time-varying feedback. The system comprises a security robot (102) provided with a detection sensor (105), a cleaning robot (103) provided with a visual sensor (106), a suction motor (107), an electrical proportional valve (109) and a brush disc pressing cylinder (108), and a central controller (101) which writes time-varying attenuation cost for the security robot (102) to plan extraction, routes video stream takeover monitoring when the cleaning robot (103) reduces speed and stays, and reduces the suction motor (107) duty cycle and reduces the brush disc pressing cylinder (108) pressure and increases the trajectory coverage times through the electrical proportional valve (109) when generating a silent constraint domain. The application can eliminate physical skid interference, solve the monitoring blind area problem, and realize night silence and environmental cleanliness balance.

Description

Technical Field

[0001] This application relates to the field of multi-agent collaboration and automated control technology, conforms to the G05B classification, and specifically refers to a collaborative system and method for security and cleaning robots based on state time-varying feedback. Background Technology

[0002] In the nighttime management of modern hospital inpatient wards, complex situations often arise, such as patients suddenly wandering off, leaving medical waste or bodily fluid stains on the floor. Security robots and cleaning robots, as two common automated multi-agent scheduling devices, undertake the tasks of monitoring and cleaning. Currently, it is difficult to effectively integrate the operational logic of these two robots with different functions. In the closed and quiet environment at night, their independent operation presents significant physical and control conflicts.

[0003] Existing robot collaborative scheduling systems typically employ a rule-based, loosely coupled command issuance mechanism. When a security robot detects a stain, it only notifies the cleaning robot to proceed to the target point. This mechanism lacks a closed-loop response with state machine transitions, resulting in the following drawbacks: First, after the cleaning robot performs its cleaning operation, the floor remains slippery for a period. The security robot's underlying motion controller cannot detect this change in physical properties, making it prone to wheel slippage or trajectory deviation when moving at the original planned speed. Second, when dealing with stubborn stains, the cleaning robot needs to repeatedly work in the same location, keeping its chassis stationary. This time consumption reduces the overall field of view coverage of the entire ward area. Third, hospitals have strict noise threshold requirements at night, and existing cleaning robots typically operate at a fixed power level, unable to implement servo electromechanical interventions based on environmental and personnel activity levels to dynamically suppress noise. Summary of the Invention

[0004] This invention solves the technical problems of interference from ground physical conditions, expansion of monitoring blind spots, and excessive operating noise caused by the independent operation of security and cleaning robots in the prior art.

[0005] This application provides a collaborative system for security and cleaning robots based on state time-varying feedback, including a security robot equipped with detection sensors; a cleaning robot equipped with a vision sensor, a suction motor, an electro-proportional valve, and a brush pressing cylinder; and a central controller that communicates with both the security robot and the cleaning robot. The detection sensors acquire environmental state data and send it to the central controller. The central controller issues cleaning tasks to the cleaning robot based on the environmental state data, and writes a slip cost with time-varying decay attributes into the corresponding area of ​​the global two-dimensional map after the cleaning robot leaves. The security robot extracts the slip cost during path planning. After the central controller detects that the chassis linear speed of the cleaning robot is lower than a first threshold and remains below a first duration, it routes the video stream from the vision sensor to the monitoring network. When the security robot detects a trigger event, it generates a silent constraint domain, which is a dynamic physical constraint boundary formed by the security robot as the center and a first preset distance as the radius. When the cleaning robot is in the silent constraint domain, it lowers the pulse width modulation duty cycle of the suction motor, reduces the pressure value of the brush pressing cylinder through the electro-proportional valve, and increases the coverage of the working trajectory.

[0006] This application also provides a collaborative method for security and cleaning robots based on state time-varying feedback, applied to a system including a security robot, a cleaning robot, and a central controller. The security robot is equipped with a detection sensor, and the cleaning robot is equipped with a vision sensor, a suction motor, an electro-proportional valve, and a brush pressing cylinder. The method includes the following operations: the detection sensor collects environmental state data and sends it to the central controller; the central controller issues a cleaning task to the cleaning robot based on the environmental state data and, after the cleaning robot leaves, writes a slip cost with time-varying decay attributes into the corresponding area of ​​the global two-dimensional map; the security robot extracts the slip cost during path planning; after the central controller detects that the chassis linear velocity of the cleaning robot is lower than a first threshold for a first duration, it routes the video stream from the vision sensor to the monitoring network; when the security robot detects a trigger event, it generates a silent constraint domain, which is a dynamic physical constraint boundary formed by the security robot as the center and a first preset distance as the radius; when the cleaning robot is in the silent constraint domain, the cleaning robot lowers the pulse width modulation duty cycle of the suction motor, reduces the pressure value of the brush pressing cylinder through the electro-proportional valve, and increases the coverage of the working trajectory.

[0007] By introducing a time-varying attenuation cost mapping mechanism, this application can quantitatively reflect the dynamic changes in the physical friction properties of the ground and automatically avoid the risk of slippage at the navigation algorithm level, eliminating physical interference caused by independent operations. Utilizing the fixed-point dwelling characteristics of the cleaning robot for video stream routing, it can compensate for the field-of-view coverage of the security monitoring network in a specific area without increasing hardware nodes. Based on feedforward acoustics and activity event-triggered servo electromechanical intervention, controlling the power of noise sources and supplementing with trajectory coverage compensation, a high-quality operating state meeting nighttime requirements is achieved. Attached Figure Description

[0008] Figure 1 This is a logical structure block diagram of a security and cleaning robot collaborative system based on state time-varying feedback provided in the embodiments of this application.

[0009] Figure 2 This is a flowchart of a collaborative method for security and cleaning robots based on state time-varying feedback provided in an embodiment of this application.

[0010] Explanation of reference numerals in the attached figures:

[0011] 101 Central Controller, 102 Security Robot, 103 Cleaning Robot, 104 Local Area Network Node Device, 105 Detection Sensor, 106 Vision Sensor, 107 Sewage Suction Motor, 108 Brush Disc Pressing Cylinder, 109 Electro-proportional Valve Detailed Implementation

[0012] To enable those skilled in the art to better understand the technical solution of this application, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0013] Specifically, a collaborative system for security and cleaning robots based on state-time-varying feedback is provided. This system is suitable for scenarios with high requirements for quiet operation and complex environmental changes, such as hospitals at night. Figure 1 As shown, the control system mainly comprises a central controller 101, a security robot 102, and a cleaning robot 103 in terms of physical entities. The central controller 101 is deployed on an edge computing node and establishes a communication network through a local wireless network node device 104. Both the security robot 102 and the cleaning robot 103 are connected to the local wireless network node device 104 via the TCP / IP communication protocol, enabling state machine transitions and data interaction with the central controller 101.

[0014] The security robot 102 is equipped with a detection sensor 105. The detection sensor 105 is a combination of a two-dimensional LiDAR component and a depth camera component. The security robot 102 also includes an acoustic array for environmental sound field monitoring and an infrared detector for detecting human activity. The depth camera component preferably uses a depth camera, and its video output is connected to the anomaly detection interface of the central controller 101 for real-time reporting of visual anomaly frames. The security robot 102 internally includes a dynamic window local obstacle avoidance algorithm module V. 02 Its underlying controller can continuously publish abnormal event messages containing spatial coordinates and timestamps through the runtime environment.

[0015] Cleaning robot 103 is configured to perform floor cleaning and polishing tasks, and includes a mobile chassis module V. 03 Cleaning Execution Module V 04 and auxiliary vision module V 05 A vision sensor 106 mounted on the main body provides forward-looking or surround-view video acquisition capabilities. Cleaning execution module V 04 It includes a suction motor 107 driven by a brushless DC motor, and a brush disc pressing cylinder 108 whose downward pressure is controlled by an electro-proportional valve 109. The cleaning robot 103 is internally equipped with a low-level communication gateway V. 06 It is used to receive control commands and convert them into low-level servo control signals for motors and pneumatic actuators.

[0016] Based on the aforementioned hardware infrastructure, the system's operation and control logic achieves multi-agent scheduling through closed-loop state responses. The detection sensor 105 acquires environmental state data in real time and sends it to the central controller 101. When the security robot 102 detects liquid stains on the ground at coordinates (x0, y0) using its depth camera component, its underlying controller generates and publishes an abnormal event message. The central controller 101 parses this message and generates a cleaning request task frame. Based on the spatial location of each robot, it sends the cleaning task to the cleaning robot 103 closest to coordinates (x0, y0). After receiving the instruction, the cleaning robot 103 activates the mobile chassis module V. 03 Proceed to the target location to carry out the cleaning operation.

[0017] After the cleaning robot 103 completes the cleaning of the target area and leaves the area, the coefficient of friction of the ground in that area will decrease significantly due to the physical cleaning process accompanied by water spraying. To prevent the security robot 102 from slipping, a map mapping module V is installed in the central controller 101. 01 Write the slippery cost with time-varying decay properties into the corresponding area of ​​the global two-dimensional map.

[0018] Map mapping module V 01Obtain the end time t of the cleaning robot 103's water spraying and suction operations in the corresponding area. clean The task ends at time t. clean As a baseline time point, reflecting the natural evaporation pattern of moisture over time, the map mapping module V... 01 The preset initial slip cost is multiplied by a follow-up coefficient with exponential decay characteristics to generate a dynamic slip cost, which is then superimposed onto the grid nodes and their expanded regions of the global 2D map to update the map cost value. To improve the accuracy of this dynamic cost map, the map mapping module V... 01 The actual water spray volume of the cleaning robot 103 in the area surrounding the target coordinate point (x0, y0) is obtained, and the current temperature and humidity data of the environment are obtained through temperature and humidity sensors distributed in the environment.

[0019] Based on the above parameters, the map mapping module V 01 The preset initial slippage cost is determined by multiplying the actual water spray volume by a first preset coefficient. For the follow-up coefficient, the system extracts the natural constant e as the first base and the negative time variable feature as the first exponent, combining the two to construct the follow-up coefficient. The time variable feature is composed of the product of the evaporation coefficient and the duration from the end of the operation. Map mapping module V 01 Based on the current ambient temperature and humidity data, the evaporation coefficient is determined by querying the system's built-in calibration mapping table. The system then performs a calculation of the slippery surface cost, using the following formula:

[0020]

[0021] The specific calculation formula for the aforementioned operation is as follows: , where C slip (x,y,t) represents the slippery cost generated at the corresponding raster coordinates (x,y) at time t; K s This represents the preset initial slip resistance value determined by the actual water spray volume; that is... The evaporation coefficient is determined by looking up the ambient temperature and humidity data in a table; t represents the current operating time of the system; t clean This indicates the end time of the water spraying and suction operations within the corresponding area; that is... This represents an exponential function with the natural constant as the base. The slippery cost calculated using the above formula exhibits a non-linear, exponential decay over time. In a specific set of parameter application examples, K... s The value of can be set to a dimensionless value from 50 to 150 to ensure that its magnitude has a sufficiently strong penalty effect on the path planning algorithm.

[0022] When map mapping module V 01After completing the above calculations, a mapping relationship will be established between the cost of grid nodes and the motion control parameters of the security robot 102. The security robot 102 extracts this slippery cost during local path planning and inputs it into the dynamic window local obstacle avoidance algorithm module V. 02 The algorithm module's evaluation function incorporates a penalty term representing the slippery cost. When the slippery cost C in the map grid... slip When (x, y, t) exceeds a preset safe passage threshold, the control logic of the security robot 102 treats this area as a high-cost region. Based on the dynamically changing slippery cost, the security robot 102 adjusts the maximum allowable linear speed threshold boundary within this grid node in real time. The system forcibly limits the maximum linear speed of the security robot 102 in this coordinate region to below 0.2 m / s, or controls the chassis to perform a detour operation. By introducing this time-varying cost mechanism, the system cuts off the yaw loss path caused by drastic changes in friction at the physical state information level, achieving safe collaborative control.

[0023] During multi-agent scheduling, considering that the cleaning robot 103 needs to perform low-speed or stationary reciprocating movements within a 1-meter radius when handling stains, this mechanical dwell state would limit local security resources. Therefore, the system implemented a dynamic field-of-view takeover strategy. The central controller 101 continuously monitors the chassis linear velocity v of the cleaning robot 103. clean When the central controller 101 detects the chassis linear velocity v of the cleaning robot 103... clean If the video falls below a first threshold and persists for a first duration, the internal video scheduling bus V is triggered. 07 .

[0024] Video scheduling bus V 07 Upon activation, the cleaning robot 103 is determined to have entered a deep-stay cleaning state. To conserve local storage and processing resources, the video scheduling bus V... 07 The default local storage process of the cleaning robot 103 is forcibly interrupted, and a real-time streaming protocol connection channel is established between the vision sensor 106 and the security video processing server, directly routing the video stream of the cleaning robot 103 to the monitoring network. The central controller 101 sets the current spatial coordinates of the cleaning robot 103 as a static monitoring station.

[0025] After the static monitoring station is established, the central controller 101 extracts the spatial field of view coverage of the visual sensor 106 at the static monitoring station. The central controller 101 reads the current global patrol task queue of the security robot 102 and traverses each patrol target point in the queue. The central controller 101 calculates the spatial intersection area between the coordinate region of each patrol target point and the spatial field of view coverage of the visual sensor 106. When the spatial intersection area is greater than the preset overlap area boundary value, it is determined that the video monitoring of the area has been taken over by the cleaning robot 103. The central controller 101 removes the matching patrol target point from the global patrol task queue to unload the monitoring and dwelling task of the security robot 102 in the area. The task node information after removal is sent to the security robot 102, controlling it to reconstruct the global traversal topology path of the remaining spatial nodes and prioritize patrolling other edge areas of the floor. Through this resource reuse method, the system effectively improves the overall security monitoring coverage of the nighttime environment without increasing hardware investment.

[0026] In terms of noise suppression during nighttime operations, the security robot 102 acts as a sentinel detector, monitoring the environment in real time. When the security robot 102 detects a trigger event, it sends information to the central controller 101 and generates a silence constraint domain. Trigger events include transient sound pressure levels in the environment collected by the acoustic array exceeding a preset silence standard value, infrared detectors and depth camera components capturing heat sources from human activity, and detecting that a ward door is open. After receiving the signal, the system generates a dynamic physical constraint boundary in the map coordinate system, with the current coordinates of the security robot 102 as the center and a first preset distance as the radius. This physical space is the silence constraint domain.

[0027] When the cleaning robot 103 enters or is within this quiet constraint zone, its internal underlying communication gateway V 06 The acoustic constraint identifier was received from the system. The underlying communication gateway V... 06 The system forcibly truncates and overrides the default operating power command it is currently executing. Closed-loop adjustments are performed through underlying servo electromechanical intervention. The underlying communication gateway V... 06 The difference between the obtained transient sound pressure level and the preset silence standard value is calculated, and the corresponding attenuation gain is extracted based on this difference. (Underlying communication gateway V) 06 The pulse width modulation control signal controlling the brushless DC motor is adjusted, and the pulse width modulation duty cycle of the suction motor 107 is reduced to a range that matches the attenuation gain. The system forcibly reduces its duty cycle from 80% of the rated operating state to 45%, reducing turbulence noise caused by high-speed gas flow.

[0028] While reducing airflow noise, the system performs drag reduction and voltage regulation using mechanical structures. The underlying communication gateway V... 06The output regulating voltage is sent to the electro-proportional valve 109, which actively releases compressed air, reducing the pressure threshold of the brush disc pressing cylinder 108 to a preset low-pressure working position. The system lowers the brush disc pressing pressure from the rated P... norm Reduced to 0.6×P norm This operation effectively reduces the rigid friction noise when the brush bristles come into contact with hard ground materials.

[0029] Since the reduction in electromechanical power and pressure inevitably leads to a decrease in the cleaning effect per cycle, in order to ensure the final cleaning quality, the underlying communication gateway V... 06 Simultaneously, the system intervenes in its kinematic controller, altering the trajectory planning algorithm of the cleaning robot 103. The cleaning robot 103 reduces the lateral overlap between adjacent parallel cleaning paths in its original trajectory to a preset first gap and increases the number of coverage operations. The system changes the original single-pass bow-shaped coverage to a triple-overlapping bow-shaped coverage. By superimposing the number of cleaning operations over time to reconstruct the trajectory, the cleanliness requirements of the floor are met while strictly limiting instantaneous physical noise. Through the combination of the aforementioned feedforward detection and underlying servo electromechanical intervention, the problem of mutual constraints among multiple agents in harsh environments is resolved.

[0030] Figure 2 A collaborative method for security and cleaning robots based on state-time-varying feedback is also provided. This method relies on the aforementioned system architecture comprising a security robot 102, a cleaning robot 103, and a central controller 101, achieving global control through clearly defined time and logical steps. For example... Figure 2 As shown, this control method mainly includes the following operation nodes:

[0031] Step S201: Acquire environmental status data and send it to the central controller 101. This step relies on the real-time spatial scanning of the security robot 102's lidar component and depth camera during its movement to identify abnormal objects within the area.

[0032] Step S202: A cleaning task is issued and a slippery cost with time-varying decay attributes is written. The cleaning task is issued to the cleaning robot 103 based on environmental state data. After the cleaning robot 103 completes the cleaning action within the specified range and leaves the coordinate point, a slippery cost with time-varying decay attributes is written to the corresponding area on the global two-dimensional map. This further includes obtaining the end time t of the cleaning robot 103's water spraying and suction operations within the corresponding area. clean ; with the end time t cleanAs a baseline time point, the preset initial slip cost is multiplied by a follow-up coefficient with exponential decay characteristics to generate the dynamic slip cost at the current running time. The calculated slip cost is used as a penalty weight and superimposed on the corresponding grid node of the global two-dimensional map and its surrounding expansion area to update the overall map cost.

[0033] Step S203: Extract the slippery cost from the path planning of the security robot 102. When the local path generator of the security robot 102 detects an increase in the passage cost of the grid ahead, it will automatically issue a deceleration command or guide the chassis to detour to avoid tire slippage.

[0034] Step S204: The video stream is routed to the monitoring network, and the central controller 101 monitors the chassis linear velocity v of the cleaning robot 103. clean If the stain falls below a first threshold and persists for a first duration, it is determined to be in the stubborn stain treatment stage. Its internal storage is interrupted, and a connection link is established with the video stream from the vision sensor 106 carried by the cleaning robot 103, routing it to the security network to fill in the gaps in the monitoring footage.

[0035] Step S205: When a triggering event is detected, a silent constraint domain is generated. When the security robot 102 detects a triggering event during its forward patrol, a silent constraint domain centered on the robot and with a radius of a specific distance is generated on the global map.

[0036] Step S206: Reduce the motor voltage power and implement trajectory coverage intervention. When the cleaning robot 103 in operation is determined to be within the quiet constraint range, pressure reduction and noise reduction measures are triggered. The system forcibly reduces the pulse width modulation duty cycle of the cleaning robot 103's suction motor 107 and reduces the working pressure value of its brush disc pressing cylinder 108 through the electro-proportional valve 109. The sweeping spacing is also increased at the control panel level to increase the coverage of the working trajectory.

[0037] Building upon the above embodiments, to ensure the system's continuous and stable operation around the clock and provide physical-level fault tolerance, the robot's underlying power supply and communication links can be further improved. Both the security robot 102 and the cleaning robot 103 are equipped with a charge / discharge management system and a high-rate power battery pack to drive their mobile chassis and sensors. Between the central controller 101 and each robot, in addition to the main data packet transmission link, the local wireless network node device 104 also supports automatic switching to a backup channel when the main frequency band encounters strong interference, ensuring low-latency transmission of critical safety braking commands and abnormal status frames.

[0038] In summary, the method and system provided in this application, through a rigorous mathematical function of slippery conditions, eliminate interference from the ground physical state generated during operations on inspection and navigation. It maintains the continuity of the security network through a field-of-view takeover mechanism that transforms the disadvantage of dwell time into an advantage for video blind spot compensation. Based on acoustic and active heat source signal-triggered feedforward noise suppression electromechanical coordinated control, high-frequency noise is cut off at the source of motor control frequency and aerodynamic pressure, balancing acoustic performance with ground cleanliness, and achieving adaptive adjustment.

[0039] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A collaborative system for security and cleaning robots based on state time-varying feedback, characterized in that, include: Security robots are equipped with detection sensors; The cleaning robot is equipped with a vision sensor, a suction motor, an electric proportional valve, and a brush plate pressing cylinder. The central controller is communicatively connected to the security robot and the cleaning robot. The detection sensor acquires environmental status data and sends it to the central controller; The central controller issues cleaning tasks to the cleaning robot based on the environmental status data, and writes a slippery cost with time-varying decay attributes into the corresponding area of ​​the global two-dimensional map after the cleaning robot leaves. The security robot extracts the slippery cost during path planning; After the central controller detects that the chassis linear velocity of the cleaning robot is lower than a first threshold and remains so for a first duration, it routes the video stream from the vision sensor to the monitoring network. When the security robot detects a trigger event, it generates a silent constraint domain, which is a dynamic physical constraint boundary formed by the security robot as the center and a first preset distance as the radius. When the cleaning robot is in the silent constraint domain, the cleaning robot reduces the pulse width modulation duty cycle of the suction motor, reduces the pressure value of the brush cylinder through the electro-proportional valve, and increases the number of times the working trajectory is covered.

2. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 1, characterized in that, The central controller is equipped with a map mapping module; The map mapping module obtains the end time of the cleaning robot, and uses the end time as the reference time point to multiply the preset initial slip cost with the follow-up coefficient with exponential decay characteristics to generate the slip cost. The map mapping module overlays the slippery cost onto the grid nodes of the global two-dimensional map to update the map cost.

3. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 2, characterized in that, The map mapping module obtains the actual water spray volume of the cleaning robot in the corresponding area, the current ambient temperature data, and the current ambient humidity data. The map mapping module determines the preset initial slippery cost based on the product of the actual water spray volume and the first preset coefficient, extracts the natural constant as the first base, extracts the negative time variable feature as the first exponent, and combines the first base and the first exponent to construct the follow-up coefficient. The time variable feature is composed of the product of the evaporation coefficient determined by the current ambient temperature data and the current ambient humidity data, and the duration from the end of the operation. The map mapping module establishes a mapping relationship between the grid nodes and the motion control parameters of the security robot, and adjusts the maximum linear speed threshold boundary that the security robot is allowed to travel within the grid node in real time according to the dynamically changing slippery cost.

4. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 1, characterized in that, The central controller is equipped with a video scheduling bus. When it is determined that the chassis linear speed of the cleaning robot is lower than the first threshold and the first duration is met, the video scheduling bus cuts off the default local storage process of the cleaning robot and establishes a real-time streaming protocol connection channel from the vision sensor to the security video processing server. The central controller sets the current coordinates of the cleaning robot as a static monitoring station.

5. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 4, characterized in that, The central controller extracts the spatial field of view coverage of the static monitoring station and extracts the current global patrol task queue of the security robot. The central controller traverses each patrol target point in the global patrol task queue and calculates the spatial intersection area between the coordinate region of each patrol target point and the spatial field of view coverage. When the spatial intersection area is greater than the preset overlapping area boundary value, the central controller removes the matching patrol target point from the global patrol task queue and sends the removed task node information to the security robot, controlling the security robot to reconstruct the global traversal topology path of the remaining spatial nodes.

6. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 1, characterized in that, The security robot includes an acoustic array and an infrared detector; The triggering events include the transient sound pressure level collected by the acoustic array exceeding the preset silence standard value, or the infrared detector capturing a heat source of human activity.

7. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 6, characterized in that, The cleaning robot includes an internal underlying communication gateway; When the cleaning robot crosses the dynamic physical constraint boundary, the underlying communication gateway forcibly intercepts and overrides the default operating power command; The underlying communication gateway calculates the difference between the transient sound pressure level and the preset silent standard value and extracts the attenuation gain, and adjusts the pulse width modulation duty cycle of the sewage suction motor to the operating range that matches the attenuation gain; The underlying communication gateway reduces the pressure threshold of the brush plate pressing cylinder to a preset low-pressure working position through the electrical proportional valve, and controls the cleaning robot to reduce the lateral overlap distance between two adjacent parallel cleaning paths in the working trajectory to a preset first distance, and adds extra cleaning operations to reconstruct the working trajectory.

8. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 1, characterized in that, The detection sensor is composed of a two-dimensional lidar component and a depth camera component, and the video output end of the depth camera component is connected to the anomaly detection interface of the central controller.

9. The collaborative system for security and cleaning robots based on state time-varying feedback as described in claim 1, characterized in that, The security robot includes a dynamic window local obstacle avoidance algorithm module, and the security robot introduces a penalty term representing the slippery cost into the evaluation function of the dynamic window local obstacle avoidance algorithm module.

10. A collaborative method for security and cleaning robots based on state time-varying feedback, applied to a system comprising a security robot, a cleaning robot, and a central controller, wherein the security robot is equipped with a detection sensor, and the cleaning robot is equipped with a vision sensor, a suction motor, an electro-proportional valve, and a brush disc pressing cylinder, characterized in that, Includes the following operations: The detection sensor collects environmental status data and sends it to the central controller; The central controller issues cleaning tasks to the cleaning robot based on the environmental status data, and writes a slippery cost with time-varying decay attributes into the corresponding area of ​​the global two-dimensional map after the cleaning robot leaves. The security robot extracts the slippery cost during path planning; After the central controller detects that the chassis linear velocity of the cleaning robot is lower than a first threshold and remains so for a first duration, it routes the video stream from the vision sensor to the monitoring network. When the security robot detects a trigger event, it generates a silent constraint domain, which is a dynamic physical constraint boundary formed by the security robot as the center and a first preset distance as the radius. When the cleaning robot is in the silent constraint domain, the cleaning robot reduces the pulse width modulation duty cycle of the suction motor, reduces the pressure value of the brush cylinder through the electro-proportional valve, and increases the number of times the working trajectory is covered.

Images