Parking lot emergency fire protection robot cluster cooperative path dynamic programming method

Abstract

The application provides a parking lot emergency fire protection robot cluster cooperative path dynamic programming method, comprising: obtaining parking lot fire environment data and robot cluster state data, constructing a fire heat environment grid map, determining a target propulsion corridor, cooling demand and a robot role; generating a target time-space injection sequence according to a candidate injection position and an injection response, controlling the injection robot to form a mobile low-temperature window moving along the target propulsion corridor; constructing a time-space passage map based on the mobile low-temperature window trajectory, planning a target travel path and a target speed sequence of the working robot; determining a dynamic meeting point in the mobile low-temperature window according to a cumulative heat absorption amount of the working robot to predict a heat bearing limit time, planning a meeting path of the working robot and the support robot; exchanging a detachable heat absorption protection component at the dynamic meeting point, re-distributing the robot role, and updating the injection sequence according to task execution feedback data.

Description

Technical Field

[0001] This invention relates to the field of robot swarm collaborative control technology, and in particular to a dynamic path planning method for a swarm of emergency fire protection robots in a parking lot. Background Technology

[0002] Parking lots are characterized by dense vehicle traffic, narrow passageways, enclosed spaces, and difficulty in dispersing smoke. After a vehicle fire occurs, the location of the fire source, temperature distribution, heat radiation intensity, smoke movement, and traffic conditions will continuously change. Directly entering high-temperature, high-smoke areas poses a significant safety risk to firefighters, thus requiring route planning.

[0003] Existing path planning methods for firefighting robots mostly rely on parking lot maps, fire source locations, and obstacle locations to generate travel paths, primarily addressing geometric obstacle avoidance. They rarely consider ambient temperature, thermal radiation intensity, spray cooling range, and temperature recovery time simultaneously. In actual operations, the low-temperature zone created by spraying varies depending on the spray location, direction, and timing. If the spraying task and the robot's travel path are planned separately, problems may arise such as adjacent cooling areas not being continuously connected, or the cooling effect diminishing by the time the robot arrives.

[0004] Furthermore, existing robot swarm collaboration methods typically allocate robots according to detection, spraying, and transportation tasks, but lack unified coordination regarding the timing of actions of multiple spraying robots, the passage sequence of operational robots, and the arrival sequence of support robots. When the fire source expands, the direction of smoke changes, the spraying effect decreases, or the actual arrival time of robots deviates, the predetermined path is difficult to adjust in time, and operational robots may leave the safe cooling zone.

[0005] Therefore, this invention proposes a dynamic path planning method for a cluster of emergency fire protection robots in parking lots. The information disclosed in the background section is only for enhancing understanding of the background of this disclosure and may therefore contain prior art information that is not common knowledge to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a dynamic path planning method for collaborative paths of parking lot emergency fire protection robot clusters, thereby solving the technical problems mentioned in the background section.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A dynamic path planning method for collaborative paths of emergency fire protection robot clusters in parking lots includes the following steps: S1. Acquire fire scene environmental data and robot cluster status data, construct a fire scene thermal environment grid map, determine the target advance corridor and cooling requirements, and assign fire protection robots to operation robots, spraying robots and support robots; S2. Based on the cooling requirements, determine the candidate injection positions and candidate injection response sets, use the ant colony algorithm to generate the target spatiotemporal injection sequence, control the injection robot to form a moving cryogenic window that moves along the target propulsion corridor, and generate the moving cryogenic window trajectory. S3. Construct a spatiotemporal travel map of the moving cryogenic window based on the moving cryogenic window trajectory, solve the target travel path and target speed sequence of the operation robot, control the operation robot to travel within the moving cryogenic window, and correct the target spatiotemporal spray sequence according to the actual window position. S4. Based on the real-time heat absorption rate of the working robot, determine the real-time cumulative heat absorption, the expected heat tolerance limit time, and the safe heat margin time series. Determine the dynamic meeting point within the moving low temperature window trajectory, and generate the working robot meeting path, the support robot support path, and the meeting area spray control sequence. S5. Control the operation robot and support robot to exchange detachable heat-absorbing protection components at the dynamic rendezvous point. Based on the robot cluster status data after the exchange, reassign robot roles, update the target travel path, evacuation path and target spatiotemporal spray sequence, and perform path replanning based on task execution feedback data.

[0008] S1 specifically includes: acquiring visible light images, infrared thermal images, and depth images to determine the fire source boundary, obstacle boundary, smoke state, and the temperature, thermal radiation intensity, and geometric passage status of the environmental grid, constructing a fire scene thermal environment grid map; fusing visual measurement results, wheel speed encoder data, and inertial measurement data to determine the robot's position, heading, and speed, and acquiring the fire-fighting medium capacity, power supply, and initial thermal state of replaceable heat-absorbing protective components to form robot cluster status data; and determining the target advancement corridor, cooling requirements, and robot role allocation results based on the fire scene thermal environment grid map and robot cluster status data, including operation robots, spraying robots, and support robots.

[0009] S2 specifically includes: setting candidate spraying locations along the target advancement corridor; determining the effective cooling coverage area, temperature recovery time, and steam accumulation level based on trial spraying data or spraying calibration data, forming a candidate spraying response set; establishing a spatiotemporal node connection for spraying based on the overlap relationship of adjacent effective cooling coverage areas, formation sequence, spraying robot reachability, and remaining fire-fighting medium capacity, and using an ant colony algorithm to solve the target spatiotemporal spraying sequence; controlling the spraying robot to execute spraying according to the target spatiotemporal spraying sequence; identifying moving low-temperature windows based on environmental grid temperature, thermal radiation intensity, and geometric passage status, determining the window position, moving direction, moving speed, and effective duration, and forming the moving low-temperature window trajectory.

[0010] S3 specifically includes: based on the fire thermal environment grid map, robot cluster status data, target advancement corridor, and moving cryogenic window trajectory, setting passable spatiotemporal nodes and their connections to construct a spatiotemporal passage map of the moving cryogenic window; using the current node of the operating robot as the starting node and the advancement node within the moving cryogenic window as the target node, using the ant colony algorithm combined with path length, predicted cumulative heat exposure, window deviation, speed change, and heading change to determine the target travel path and target speed sequence; adjusting the speed and heading based on the longitudinal and lateral deviations between the operating robot and the moving cryogenic window, correcting the target spatiotemporal injection sequence based on the window position, and determining the heat absorption rate to form follow-up control feedback data.

[0011] S4 specifically includes: determining the real-time cumulative heat absorption and heat tolerance limit based on the initial cumulative heat absorption and real-time heat absorption rate; predicting the cumulative heat absorption based on the target spatiotemporal injection sequence and the moving cryogenic window trajectory to form the expected heat tolerance limit time and safe heat margin time series; screening candidate rendezvous nodes that meet the constraints of docking, window duration, rendezvous synchronization, and thermal state within the moving cryogenic window trajectory; determining the dynamic rendezvous point and solving the support path for the support robot; updating the target node with the dynamic rendezvous point; generating the rendezvous path and unified rendezvous time for the operating robot; adjusting the injection spatiotemporal nodes according to the rendezvous area and component exchange duration to form the rendezvous area injection control sequence and dynamic rendezvous planning results.

[0012] S5 specifically includes: controlling the operation robot and support robot to enter the dynamic rendezvous point, completing alignment based on relative pose and relative speed, exchanging detachable heat-absorbing protection components, and generating component exchange completion information based on component identification, arrival signal, and locking status; reassigning robot roles based on component exchange completion information and robot cluster status data after exchange, generating target travel path, evacuation path, and target spatiotemporal spray sequence to form an updated cluster collaborative path; controlling the robot cluster to execute the cluster collaborative path, generating task execution feedback data based on actual cooling coverage, temperature recovery time, fire source boundary, and accumulated heat absorption, and replanning based on the task execution feedback data.

[0013] The beneficial effects of this invention are as follows: This invention constructs a fire scene thermal environment grid map that includes temperature, thermal radiation intensity, smoke state, and geometric passage status by integrating visible light, infrared thermal imaging, and depth vision measurement, thereby improving the accuracy of fire scene perception and passage risk assessment in complex parking lots. By coordinating the spraying positions, spraying times, and spraying durations of multiple spraying robots, a low-temperature window that moves continuously along the target advancement corridor is formed, avoiding spatial or temporal interruptions between adjacent cooling zones.

[0014] This invention integrates the trajectory of a moving cryogenic window with the robot's motion constraints, heat exposure risks, and collision avoidance requirements into a joint model. This enables the robot to continuously follow a safe cooling zone, reducing the risk of leaving the cryogenic window or becoming trapped in high temperatures. By calculating real-time cumulative heat absorption, predicting cumulative heat absorption, and estimating the heat tolerance limit, the timing of component replacement can be determined in advance based on the thermal environment of the subsequent path, avoiding premature evacuation or overheating caused by fixed operation durations.

[0015] This invention determines a dynamic rendezvous point within a moving low-temperature window and collaboratively plans the rendezvous path of the operating robot, the support robot's support path, and the spray control sequence for the rendezvous area. This enables the replacement of heat-absorbing protective components to be exchanged under safe cooling conditions, improving continuous operation capability. By updating the spray response, robot roles, and cluster collaborative paths using the actual cooling coverage area, temperature recovery time, robot arrival status, and component heat absorption status, the robot cluster can adapt to fire source expansion, changes in spray effect, and altered access conditions, improving the safety and adaptability of emergency firefighting operations. Attached Figure Description

[0016] Figure 1 This is an overall flowchart of the dynamic path planning method for the cluster of emergency fire protection robots in parking lots according to the present invention. Figure 2 This is a structural block diagram of the parking lot emergency fire protection robot cluster collaborative control system of the present invention; Figure 3 This is a schematic diagram illustrating the formation of a mobile cryogenic window and the collaborative operation of a robot cluster in a parking lot fire environment, as described in this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1: As Figures 1 to 3 As shown in the figure, this embodiment provides a method for dynamic path planning of a cluster of emergency fire protection robots in a parking lot, including the following steps: S1. Acquire fire scene environmental data and robot cluster status data, construct a fire scene thermal environment grid map, determine the target advance corridor and cooling requirements, and assign fire protection robots to operation robots, spraying robots and support robots; S2. Based on the cooling requirements, determine the candidate injection positions and candidate injection response sets, use the ant colony algorithm to generate the target spatiotemporal injection sequence, control the injection robot to form a moving cryogenic window that moves along the target propulsion corridor, and generate the moving cryogenic window trajectory. S3. Construct a spatiotemporal travel map of the moving cryogenic window based on the moving cryogenic window trajectory, solve the target travel path and target speed sequence of the operation robot, control the operation robot to travel within the moving cryogenic window, and correct the target spatiotemporal spray sequence according to the actual window position. S4. Based on the real-time heat absorption rate of the working robot, determine the real-time cumulative heat absorption, the expected heat tolerance limit time, and the safe heat margin time series. Determine the dynamic meeting point within the moving low temperature window trajectory, and generate the working robot meeting path, the support robot support path, and the meeting area spray control sequence. S5. Control the operation robot and support robot to exchange detachable heat-absorbing protection components at the dynamic rendezvous point. Based on the robot cluster status data after the exchange, reassign robot roles, update the target travel path, evacuation path and target spatiotemporal spray sequence, and perform path replanning based on task execution feedback data.

[0019] S1 specifically includes the following sub-steps: S110. Construct a grid map of the fire scene's thermal environment. The projection of the parking lot entrance center onto the ground is used as the origin of a unified parking lot coordinate system. The directions pointing from the entrance into the parking lot, along the parking spaces, and perpendicular to the ground are used as the positive directions of the coordinate axes, with the coordinate unit being meters (m).

[0020] Visible light cameras, infrared thermal imaging devices, and depth cameras mounted on the fire protection robot are used to simultaneously perform visual measurements of driveways, parking spaces, pillars, fire doors, ramps, fire sources, and smoke areas within the parking lot. Each camera uses a unified trigger signal output from the robot controller for image acquisition; for cameras that do not support unified triggering, images with a timestamp difference of no more than 100ms are grouped into the same measurement frame.

[0021] Based on the camera device intrinsic parameters, rotation matrix, and translation vector obtained using the calibration board before the task execution, each measurement point is transformed to a unified parking lot coordinate system:

[0022] in, Let be the coordinates of the spatial point in the unified parking lot coordinate system; Let be the coordinates of the spatial point in the coordinate system of the m-th camera device; and These are the rotation matrix and translation vector from the coordinate system of the m-th camera device to the coordinate system of the unified parking lot, respectively.

[0023] The continuous pixel region in the infrared thermal image whose temperature is higher than the fire source temperature threshold is defined as the fire source boundary. The fire source temperature threshold is the larger value between the temperature value after the current ambient temperature of the parking lot is increased by 60°C and the 95th percentile value of the temperature value of all effective pixels in the current infrared thermal image.

[0024] Obstacle boundaries are determined based on depth abrupt change points in the depth image and the outlines of vehicles, pillars, and fire doors in the visible light image; smoke regions are determined based on contrast-decrease areas in the visible light image and non-uniform temperature rise areas in the infrared thermal image; and the direction of smoke movement and smoke velocity are determined based on the optical flow displacement and depth values ​​of smoke feature points in adjacent measurement frames.

[0025] The parking lot space was divided into environmental grids with a side length of 0.25m, and the coordinates, average temperature, maximum temperature, thermal radiation intensity, smoke velocity, ground height, obstacle occupancy status, geometric traffic status, and data update time of each environmental grid were recorded.

[0026] The thermal radiation intensity of each environmental grid is determined according to the following formula:

[0027] in, The thermal radiation intensity of the g-th environmental grid is expressed in W / m². 2 ; The emissivity of the heated object's surface is read from the material parameter table; The viewing angle coefficient is determined based on the distance, angle, and occlusion status between the environmental grid and the heated object. It is the Stefan constant; The surface temperature is after emissivity correction, in °C. The current ambient temperature of the parking lot is in °C.

[0028] When there are no fixed obstacles in the environmental grid, the ground height difference does not exceed the obstacle-crossing height of the fire protection robot, and the slope does not exceed the maximum allowable slope, the environmental grid is marked as a geometrically passable grid; otherwise, it is marked as a geometrically impassable grid, thus forming a fire scene thermal environment grid map.

[0029] For example, if the side temperature of a vehicle is 180℃, the current ambient temperature of the parking lot is 35℃, the emissivity is 0.90, and the viewing angle coefficient is 0.60, the corresponding thermal radiation intensity is calculated according to the above formula, and the calculation result is written into the corresponding environmental grid to avoid directly substituting the surface temperature for the thermal radiation intensity.

[0030] S120. Generate robot cluster state data. Identify the unique coded visual positioning markers on the top and sides of each fire protection robot using visual measurement to obtain the robot's identifier, outline center, and vehicle orientation. Use the travel distance output by the wheel speed encoder and the heading angle change and acceleration output by the inertial measurement unit as state prediction data, and the absolute position and heading obtained from visual measurement as state correction data. Use extended Kalman filtering to determine the current position, heading, and travel speed of each fire protection robot.

[0031] When the visible area of ​​the visual positioning mark is less than 40% of the complete area, or the reprojection error is greater than 3 pixels, the current visual measurement result is not adopted. Instead, the wheel speed encoder and inertial measurement device are used for state prediction, and the cumulative positioning error is corrected after the visual measurement is restored.

[0032] Read the outputs of the liquid level sensor in the storage tank, the pressure sensor in the spray pipeline, the flow sensor, the angle encoder of the spray gimbal, and the battery management system to obtain the remaining fire-fighting medium capacity, spray pressure, spray flow rate, spray direction adjustment range, and remaining power; read the vehicle width, minimum turning radius, maximum permissible speed, maximum permissible acceleration, maximum permissible slope, and maximum braking distance from the robot equipment parameter table.

[0033] Based on the component identification of the replaceable heat absorption protection component, the rated heat absorption capacity, mass of each heat absorption layer, specific heat capacity, latent heat of phase change, and phase change temperature range are read from the component parameter table. The initial cumulative heat absorption and initial remaining heat absorption capacity are then determined based on the output of the component temperature sensor and the component's historical heat absorption records.

[0034]

[0035] in, The initial cumulative heat absorption of the replaceable heat-absorbing protective components carried by the i-th fire protection robot; This represents the initial remaining heat absorption capacity; This refers to the number of heat-absorbing layers; , , , and These represent the mass, specific heat capacity, current temperature, latent heat of phase change, and mass percentage of the r-th layer of heat-absorbing material, respectively; The baseline temperature before the mission begins; This is the rated heat absorption capacity.

[0036] The above data are mapped according to robot identifiers to form robot cluster status data. For example, when the sensible heat absorption layer has a mass of 12 kg and a specific heat capacity of 900 J / (kg·℃), the phase change heat absorption layer has a mass of 4 kg and a latent heat of phase change of 180 kJ / kg, the reference temperature is 25℃, the current average temperature is 65℃, and the phase change ratio is 25%, the initial cumulative heat absorption is 612 kJ; when the rated heat absorption capacity is 1500 kJ, the initial remaining heat absorption capacity is 888 kJ.

[0037] As a specific implementation method, the sensible heat absorption layer can be made of high-temperature resistant aluminum silicate ceramic fiber or silica aerogel felt material; the phase change heat absorption layer can be made of high-temperature resistant phase change material with a melting point between 100°C and 200°C, such as erythritol, high-density polyethylene or specific anhydrous crystalline salt, to ensure that a stable phase change can occur and latent heat can be absorbed under the high temperature environment of the fire.

[0038] S130. Determine the robot role assignment results and target advancement corridor. Delete geometrically impassable grids from the fire thermal environment grid map, and use the sum of the circumscribed circle radius of the fire protection robot vehicle and the safe distance as the expansion radius to perform obstacle expansion processing on the vehicle, pillars, walls, and fire door boundaries; in the remaining geometrically passable grids, use an 8-neighborhood connectivity method to search from the current position of the fire protection robot cluster to the target grid outside the fire source boundary.

[0039] Candidate propulsion corridors are defined as connected grid bands that can continuously connect the starting grid and the target grid, and whose minimum width is not less than the sum of the robot body width and the safety distance on both sides. The corridor length, minimum width, maximum temperature, maximum thermal radiation intensity, flue gas reverse velocity, and length of continuous high-temperature sections are calculated for each candidate propulsion corridor. Candidate propulsion corridors whose width does not meet the passage requirements or whose length of continuous high-temperature sections exceeds the allowable high-temperature exposure length are excluded. Then, the target propulsion corridors are determined in the order of maximum thermal radiation intensity from low to high, flue gas reverse velocity from low to high, and corridor length from short to long.

[0040] The permissible high-temperature exposure length is the product of the permissible high-temperature exposure time of the fire protection robot and the predetermined initial travel speed. The permissible high-temperature exposure time, permissible ambient temperature, and permissible thermal radiation intensity are determined by the robot's temperature resistance test and thermal radiation tolerance test. Environmental grids within the target propulsion corridor whose temperature or thermal radiation intensity exceeds the corresponding permissible value are designated as cooling grids. The cooling requirements for the target propulsion corridor are formed based on the total area of ​​the cooling grids, the excessive temperature difference, the excessive thermal radiation intensity, and the required duration of maintaining a low-temperature state.

[0041] Robot roles were assigned in the following order: first, the operational robot; then, the spraying robot; and finally, the support robot. The operational robot was selected from those with initial remaining heat absorption capacity sufficient to meet the predicted heat absorption of the initial propulsion section, located close to the entrance of the target propulsion corridor, and with a minimum turning radius meeting the corridor requirements.

[0042] The minimum number of spraying robots is determined based on the length of the continuous cooling section and the effective cooling coverage length of a single spraying robot. If the minimum number is less than 2, then 2 is selected. From the remaining fire protection robots, robots with spray pressure, spray flow rate, remaining fire medium capacity, and arrival capability that meet the cooling requirements of the target advance corridor are selected as spraying robots. From the remaining fire protection robots, robots carrying low heat load replaceable heat-absorbing protection components and requiring a short time to reach the entrance of the target advance corridor are selected as support robots.

[0043] A low-heat-load replaceable heat-absorbing protection component refers to a replaceable heat-absorbing protection component whose initial cumulative heat absorption is no more than 30% of the rated heat absorption capacity and whose initial remaining heat absorption capacity is greater than the heat absorption capacity required for one expected rendezvous.

[0044] The final result is the robot role allocation and target advancement corridor. The fire thermal environment grid map, robot cluster status data, robot role allocation result, target advancement corridor and target advancement corridor cooling requirements are output to S210.

[0045] S2 specifically includes the following sub-steps: S210. Establish a candidate spray response set. Using the centerline of the target advancement corridor as a reference, set the longitudinal sampling interval to 40% to 70% of the effective spray distance of a single spray robot, and generate candidate spray positions on both sides of each sampling position. A candidate spray position refers to the position of the spray robot's center in a unified parking lot coordinate system when the spray robot can safely reach and stably stop, and the spray gimbal can cover at least one grid to be cooled within the allowable adjustment range.

[0046] Delete locations where the stopping area is too narrow, the ground slope exceeds the maximum allowable slope, the stopping area blocks the robot's advance, there is no connecting path between the robot and adjacent candidate spraying locations, or the spraying gimbal cannot cover the grid to be cooled.

[0047] For each retained candidate spray location, the spraying robot is preferentially controlled to conduct trial spraying for 1 to 3 seconds at 10% to 20% of the rated spray flow rate. The temperature drop and recovery processes of each environmental grid are measured using an infrared thermal imaging device. When trial spraying is not possible, the temperature drop data corresponding to the same spray pressure, spray distance, spray angle, initial temperature, and smoke velocity are read from the spray test parameter table generated before the mission, and correction is performed using linear interpolation of adjacent calibration data. The spray test parameter table is obtained by conducting tests in a standard fire test field with different pressures, angles, and initial environmental temperatures, and recording and fitting the temperature change curves of the corresponding grids over time using an infrared thermal imager array.

[0048] The effective cooling coverage area is defined as a continuous set of environmental grids that simultaneously meet the requirements of ambient temperature not exceeding the allowable ambient temperature, thermal radiation intensity not exceeding the allowable thermal radiation intensity, and maintaining geometric passability for a continuous duration. The continuous duration is not less than the time required for the robot to pass through this effective cooling coverage area, plus a 2-second control margin. The time elapsed after spraying stops, when the ambient temperature or thermal radiation intensity first exceeds the corresponding allowable value again, is defined as the temperature recovery time, and is taken as the minimum temperature recovery time of all environmental grids within the effective cooling coverage area.

[0049] The steam accumulation level is determined based on the contrast reduction rate of the visible light image and the output of the humidity sensor: Level 1 is when the contrast reduction rate is no more than 20% and the relative humidity is below 85%; Level 2 is when the contrast reduction rate is greater than 20% but no more than 40%, or the relative humidity is between 85% and 95%; and Level 3 is when the contrast reduction rate is greater than 40%, or the relative humidity is above 95%.

[0050] Candidate injection locations with a steam accumulation level of 3 are not included in the subsequent ant colony algorithm solution. A candidate injection response set is formed by associating the candidate injection location, injection robot identifier, injection direction, injection pressure, injection flow rate, injection duration, effective cooling coverage area, temperature recovery time, expected fire-fighting medium consumption, and steam accumulation level.

[0051] For example, if the effective cooling coverage length of candidate injection position P1 is 4.2m and the temperature recovery time is 7.5s, and the effective cooling coverage length of candidate injection position P2 is 3.8m and the temperature recovery time is 10.2s, then the effective cooling coverage area corresponding to P2 must be formed before the effective cooling coverage area corresponding to P1 fails.

[0052] S220. The target spatiotemporal spray sequence is generated using the ant colony algorithm. A spray spatiotemporal node is defined as the state combination of a given spraying robot performing spraying according to given spraying parameters at a given candidate spraying position and within a given time period.

[0053] A directed connection is established between two spraying spatiotemporal nodes only when there is a continuous common grid band in the effective cooling coverage area corresponding to adjacent spraying spatiotemporal nodes, the width of the common grid band is not less than the safe passage width of the operating robot, the time when the subsequent spraying spatiotemporal node begins to form an effective cooling coverage area is not later than the time when the effective cooling coverage area corresponding to the previous spraying spatiotemporal node fails, the spraying robot can arrive on time, the predicted area occupied by the vehicle body will not conflict, and the cumulative fire-fighting medium consumption does not exceed the corresponding remaining fire-fighting medium capacity.

[0054] The time difference between the failure time of the effective cooling coverage area corresponding to the previous injection spatiotemporal node and the time when the effective cooling coverage area begins to form at the next injection spatiotemporal node is defined as the phase safety margin.

[0055] In the h-th iteration, the probability that the a-th artificial ant moves from the jet spatiotemporal node u to the jet spatiotemporal node v is:

[0056] in, The node transition probability; The connection pheromone concentration between node u and node v at the h-th iteration; As a heuristic value; The pheromone influence coefficient; This is the heuristic influence coefficient; Let z be the set of possible nodes for the a-th artificial ant that satisfy all constraints at node u. Any candidate node in; The connection pheromone concentration between node u and candidate node z at the h-th iteration; Let be the heuristic value corresponding to node u to candidate node z.

[0057] The heuristic value is determined by a weighted average of the effective cooling coverage overlap length, phase safety margin, expected fire-fighting medium consumption, steam accumulation level, and spray robot movement distance after normalization. The initial pheromone for all feasible connections is set to 1, the number of artificial ants is set to 20 to 60, the maximum number of iterations is set to 50 to 150, and the pheromone volatility coefficient is set to 0.1 to 0.4.

[0058] After each round of searching, pheromones are added to the spatiotemporal spraying sequences that can continuously cover the target advancement corridor and do not involve interruptions in the moving cryogenic window, level 3 steam accumulation, or robot path conflicts. Among all feasible spatiotemporal spraying sequences, the target spatiotemporal spraying sequence is determined in the following order: longer continuous duration of the moving cryogenic window, lower consumption of fire-fighting medium, and shorter total distance traveled by the spraying robot. If no feasible spatiotemporal spraying sequence is obtained, the number of spraying robots is increased, the single planning segment is shortened, or the next candidate advancement corridor is selected in sequence; if no feasible result is still found, the operating robot is prohibited from entering the target advancement corridor.

[0059] S230. Form and correct the moving cryogenic window trajectory. Control each spraying robot to move to the candidate spraying position corresponding to the target spatiotemporal spraying sequence, and execute spraying according to the target spatiotemporal spraying sequence. Continuously update the temperature, thermal radiation intensity and geometric passage status of each environmental grid through visual measurement.

[0060] The largest connected region consisting of adjacent environmental grids within the same measurement time, whose temperature and thermal radiation intensity do not exceed the corresponding allowable values, whose minimum lateral width is not less than the safe passage width of the operation robot, and whose longitudinal length is not less than the length of the operation robot body, is defined as the moving cryogenic window. Connected regions meeting these conditions are identified using an 8-neighborhood connectivity method, and the connected region with the largest overlap area with the moving cryogenic window at the previous measurement time is determined as the current moving cryogenic window.

[0061] The centerline of the moving cryogenic window is formed by sequentially connecting the midpoints of the left and right boundaries obtained on each transverse section of the moving cryogenic window. The midpoint of the overlapping section between the target advancement corridor centerline and the moving cryogenic window is defined as the window position. The window movement direction and speed are determined based on the window position at three consecutive measurement times. The shortest remaining temperature recovery time in the environmental grid corresponding to the narrowest section of the moving cryogenic window is determined as the effective window duration. The moving cryogenic window trajectory is formed by combining the window position, window boundary, window centerline, window movement direction, window movement speed, and effective window duration in chronological order.

[0062] When the measured window position lags behind the predicted window position by more than 0.5m along the target advancement direction, the spray start time of the next spraying robot is advanced by 0.2s to 1s, and the spray duration of the current spraying robot is extended by 5% to 20%; when the measured window position is ahead, the spray start time of the next spraying robot is delayed; when the minimum lateral width of the moving cryogenic window is less than the safe passage width of the operation robot, the operation robot's entry is suspended, and the spray duration of the adjacent spraying robot is extended.

[0063] If a complete infrared thermal image or depth image cannot be obtained for two consecutive measurement cycles, the system automatically switches to the multi-sensor fusion blind zone estimation mode. It uses millimeter-wave radar data located around the vehicle to identify nearby obstacles and combines this data with data from the wheel speed encoder and inertial measurement unit to estimate the robot's position. Simultaneously, it does not update the moving cryogenic window trajectory and controls the robot to stop. If the visual measurement failure hold time (determined by the sum of the robot's safe braking time and control communication delay) is exceeded and the system still has not recovered, S210 is re-executed. Finally, the candidate injection response set, the target spatiotemporal injection sequence, and the corrected moving cryogenic window trajectory are output to S310.

[0064] S3 specifically includes the following sub-steps: S310. Construct a spatiotemporal access map for the moving cryogenic window. Read the continuously updated fire thermal environment grid map, robot cluster status data, target advancement corridor, target spatiotemporal spray sequence, and corrected moving cryogenic window trajectory. Define the positions that are within the moving cryogenic window at the corresponding time and can accommodate the operating robot body as passable spatiotemporal nodes.

[0065] Each passable spatiotemporal node records its node number, corresponding time, unified parking lot coordinates, candidate node speed, candidate node heading, ambient temperature, thermal radiation intensity, lateral distance from the centerline of the moving cryogenic window, minimum distance from the window boundary, permissible speed range, permissible heading range, and remaining effective window duration. The time discrete interval is the smaller of 1 / 10 of the shortest effective duration of the moving cryogenic window and 5 times the robot control cycle.

[0066] For two traversable spatiotemporal nodes at adjacent time points, a directed connection is established only if kinematic constraints, window constraints, and collision avoidance constraints are satisfied:

[0067] in, This represents the spatial distance between node c and node d. It represents the difference between the corresponding times of the two nodes; and These are the candidate velocities for nodes c and d, respectively. and These are the maximum permissible speed and the maximum permissible acceleration, respectively. and These are the candidate headings for nodes c and d, respectively. This represents the maximum permissible rate of change of heading angle.

[0068] The window constraint requires that the connecting line segments between nodes always remain within the moving low-temperature window during the corresponding time period, and that the remaining effective duration of the window after the working robot reaches the node is greater than the sum of the braking time and the control margin. The collision avoidance constraint requires that the predicted vehicle body occupancy areas of the working robot and other fire protection robots do not overlap, and that a safe collision avoidance distance is maintained. The predicted position of the spraying robot is determined by the target spatiotemporal spray sequence, while the predicted positions of other fire protection robots are determined by their current position, current speed, and current control command.

[0069] For example, when the time discrete interval is 0.5s, the distance between nodes is 0.55m, and the maximum allowable speed is 1.4m / s, although the distance constraint is met, if the connecting line segment between nodes passes through an area where the width of the moving cryogenic window is insufficient, the connection is still determined to be an infeasible connection. This forms the spatiotemporal travel map of the moving cryogenic window.

[0070] S320. The ant colony algorithm is used to solve for the target travel path and target velocity sequence. The traversable spatiotemporal node that is closest to the current actual position of the robot, whose corresponding time is no earlier than the current control time, and which satisfies the current heading constraints is determined as the starting node. The traversable spatiotemporal node that is closest to the fire source boundary along the center line of the target advancement corridor within the current moving cryogenic window trajectory coverage area, and whose remaining effective window duration is greater than the robot's safe braking time, is determined as the target node. The ant colony algorithm transfer mechanism in S220 is used to search for candidate paths from the starting node to the target node in the spatiotemporal travel map of the moving cryogenic window.

[0071] The overall evaluation value of the candidate paths is:

[0072] in, Candidate paths; This is the comprehensive evaluation value of the candidate paths; This represents the normalized path length. To normalize the predicted cumulative heat exposure; This is the normalized window center deviation; To normalize the window boundary risk, its value is negatively correlated with the minimum distance from the path node to the window boundary; This represents the normalized rate of change. This represents the normalized change in heading. , , , , , These are the corresponding weights, and the sum of the weights is 1.

[0073] In the embodiments, values ​​of 0.13, 0.30, 0.20, 0.15, 0.12, and 0.10 can be used respectively. Candidate paths with robot path conflicts, interrupted movement cryogenic windows, or inability to safely brake before the window expires are determined as infeasible paths. The feasible path with the lowest comprehensive evaluation value is determined as the target travel path. The initial target speed is obtained by dividing the distance between adjacent path nodes by the corresponding time interval, and the maximum permissible speed, maximum permissible acceleration, and end-effector safety braking are checked sequentially to form a target speed sequence.

[0074] For example, candidate path A is 8.0m long, but many nodes are less than 0.15m from the window boundary; candidate path B is 8.8m long, but all nodes are more than 0.40m from the window boundary and the predicted cumulative heat exposure is low. When the comprehensive evaluation value of path B is smaller, path B is selected.

[0075] If no feasible path is found, the robot is controlled to remain within the current cryogenic window, and S230 is invoked to adjust the injection phase and reconstruct the spatiotemporal passage map of the cryogenic window; if no feasible path is still found, the robot is controlled to retreat in the reverse direction along the nodes that have already been passed.

[0076] S330, Execute the moving cryogenic window follow-up control and generate follow-up control feedback data. The robot controller executes follow-up control according to a control cycle of 0.05s to 0.20s. The projected distance between the current position of the robot and the current window position in the direction of the target advancement corridor centerline is defined as the longitudinal deviation, and the shortest distance from the center of the robot body to the centerline of the moving cryogenic window is defined as the lateral deviation.

[0077] When the longitudinal lag exceeds the allowable longitudinal deviation, the robot's speed is increased, provided it does not exceed the maximum allowable acceleration. When the robot is less than the safe braking distance from the front boundary of a window, the speed is reduced. When the lateral deviation exceeds the allowable lateral deviation, the heading is corrected according to the direction of the deviation. The robot stops advancing when it is less than the emergency boundary distance from any window boundary. The allowable longitudinal deviation, allowable lateral deviation, and emergency boundary distance are determined based on visual measurement error, robot body width, and maximum braking distance.

[0078] If the difference between the actual temperature and the allowable ambient temperature of the operating robot's environment grid is no greater than the temperature warning margin, and the operating robot is lagging longitudinally, extend the spraying duration of the current spraying robot. If there is only a longitudinal lag, and the minimum lateral width of the moving low-temperature window in front is no less than the safe passage width of the operating robot in two consecutive control cycles, and the change in window movement speed does not exceed the allowable change in window speed, only increase the operating robot's travel speed. When the operating robot approaches the front boundary of the window, start the forward spraying robot in advance. When the steam accumulation level reaches level 3, reduce the spray flow and stop the operating robot. The temperature warning margin and the allowable change in window speed are determined by simulated fire follow-up tests.

[0079] The simulated fire-following test specifically involves: constructing a simulated fire source and propulsion channel in the test field; controlling the robot to move within a preset cooling window at different speeds and with varying lateral and longitudinal deviations; recording the extreme values ​​of the ambient temperature difference and window speed change when the robot's outer surface temperature reaches the safety threshold; and calibrating the aforementioned warning margin and allowable variation based on this data. The real-time heat absorption rate of the operating robot is calculated using the following formula:

[0080] in, Let be the real-time heat absorption rate of the i-th working robot at time t, in W; The effective heating area is read from the robot's structural parameter table; The radiation absorption correction factor is determined by the thermal radiation calibration test of the replaceable heat-absorbing protection component; The actual thermal radiation intensity of the grid surrounding the robot. The convective heat transfer coefficient was obtained from a table of thermal test parameters generated at different flue gas velocities. The actual ambient temperature at the location of the robot. The outer surface temperature of the replaceable heat-absorbing protective component is shown. The thermal radiation calibration test and thermal test parameter tables were obtained by applying standard radiant heat sources of different intensities and high-temperature flue gas with different flow rates in a closed test chamber, and by comparing the measured heat absorption rate with the theoretically calculated value using a heat flow meter.

[0081] The measurement time, actual position of the robot, actual heading, actual speed, longitudinal deviation, lateral deviation, actual temperature, actual thermal radiation intensity, component outer surface temperature, real-time heat absorption rate, actual trajectory, corrected target spatiotemporal injection sequence, and corrected moving cryogenic window trajectory are combined according to a unified timestamp to form follow-up control feedback data, which is then output to S410.

[0082] S4 specifically includes the following sub-steps: S410: Predict the expected heat tolerance limit moment and form a time series of safe heat margin. Read the initial cumulative heat absorption, rated heat absorption capacity and component parameters formed in S120, read the candidate injection response set formed in S230, and read the follow-up control feedback data, the corrected target spatiotemporal injection sequence and the corrected moving cryogenic window trajectory formed in S330.

[0083] Starting from the initial accumulated heat absorption, discrete integration is performed according to the real-time heat absorption rate at the effective measurement time:

[0084] in, The real-time cumulative heat absorbed by the i-th robot at the k-th measurement time; This represents the initial cumulative heat absorption; For the first Real-time heat absorption rate at each measurement moment; This is the time interval between adjacent valid measurement moments.

[0085] If data is missing for one control cycle, the larger of the two most recent effective real-time heat absorption rates will be used for conservative updates; if the continuous missing time exceeds 1 second, the robot will be stopped. The minimum value among the rated heat absorption capacity minus 10% safety reserve and the cumulative heat absorption corresponding to the battery, controller, vision measurement device, and drive motor reaching the allowable temperature is defined as the heat tolerance limit; the difference between the rated heat absorption capacity and the real-time cumulative heat absorption is defined as the real-time remaining heat absorption capacity.

[0086] Based on the corrected target spatiotemporal injection sequence, candidate injection response set, and fire thermal environment grid map, the predicted temperature and predicted thermal radiation intensity for each subsequent time period are obtained. The predicted heat absorption rate is then calculated using the S330 heat absorption rate relationship, thereby determining the predicted cumulative heat absorption for future times.

[0087] in, For the future The predicted cumulative heat absorption; This represents the current real-time cumulative heat absorption. For the future The predicted heat absorption rate.

[0088] The moment when the predicted cumulative heat absorption first reaches the heat tolerance limit is defined as the predicted heat tolerance limit moment, and a safety margin is calculated for each future predicted moment:

[0089] in, For work robots in the future Safety heat margin; This represents the heat resistance limit.

[0090] This generates a time series of safety margins, which the S420 uses to read the safety margins at different times for different candidate rendezvous nodes.

[0091] S420. Determine the dynamic rendezvous point and generate a support path for the support robot. Within the corrected moving cryogenic window trajectory, a passable spatiotemporal node that can simultaneously accommodate the work robot and the support robot, and allows for the exchange of replaceable heat-absorbing protective components, is defined as a candidate rendezvous node.

[0092] The effective width of the candidate rendezvous node shall not be less than the sum of the widths of the two robot bodies, the component exchange interval, and the safety distances on both sides. The ground slope shall not exceed the allowable slope determined by the calibration test of the component exchange mechanism. The remaining effective window duration shall not be less than the sum of the arrival time difference of the two robots, the component exchange duration, and the departure time. The component exchange duration shall be the maximum completion time in the alignment, unlocking, component transfer, locking, and status verification tests, plus a 2-second control margin.

[0093] The critical narrow passage is defined as a continuous corridor section in which only one fire protection robot is allowed to pass through and there is no space on either side for another fire protection robot to stop or avoid; the junction of the area occupied by the spray robot body and the space sweep area formed by the spray gimbal within the allowable turning angle is defined as the spray turning area.

[0094] Based on the fire scene thermal environment grid map and the trajectory of the moving cryogenic window, a spatiotemporal path map for the support robot is established, and the ant colony algorithm of S320 is used to solve the support robot's support path. The support path must not occupy critical narrow passages, must not enter the spray turning area, and when the support robot reaches the candidate rendezvous node, its predicted cumulative heat absorption must not exceed the thermal tolerance limit, and its replaceable heat absorption protection components should still meet the low heat load judgment conditions.

[0095] The difference between the estimated arrival time of the task robot and the estimated arrival time of the support robot is defined as the rendezvous and synchronization error.

[0096] in, Let be the rendezvous and synchronization error of the j-th candidate rendezvous node; and These are the estimated times when the working robot and the support robot arrive at the j-th candidate rendezvous node, respectively.

[0097] Candidate rendezvous nodes that are deleted if the rendezvous synchronization error exceeds the rendezvous synchronization threshold, if the safe heat margin corresponding to the expected completion time of component exchange of the working robot is less than the heat margin required for component exchange and emergency evacuation, or if the cumulative heat absorption predicted by the support robot exceeds the heat tolerance limit.

[0098] Among the remaining candidate rendezvous points, the nodes with the largest safety margin for the work robot, the earliest arrival time for the two robots, and the shortest support path are selected as dynamic rendezvous points in sequence. If no feasible candidate rendezvous point is found, the subsequent advance distance of the work robot is shortened, the spray duration in the current area is extended, or the support robot is replaced in sequence; if no feasible point is still found, the work robot is controlled to withdraw.

[0099] S430: Jointly update the rendezvous path, target velocity sequence, and rendezvous area injection control sequence. Replace the original target node in S320 with the passable spatiotemporal node corresponding to the dynamic rendezvous point; when the dynamic rendezvous point is on the original target's travel path, intercept the original target's travel path and re-execute reachability, speed, and braking checks; when the dynamic rendezvous point deviates from the original target's travel path, resolve the robot's rendezvous path in the moving cryogenic window spatiotemporal travel map.

[0100] The rendezvous time is determined based on the shortest safe arrival time of the two robots, and the rendezvous synchronization error is eliminated by adjusting the target speed within the permissible waiting section. The permissible waiting section refers to the path section where the temperature, heat radiation intensity, and remaining effective window duration all meet the robot's dwell requirements, and do not obstruct the passage of other fire protection robots.

[0101] Based on the effective area of ​​the dynamic rendezvous point, the duration of component exchange, and the area occupied by the two robot vehicles, select a spatiotemporal node from the target spatiotemporal spray sequence that can cover the rendezvous area, so that the rendezvous area forms an effective cooling coverage area before the first robot arrives, and continues until the component exchange is completed and the two robots leave; the spray direction must not be directly opposite the component exchange interface, and the steam accumulation level must not reach level 3.

[0102] When the original target spatiotemporal injection sequence cannot cover the rendezvous area, a local spatiotemporal injection node set is established using candidate injection positions within a 5m radius before and after the dynamic rendezvous point. The ant colony algorithm from S220 is then used to adjust the injection start time, injection duration, and injection robot stopping position, without modifying injection spatiotemporal nodes that have already been completed or cannot be safely interrupted. If the injection control adjustment causes changes in the predicted temperature, predicted thermal radiation intensity, or predicted heat absorption rate, the process returns to S410 to recalculate the safety margin time series.

[0103] The final dynamic rendezvous planning result includes the dynamic rendezvous point, unified rendezvous time, rendezvous path and target speed sequence of the operating robot, support path and target speed sequence of the support robot, expected component exchange duration, expected safe heat margin of the operating robot, expected remaining heat absorption capacity of the support robot, spray control sequence of the rendezvous area, and rendezvous failure evacuation path, and is output to S510.

[0104] S5 specifically includes the following sub-steps: S510: Control the operation robot and support robot to complete the exchange of replaceable heat-absorbing protection components. Read the dynamic rendezvous planning results and read the component exchange interface position, visual positioning mark parameters, and component exchange allowable error from the robot equipment parameter table. Control the operation robot and support robot to enter the dynamic rendezvous point according to the corresponding path and target speed sequence. Use at least three non-collinear visual positioning marks set on the two robot bodies to perform visual measurements to obtain the relative position, relative heading, and relative speed of the two robots.

[0105] The relative position error and the relative heading error are determined according to the following formula:

[0106]

[0107] in, This refers to the relative position error; This refers to the relative heading error; , , These are the location and heading of the support robot; , , These are the position and heading of the robot. , , These are the target's relative position and target's relative heading when the component exchange interface reaches the alignment state.

[0108] Component exchange is initiated only when the relative position error is no greater than 20 mm, the relative heading error is no greater than 2°, the relative speed is no greater than 0.02 m / s, and the dynamic rendezvous point is still within the moving cryogenic window. The component exchange mechanism includes a component carrier slot, a locking mechanism, an exchange actuator, a position sensor, a locking position sensor, and a component identification reading device.

[0109] During the exchange, the support robot's exchange actuator first supports the high-heat-load replaceable heat-absorbing protective component carried by the working robot. Then, the locking mechanism of this component is released, and it is transferred to an empty component-bearing slot on the support robot. Subsequently, the low-heat-load replaceable heat-absorbing protective component is inserted into the working robot's component-bearing slot and locked. As a specific hardware implementation, the locking mechanism employs a spring-loaded latch structure linked by an electromagnetic push rod; the exchange actuator includes a telescopic push rod and a lateral translation guide rail. When the two robots are aligned, the support robot's telescopic push rod extends, using its end electromagnet to attract the high-heat-load component on the working robot. The working robot's electromagnetic push rod retracts, unlocking the spring-loaded latch. The telescopic push rod then retracts, dragging the component along the lateral translation guide rail into the support robot's empty component-bearing slot. Next, the reverse action is performed, pushing the low-heat-load component into the working robot's component-bearing slot. The working robot's electromagnetic push rod is de-energized, resetting the spring-loaded latch and completing the mechanical locking.

[0110] High heat load replaceable heat absorption protection components refer to high heat load ratios that reach the rated heat absorption capacity in real time, or replaceable heat absorption protection components with a safety heat margin lower than the heat margin required for component exchange and emergency evacuation; the high heat load ratio is determined by component thermal cycling test.

[0111] The component exchange is considered complete and a component exchange completion message is generated only when the component identifier of the low-heat-load replaceable heat-absorbing protection component matches the predetermined component identifier, the positioning sensor outputs a valid signal, the locking position sensor outputs a locking completion signal, the installation interface temperature is lower than the allowable interface temperature, and the robot controller can read the component temperature and component parameters. If two consecutive alignment or locking failures occur, the most recent stable locking state is restored and the rendezvous failure evacuation path is executed.

[0112] S520: Based on the component exchange completion information, reassign robot roles and update the cluster collaborative path. Read the current component identifier, real-time cumulative heat absorption, real-time remaining heat absorption capacity, current position, heading, travel speed, remaining fire-fighting medium capacity, spray pressure, spray flow rate, remaining power, and estimated arrival time of each fire protection robot to form the robot cluster status data after the exchange.

[0113] The next advancement segment is defined as the target advancement corridor segment extending from the current dynamic rendezvous point along the updated moving cryogenic window trajectory to the next dynamic rendezvous point or the current planning time domain endpoint; the next planning period is defined as the time interval between the current control command issuance time and the next periodic replanning time.

[0114] The new operational robot should carry a low-heat-load replaceable heat-absorbing protection component, and the real-time remaining heat-absorbing capacity should be greater than the sum of the predicted heat absorption and safety reserve for the next propulsion section; the spraying robot should meet the spraying pressure, spraying flow rate and fire-fighting medium capacity requirements for the next planning cycle; the robot carrying a high-heat-load replaceable heat-absorbing protection component is identified as an evacuation support robot.

[0115] When multiple fire protection robots meet the same role conditions, calculate the role assignment evaluation value:

[0116] in, To assume the role of the i-th fire protection robot Evaluation value; Indicates the robot character type; To satisfy the normalization capability; To normalize the remaining power; To normalize arrival timeliness, the shorter the arrival time, the larger this value; This is a normalized path safety margin; , , , These are the corresponding weights, and the sum of the weights is 1.

[0117] Among the fire protection robots that meet the role qualification requirements, the robot with the highest role assignment evaluation value is selected to assume the corresponding role; if the evaluation values ​​are the same, the robots with shorter arrival times and larger real-time remaining heat absorption capacity are selected in order. Using the dynamic rendezvous point as the new starting position, S310 and S320 are executed again to generate a new target travel path and target speed sequence for the working robots.

[0118] Based on the fire thermal environment grid map, an evacuation time and space access map is established. Areas where the ambient temperature and thermal radiation intensity are both below the safe transfer threshold of components, can accommodate the evacuation support robot to dock, and allow the operation of external cooling equipment are defined as low temperature receiving areas. Evacuation paths that avoid new operation robot target travel paths, key narrow passages, and spray turning areas are generated.

[0119] When the role of the spraying robot changes, the remaining fire-fighting medium capacity is insufficient, or the occupancy status of the candidate spraying positions changes, S220 is re-executed based on the updated robot role allocation results to obtain the updated target spatiotemporal spraying sequence. The updated robot role allocation results, the new target travel path, the target speed sequence, the evacuation path, and the updated target spatiotemporal spraying sequence are collectively determined as the updated cluster cooperative path.

[0120] S530: Update the cluster collaboration path based on task execution feedback data. Control the new operation robot to follow the moving cryogenic window towards the fire source area along the new target travel path, control the evacuation support robot to transport the high heat load replaceable heat-absorbing protection components to the cryogenic receiving area along the evacuation path, and control the spraying robot to perform spraying according to the updated target spatiotemporal spraying sequence.

[0121] By using visual measurement, infrared thermal imaging devices, depth cameras, component temperature sensors, jet pipeline pressure sensors, and robot controller timestamps, the actual cooling coverage area, actual temperature recovery time, fire source boundary, moving low-temperature window position, robot actual arrival time, and actual cumulative heat absorption are obtained, forming task execution feedback data.

[0122] For the same candidate injection location, injection pressure, injection direction, and similar environmental conditions, the response parameters in the candidate injection response set are updated recursively:

[0123] in, The response parameters are updated for the p-th candidate injection position; These are the response parameters before the update. These are the actual response parameters measured in this study; The update coefficient is determined based on the confidence level of visual measurements; response parameters include effective cooling coverage length, effective cooling coverage width, or temperature recovery time.

[0124] The actual formation of continuous moving cryogenic window spatiotemporal node connections and passable spatiotemporal node connections increases pheromones, while connections that experience interruptions in the moving cryogenic window, temperature exceeding limits, robot delays, or path conflicts decrease pheromones.

[0125] If the fire source boundary moves more than one environmental grid side length relative to the previous planning result, or if the actual temperature recovery time is more than 20% lower than the predicted value, return to S210 to replan the injection spatiotemporal sequence; if the minimum lateral width of the moving low-temperature window is less than the safe passage width of the operating robot, or if a new geometrically impassable grid appears, return to S310 to replan the target travel path; if the safe heat margin is lower than the heat margin required for component exchange and emergency evacuation, return to S410 to redetermine the expected heat tolerance limit time and dynamic rendezvous point; if the robot's role capabilities or the status of the replaceable heat-absorbing protection components change, return to S520 to reassign the robot role.

[0126] When the fire source boundary no longer expands within a preset duration, and the maximum temperature and maximum heat radiation intensity of the target fire source area are both below the task completion threshold, the current firefighting task ends; when the parking lot structural safety monitoring system issues a collapse alarm, or visual measurement and depth measurement confirm that the top plate, columns or ground have displaced beyond the structural safety threshold, or when the operating robot is about to reach its thermal tolerance limit and there is no feasible dynamic rendezvous point, the fire protection robot cluster is controlled to carry out emergency evacuation according to its respective evacuation path.

[0127] All the above formulas are performed using dimensionless numerical calculations; the relevant formulas are based on empirical models that approximate the real situation, obtained through extensive data collection and software simulation fitting. The preset parameters and thresholds involved in the formulas can be conventionally set and adjusted by those skilled in the art according to the physical constraints of the actual application scenario.

[0128] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0129] 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. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0130] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic path planning method for collaborative paths of emergency fire protection robot clusters in parking lots, characterized in that, Includes the following steps: S1. Acquire fire scene environmental data and robot cluster status data, construct a fire scene thermal environment grid map, determine the target advance corridor and cooling requirements, and assign fire protection robots to operation robots, spraying robots and support robots; S2. Based on the cooling requirements, determine the candidate injection positions and candidate injection response sets, use the ant colony algorithm to generate the target spatiotemporal injection sequence, control the injection robot to form a moving cryogenic window that moves along the target propulsion corridor, and generate the moving cryogenic window trajectory. S3. Construct a spatiotemporal travel map of the moving cryogenic window based on the moving cryogenic window trajectory, solve the target travel path and target speed sequence of the operation robot, control the operation robot to travel within the moving cryogenic window, and correct the target spatiotemporal spray sequence according to the actual window position. S4. Based on the real-time heat absorption rate of the working robot, determine the real-time cumulative heat absorption, the expected heat tolerance limit time, and the safe heat margin time series. Determine the dynamic meeting point within the moving low temperature window trajectory, and generate the working robot meeting path, the support robot support path, and the meeting area spray control sequence.

2. The dynamic path planning method for collaborative paths of parking lot emergency fire protection robot clusters according to claim 1, characterized in that, Also includes: S5. Control the operation robot and support robot to exchange detachable heat-absorbing protection components at the dynamic rendezvous point. Based on the robot cluster status data after the exchange, reassign robot roles, update the target travel path, evacuation path and target spatiotemporal spray sequence, and perform path replanning based on task execution feedback data.

3. The parking lot emergency fire protection robot cluster collaborative path dynamic planning method according to claim 1, characterized in that, S1 specifically includes: Collect visible light images, infrared thermal images, and depth images to determine the fire source boundary, obstacle boundary, smoke state, and the temperature, thermal radiation intensity, and geometric access status of the environmental grid, and construct a fire thermal environment grid map. By integrating visual measurement results, wheel speed encoder data, and inertial measurement data, the robot's position, heading, and speed are determined, and the initial thermal state of the fire-fighting medium capacity, power supply, and replaceable heat-absorbing protective components is obtained to form robot cluster state data.

4. The parking lot emergency fire protection robot cluster collaborative path dynamic planning method according to claim 3, characterized in that, Also includes: Based on the fire thermal environment grid map and robot cluster status data, the target advance corridor, cooling requirements, and robot role allocation results were determined. The robot roles include operation robots, spraying robots, and support robots.

5. The dynamic path planning method for collaborative paths of parking lot emergency fire protection robot clusters according to claim 1, characterized in that, S2 specifically includes: Candidate injection locations are set up along the target advance corridor. Based on the test injection data or injection calibration data, the effective cooling coverage area, temperature recovery time, and steam accumulation level are determined to form a set of candidate injection responses. Based on the overlapping relationship of adjacent effective cooling coverage areas, formation sequence, reachability of spraying robots, and remaining fire-fighting medium capacity, a spatiotemporal node connection for spraying is established, and the target spatiotemporal spraying sequence is solved using the ant colony algorithm. The control robot executes spraying according to the target spatiotemporal spraying sequence. Based on the environmental grid temperature, thermal radiation intensity, and geometric passage status, it identifies the moving cryogenic window, determines the window position, moving direction, moving speed, and effective duration, and forms the moving cryogenic window trajectory.

6. The dynamic path planning method for collaborative paths of parking lot emergency fire protection robot clusters according to claim 1, characterized in that, S3 specifically includes: Based on the fire thermal environment grid map, robot cluster status data, target advancement corridor and moving cryogenic window trajectory, set passable spatiotemporal nodes and their connections, and construct a spatiotemporal passability map of the moving cryogenic window. Using the current node of the robot as the starting node and the propulsion node within the low-temperature window as the target node, the ant colony algorithm is used in combination with path length, predicted cumulative heat exposure, window deviation, speed change and heading change to determine the target travel path and target speed sequence.

7. The parking lot emergency fire protection robot cluster collaborative path dynamic planning method according to claim 6, characterized in that, Also includes: Adjust the speed and heading based on the longitudinal and lateral deviations between the robot and the moving cryogenic window, correct the target spatiotemporal injection sequence based on the window position, determine the heat absorption rate, and generate follow-up control feedback data.

8. The dynamic path planning method for collaborative paths of parking lot emergency fire protection robot clusters according to claim 1, characterized in that, S4 specifically includes: The real-time cumulative heat absorption and heat tolerance limit are determined based on the initial cumulative heat absorption and the real-time heat absorption rate. The cumulative heat absorption is predicted based on the target spatiotemporal injection sequence and the trajectory of the moving low-temperature window, forming the time series of the expected heat tolerance limit moment and the safe heat margin. Within the moving cryogenic window trajectory, candidate rendezvous nodes that meet the constraints of docking, window duration, rendezvous synchronization, and thermal state are selected, dynamic rendezvous points are determined, and the support path of the support robot is solved. The target node is updated with dynamic rendezvous points to generate the rendezvous path and unified rendezvous time of the operation robot. The spraying spatiotemporal nodes are adjusted according to the rendezvous area and the duration of component exchange to form the spraying control sequence of the rendezvous area and the dynamic rendezvous planning results.

9. The dynamic path planning method for collaborative paths of parking lot emergency fire protection robot clusters according to claim 2, characterized in that, S5 specifically includes: The control robot and support robot enter the dynamic rendezvous point, complete the alignment according to the relative posture and relative speed, exchange the replaceable heat-absorbing protection components, and form component exchange completion information based on component identification, positioning signal and locking status. Based on the information from the completion of component exchange and the status data of the robot cluster after the exchange, robot roles are reassigned, and target travel paths, evacuation paths, and target spatiotemporal jet sequences are generated to form an updated cluster collaborative path.

10. The parking lot emergency fire protection robot cluster collaborative path dynamic planning method according to claim 9, characterized in that, Also includes: The robot swarm is controlled to execute a collaborative path. Based on the actual cooling coverage, temperature recovery time, fire source boundary, and cumulative heat absorption, task execution feedback data is generated, and replanning is performed based on the task execution feedback data.

Images