Curtain wall cleaning robot control system

Abstract

The invention belongs to the technical field of intelligent control, particularly relates to a curtain wall cleaning robot control system, and aims at solving the problems that an existing curtain wall cleaning robot is low in cleaning efficiency, safety and automation level. The system comprises a main controller; the main controller is configured to receive data of each sensor of the curtain wall cleaning robot and a remote control instruction sent by a remote controller; based on sensor data and remote control instructions, multi-thread programming is adopted, multiple tasks are processed in parallel, and then the curtain wall cleaning robot is controlled to complete curtain wall cleaning work. The efficiency, the safety and the automation level of curtain wall cleaning operation are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent control, and particularly relates to a control system for a curtain wall cleaning robot. Background Art

[0002] With the development of modern urban architecture, curtain walls, as important architectural elements, have been widely used. However, curtain walls are long-term exposed to the outdoor environment and are prone to pollution and corrosion, thus requiring regular cleaning. The traditional manual cleaning method is not only inefficient but also poses a relatively high safety risk. In recent years, significant progress has been made in robot technology, especially in aspects such as movement, adsorption, operation, and perception. These technological breakthroughs provide strong support for the research and development of curtain wall cleaning robots, enabling robots to more stably and flexibly adapt to various curtain wall cleaning scenarios. With the continuous development of intelligent and automated technologies, all industries are seeking to improve production efficiency and operation quality through intelligent means. The curtain wall cleaning industry is no exception. By introducing intelligent and automated technologies, precise control and optimization of the cleaning process can be achieved, improving the overall operation level and efficiency. Summary of the Invention

[0003] In order to solve the above problems in the prior art, that is, to solve the problems of low efficiency, low safety, and low automation level in the cleaning operation of existing curtain wall cleaning robots, the present invention provides a control system for a curtain wall cleaning robot. The curtain wall cleaning robot includes a cleaning trolley and a winch module. The control system for the curtain wall cleaning robot includes: a main controller;

[0004] The main controller is configured to receive data from various sensors of the curtain wall cleaning robot and remote control instructions sent by a remote controller;

[0005] Based on the input data and / or the remote control instructions, multi-threaded programming is adopted to parallelly process multiple tasks, and then control the curtain wall cleaning robot to complete the curtain wall cleaning operation;

[0006] Among them, the threads in multi-threaded programming include a motor control thread, a remote controller instruction receiving thread, a cable reel controller thread, a sensor reading thread, an air pump and water valve control thread, and a system operation monitoring thread.

[0007] In some preferred embodiments, the motors equipped on the curtain wall cleaning robot include a lifting motor, a traveling motor, and a rotary brush motor;

[0008] The motor control thread is configured to determine whether the lifting motor is at the initial point. If so, the motor control thread is started; otherwise, the lifting motor is controlled to return to the initial point, and then the motor control thread is started.

[0009] After the motor control thread is started, if the remote control instruction is received, the remote control instruction is parsed to obtain the type of the remote control instruction, and then the corresponding motor is controlled to perform corresponding actions according to the type.

[0010] In some preferred embodiments, the sensors equipped on the curtain wall cleaning robot include edge sensors, ranging sensors, and infrared sensors;

[0011] The sensor reading thread is configured to start a timer and determine whether the time set by the timer has arrived. If so, a reading program is started to read and integrate the data of each sensor, and the integrated data is sent to the message queue.

[0012] In some preferred embodiments, the air pump and water valve control thread is configured to parse the remote control instruction. If the remote control instruction is a water pump control instruction, the curtain wall cleaning robot is controlled by the water pump to complete the water flow during the curtain wall cleaning operation;

[0013] If the remote control instruction parsed is an air pump control instruction, the air pump motor is controlled to act, and then it is determined whether the air pump control is started. If so, it is determined whether the air pump pressure reaches the set pressure. If the set pressure is not reached, the air pump is controlled to pressurize until the set pressure is reached.

[0014] In some preferred embodiments, the remote control instruction receiving thread is configured to receive the remote control instruction of the remote controller and perform verification on the remote control instruction:

[0015] The method for performing verification on the remote control instruction is:

[0016] A10, set the read timeout parameter according to the configuration file;

[0017] A20, determine whether the remote controller reception times out according to the read timeout parameter. If so, clear the instruction reception buffer. If not, read the received data and perform CRC verification;

[0018] A30, if the CRC verification result is correct, analyze the valid instructions in the remote control instruction in groups and count them separately; otherwise, clear the instruction reception buffer, reset the read timeout parameter, and jump to A20;

[0019] A40, determine whether the count of the valid instructions exceeds the set count threshold. If so, jump to A50. Otherwise, reset the read timeout parameter according to the configuration file and jump to A20;

[0020] For A50, if it is a wrong combination instruction, it is excluded, the read timeout parameter is reset, and it jumps to A20. Otherwise, the valid instruction is output, the valid instruction in the remote control instruction is normally executed, and the read timeout parameter is reset, then it jumps to A20.

[0021] In some preferred embodiments, the cable winder control thread is configured to parse the remote control instruction and control the cable winder to wind and unwind and tow the pipeline; the pipeline includes a power line, a signal line, and a water pipe.

[0022] In some preferred embodiments, the system operation monitoring thread is configured to set the time of the watchdog timer, block and wait for the watchdog timer, and obtain the cumulative dog feeding flag;

[0023] Judge whether dog feeding is required according to the cumulative dog feeding flag. If so, perform the dog feeding operation and read the status data of each motor. Otherwise, directly read the status data of each motor;

[0024] Based on the status data, analyze the status of each motor. If the status is normal, reset the time of the watchdog timer and perform status judgment. Otherwise, send a warning alarm.

[0025] In some preferred embodiments, to control the curtain wall cleaning robot to complete the curtain wall cleaning operation, the method is as follows:

[0026] Transport the cleaning trolley and the winch module to the roof of the building and fix the winch module;

[0027] Connect the power supply and water source interfaces of the winch module to the external power supply and water source;

[0028] Turn on the switches of the water source and the power supply, and adsorb the cleaning trolley on the edge of the curtain wall of the roof, with the head facing down;

[0029] Input the running track data of the cleaning trolley into the remote controller and click to start running; the running track data includes the running height and the overall running width of the cleaning trolley;

[0030] Control the cleaning trolley to run according to the remote control instruction. When the cleaning trolley moves downward, the cleaning components of the cleaning trolley start the cleaning action. During the downward movement, sensor data is obtained through the detection sensor to perform obstacle avoidance actions;

[0031] When the cleaning trolley runs to the set running height, control the cleaning trolley to turn 180° and move upward. The cleaning action is turned off during the upward movement;

[0032] When the cleaning trolley reaches the top of the curtain wall, control the trolley to turn 90°, move horizontally a distance equal to the width of one vehicle body, turn 90° again, and the trolley moves downward;

[0033] The cleaning trolley moves up and down reciprocally. When the lateral movement distance reaches the set overall operating width, it performs the downward movement for the last time. After the downward movement is completed, the cleaning trolley is controlled to stop moving, and the curtain wall cleaning operation is completed.

[0034] In some preferred embodiments, the cleaning trolley is connected to the remote controller by a 433 wireless communication method.

[0035] In some preferred embodiments, when the curtain wall cleaning robot performs the curtain wall cleaning operation, it also includes autonomous path planning, specifically:

[0036] After the curtain wall cleaning robot is started, it collects images and data of the curtain wall and the surrounding environment as input data;

[0037] According to the input data, an environmental map is generated, and then obstacles that may hinder the movement of the curtain wall cleaning robot are identified, and information about the obstacles is determined; the information about the obstacles includes the position and size;

[0038] Combined with the obstacle information, an optimal cleaning path is generated through an improved path planning algorithm:

[0039] S10, combined with the obstacle information, the environmental map is divided into grids of equal size, and the state attribute values of each grid are set, thereby forming a grid map;

[0040] S20, according to the grid where the curtain wall cleaning robot is currently located, the state attribute values of the covered grids are modified, and the next feasible grid is obtained by combining the improved A* algorithm;

[0041] The heuristic rule of the improved A* algorithm is:

[0042] Among them, r j represents the position of the next moving grid, f j represents the attribute value of grid j, grid j belongs to the neighborhood grid of the current grid i of the robot, c represents the uncovered area coefficient, which represents the influence degree of the side with the least number of uncovered grids among the two sides of the original movement direction on the robot's selection of the next grid, G c represents the sum of the number of uncovered grids on the side of the direction of the grid to be moved and the number of uncovered grids in the original movement direction, G max represents the total number of uncovered grids when the robot is in the current grid, w ij represents the distance between the current grid i and the grid j to be moved, s j represents the steering function;

[0043] S30. If the curtain wall cleaning robot gets stuck in a dead zone and there is a grid with a state attribute value of 1 among the next feasible grids, jump to S40. If the curtain wall cleaning robot gets stuck in a dead zone and there is no grid with a state attribute value of 1 among the next feasible grids, jump to S60. Otherwise, loop and execute S20;

[0044] S40. Calculate the distance between the grid where the curtain wall cleaning robot is currently located and the next feasible grid with a state attribute value of 1, and select the grid with the shortest distance as the temporary search point;

[0045] S50. Add a distance to the target point item to the state transition rule of the ant colony algorithm, and then the robot uses it to plan the optimal route to escape the dead zone and jump to S20;

[0046] S60. End the search and obtain the optimal cleaning path.

[0047] In some preferred embodiments, the optimal route to escape the dead zone is planned by an improved ant colony algorithm. The method is as follows:

[0048] When performing state transition, first generate a random number q, where 0 ≤ q ≤ 1, and select the next position according to the following rules:

[0049] where S represents selecting the next position through roulette, τ ij represents the pheromone concentration on the edge, η ij represents the heuristic information, y jg represents the Euclidean distance between the allowed transfer position and the target position, α, β, and λ respectively represent the relative weights of τ ij , η ij , y jg in the path selected by the ant, q0 represents the initial position point, t represents time, and allowed k represents the positions allowed for the next step of ant k;

[0050] When moving from i to j, the local pheromone update process is: τ ij (t + 1) = (1 - μ)τ ij (t) + μτ0

[0051] where μ represents the pheromone evaporation coefficient during local update, and τ0 represents a small normal constant;;

[0052] The global pheromone update process is: τ ij (t + n) = (1 - ρ)τ ij (t) + ρΔτ ij

[0053] Among them, ρ represents the pheromone evaporation coefficient during global update, and L ib represents the optimal path of the current iteration.

[0054] Advantages of the present invention:

[0055] 1) The control system of the present invention is based on the embedded Linux operating system. This choice not only provides a stable and efficient operating environment for the robot, but also lays a solid foundation for its subsequent function expansion and performance optimization;

[0056] 2) The present invention adopts a multi-threaded programming method. By parallel processing multiple tasks, it not only effectively improves the concurrent processing ability of the system, but also significantly shortens the switching time between tasks, enabling the robot to respond quickly in the face of emergencies, ensuring the efficiency and safety of the cleaning operation, and improving the automation level. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] By reading the detailed description of the non-limiting embodiments with reference to the following drawings, other features, objects, and advantages of the present application will become more apparent:

[0058] Figure 1 is a schematic framework diagram of a curtain wall cleaning robot control system according to an embodiment of the present invention;

[0059] Figure 2 is a schematic structural diagram of a cleaning cart according to an embodiment of the present invention;

[0060] Figure 3 is a schematic structural diagram of a walking component according to an embodiment of the present invention;

[0061] Figure 4 is a schematic structural diagram of an adsorption component according to an embodiment of the present invention;

[0062] Figure 5 is a schematic structural diagram of a cleaning component according to an embodiment of the present invention;

[0063] Figure 6 is a schematic structural diagram of a detection component according to an embodiment of the present invention;

[0064] Figure 7 is a schematic structural diagram of a water and power supply component according to an embodiment of the present invention;

[0065] Figure 8 is a schematic structural diagram of a hoisting module according to an embodiment of the present invention;

[0066] Figure 9 is a schematic framework diagram of multi-threaded programming according to an embodiment of the present invention;

[0067] Figure 10 It is a schematic diagram of the control process of the motor control thread in an embodiment of the present invention;

[0068] Figure 11 It is a schematic diagram of the reading process of the sensor reading thread in an embodiment of the present invention;

[0069] Figure 12 It is a schematic diagram of the control process of the air pump and water valve control thread in an embodiment of the present invention;

[0070] Figure 13 It is a schematic diagram of the process of verifying remote control instructions by the remote control instruction receiving thread in an embodiment of the present invention;

[0071] Figure 14 It is a schematic diagram of the control process of the wire winder control thread in an embodiment of the present invention;

[0072] Figure 15 It is a schematic diagram of the operation process of the system operation monitoring thread in an embodiment of the present invention;

[0073] Figure 16 It is a schematic diagram of the path when the cleaning cart performs curtain wall cleaning operations in an embodiment of the present invention;

[0074] Figure 17 It is a schematic diagram of the process of the improved path planning algorithm in an embodiment of the present invention.

[0075] Figure 18 It is a schematic diagram of the robot grid map in an embodiment of the present invention.

[0076] Figure 19 It is a schematic diagram of the process of determining transition points in an embodiment of the present invention. Detailed implementation manners

[0077] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0078] The following further elaborates on the present application with reference to the accompanying drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the relevant invention and are not intended to limit the invention. Additionally, it should be noted that for the sake of description, only parts related to the relevant invention are shown in the accompanying drawings.

[0079] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The following will describe the present application in detail with reference to the drawings and in combination with the embodiments.

[0080] To more clearly illustrate the curtain wall cleaning robot control system of the present invention, each module in the system embodiment of the present invention will be described in detail below with reference to the drawings.

[0081] For the completion cleaning operations of conventional exterior wall decoration materials such as glass, aluminum plates, and stones on high-rise buildings with non-shaped and simple-shaped exterior walls, a curtain wall cleaning robot has been developed and designed. This robot is applicable to high-altitude curtain walls with a facade angle not greater than 90°. Through an intelligent control method, it can achieve automatic or semi-automatic control of the curtain wall robot, improve the construction conditions of workers, greatly reduce the labor intensity of workers, and can significantly reduce the cleaning costs.

[0082] The curtain wall cleaning robot includes a cleaning cart and a hoisting module;

[0083] The cleaning cart includes a traveling component, an adsorption component, a cleaning component, a detection component, water, and a power supply component, as Figure 2 shown;

[0084] Traveling component: It completes functions such as the vertical and horizontal movement, turning, and obstacle crossing of the cart; the traveling component realizes the traveling function by two groups of wheels separately driven by two motors. When the motors on both sides of the cart rotate in the same direction, vertical or horizontal movement can be achieved. When the motors on both sides rotate in opposite directions, the turning function is realized; the traveling component realizes the traveling function by two groups of wheels separately driven by two motors. When the motors on both sides of the cart rotate in the same direction, vertical or horizontal movement can be achieved. When the motors on both sides rotate in opposite directions, the turning function is realized; as Figure 3 shown;

[0085] Adsorption component: It completes the adsorption of the cart on the curtain wall and also has the function of obstacle crossing; the adsorption component is mainly composed of 4 groups of vacuum suction cups and a vacuum generator. Each group of suction cups can move up and down independently, and each suction cup has a certain degree of compressibility; when the cart is traveling on a flat surface, each group of suction cups extends the same length at the same time to ensure complete adsorption; when the cart is crossing an obstacle, the cart will have a certain inclination angle, and it relies on the compressibility of the suction cups to adapt to the angle change; when the suction cup passes over the obstacle, the suction cup actively retracts to the initial position to avoid the obstacle; as Figure 4 shown;

[0086] Cleaning component: It completes the cleaning of the curtain wall and includes components such as atomizing nozzles, rolling brushes, and scrapers, as Figure 5As shown, the stain and dust are washed down from the curtain wall surface through the atomizing nozzle, and the rotary brush and scraper are used to further remove the residues to ensure the cleanliness of the curtain wall surface. The cleaning component is connected to the car body through the servo motor and can be actively lifted during obstacle crossing.

[0087] Detection component: It has the functions of curtain wall edge detection and obstacle detection, and is linked with the car walking and adsorption components for control. The detection component mainly includes the detection sensors (or edge detection sensors) and their brackets located at the four corners of the car, 2 distance sensors, and 4 infrared sensors. When the detection component detects the curtain wall edge, the car will actively stop. When the detection component detects an obstacle, it will be linked with the walking and adsorption components to perform obstacle avoidance actions. As Figure 6 shown;

[0088] Water and power supply component: The water supply and power supply component of the car can realize continuous water supply and power supply during the car turning process. The water and power supply component mainly includes the slip ring on the top of the car and the integrated water and power cable. As Figure 7 shown;

[0089] In addition, it also includes a structural bracket, which is the main framework of the curtain wall cleaning robot and is used to support and connect other components. It is usually made of high-strength and lightweight materials to ensure the stability and durability of the robot, while reducing the overall weight to facilitate movement on the curtain wall surface. Preferably, the weight of the cleaning car does not exceed 30 kg, the winch module does not exceed 110 kg, the power is less than or equal to 1.5 Kw, the winch module is less than or equal to 2.5 Kw, the adsorption force is greater than 50 kg, the wind resistance / wind resistance level: 17.2 - 20.7 m / s, grade eight, the waterproof level: IP42, the traveling mode: wheeled movement, the boundary sensor protects against falling, and it has autonomous obstacle avoidance; the cleaning efficiency: ≥1000 sqm / day, the adsorption method: vacuum suction cup, the working surface angle: 0 - 90°.

[0090] The winch module includes a wire winding component and a frame component, as Figure 8 shown;

[0091] Wire winding component: It completes the winding function of the water and electric wires, and at the same time has the function of preventing the car from falling, and is provided with water and power supply interfaces. The wire winding component includes a wire winding drum, a drum drive motor, a wire arranging device, and a tension detection device; the drum is used for winding and unwinding the wire, and a slip ring is arranged on the axis of the drum, and the slip ring is connected to the power supply and water source interfaces; the wire arranging device mainly includes a reciprocating lead screw, which is driven by the drum motor to realize the orderly arrangement of the wires and avoid winding; the tension detection device is used to detect the tension of the wires and is linked with the wire winding drum motor to realize the active winding and unwinding of the wires.

[0092] Frame component: It is provided with wheels and can be pushed. The frame component includes a frame, wheels, a wire winding bracket, and a wire winding bracket support rod. The tension detection device of the wire winding component is arranged on the wire winding bracket.

[0093] To achieve the intelligent control and management of the curtain wall cleaning robot and its supporting cable coiler, aiming to improve the efficiency, safety and automation level of curtain wall cleaning operations. During system operation, control systems for robot motion control, cable coiler control, robot status monitoring, etc. are required to ensure the system is easy to operate and maintain to the greatest extent, with stable operation and secure and reliable data storage.

[0094] The control system of the curtain wall cleaning robot in the present invention is designed with its core rooted in the embedded Linux operating system. This choice not only provides a stable and efficient operating environment for the robot but also lays a solid foundation for its subsequent function expansion and performance optimization. To further improve the real-time performance of the system and ensure rapid response and precise execution in complex and changeable cleaning tasks, the present invention ingeniously adopts a multi-threaded programming method. By processing multiple tasks in parallel, multi-threaded programming not only effectively improves the concurrent processing ability of the system but also significantly shortens the switching time between tasks, enabling the robot to quickly respond in case of emergencies and ensuring the efficiency and safety of cleaning operations. Through CAN and 485 bus communication, and through reasonable planning and analysis of system functions, six main threads are mainly created, namely the motor control thread, the remote control command receiving thread, the cable coiler control thread, the sensor reading thread, the air pump and water valve control thread, and the system operation monitoring thread. Through global variables, read-write control locks, semaphores, mutexes, etc., the read-write control of process data and the implementation of control logic are achieved among threads. Finally, the six threads work together to stably and reliably complete the system functions. Specifically as follows:

[0095] The control system of the curtain wall cleaning robot includes a main controller. The main control part uses an NXP imx6ull embedded development platform, which has interfaces such as two CANs, three 485s, one TTL, and multiple GPIOs.

[0096] The main controller is configured to receive data from various sensors of the curtain wall cleaning robot and remote control commands sent by the remote controller; based on the input data and / or the remote control commands, multi-threaded programming is adopted to process multiple tasks in parallel, and then control the curtain wall cleaning robot to complete the curtain wall cleaning operation.

[0097] Among them, the threads during multi-threaded programming include the motor control thread, the remote control command receiving thread, the cable coiler control thread, the sensor reading thread, the air pump and water valve control thread, and the system operation monitoring thread, as Figure 9 shown.

[0098] 1) Motor control thread

[0099] The curtain wall cleaning robot is equipped with a variety of motors to drive different moving parts, such as the walking mechanism, the rotary brush, the lifting mechanism, etc. These motors need to be driven by corresponding drivers, which receive instructions from the controller to control the rotation speed, steering, start and stop of the motors. The curtain wall cleaning robot has four walking motors responsible for the movement of the walking mechanism. Through the remote control, the robot can move forward, backward, rotate clockwise, rotate counterclockwise, and turn during movement.

[0100] One rotary brush motor is responsible for the movement of the rotary brush. Through the remote control, the start and stop of the rotary brush, the switching of high, medium and low speeds, etc. can be realized; two lifting motors are responsible for the obstacle crossing function of the robot. Through the remote control, the active obstacle avoidance and automatic obstacle avoidance functions can be set.

[0101] The detailed process of the motor control thread controlling the motor is as follows: first, judge whether the lifting motor is at the initial point. If so, the motor control thread starts. Otherwise, control the lifting motor to return to the initial point and the motor control thread starts;

[0102] After the motor control thread starts, if it receives the remote control instruction, it parses the remote control instruction to obtain the type of the remote control instruction, and then controls the corresponding motor to perform corresponding actions according to the type, as Figure 10 shown.

[0103] 2) Sensor reading thread

[0104] The curtain wall cleaning robot integrates a precise sensor system to ensure its efficient and safe operation in a complex working environment. Among them, edge detection is a key safety link when the robot performs tasks. For this reason, the robot is equipped with four high-sensitivity edge detection sensors, which can real-time sense and accurately identify whether the robot is approaching the edge of the curtain wall, effectively preventing the occurrence of falling accidents and greatly improving the safety of the operation. In order to accurately control the distance between the robot and potential obstacles and ensure the smooth progress of the cleaning operation, the robot also has two laser range sensors built in. These range sensors use advanced laser technology to quickly and accurately measure the distance between the robot and surrounding obstacles, providing a basis for the robot's dynamic path planning, ensuring that it can also move flexibly in narrow or complex spaces and avoid collisions. To meet the need for rapid response to close-range obstacles, the robot is additionally equipped with four infrared sensors, two of which are dedicated to close-range detection. These infrared sensors feature fast response speed and high detection accuracy, and can effectively identify and feedback the presence of obstacles within 5 cm, providing immediate obstacle avoidance instructions for the robot.

[0105] The specific process of the sensor reading thread reading is as follows: start the timer, and judge whether the time set by the timer is reached. If so, start the reading program, read the data of each sensor and integrate them, and send the integrated data to the message queue, asFigure 11 as shown

[0106] 3) Air pump and water valve control thread

[0107] The curtain wall cleaning robot body has a water valve and two air pumps; the two air pumps respectively control six air valves, and the opening and closing of the water valve and air pumps can be controlled through a remote controller.

[0108] such as Figure 12 as shown, the specific control process of the air pump and water valve control thread is: parse the remote control instruction, if the remote control instruction is a water pump control instruction, then control the curtain wall cleaning robot to complete the water flow during the curtain wall cleaning operation through the water pump;

[0109] if the remote control instruction is an air pump control instruction, then control the air pump motor to act, and then judge whether the air pump control is started. If so, judge whether the air pump pressure reaches the set pressure. If it does not reach the set pressure, control the air pump to pressurize until the set pressure is reached.

[0110] 4) Remote controller instruction receiving thread

[0111] The remote controller instruction receiving thread of the curtain wall cleaning robot is a key component in the curtain wall cleaning robot control system. This thread is mainly responsible for receiving instructions from the ground remote controller and transmitting these instructions to the main controller of the robot to achieve remote control and monitoring of the robot. By adopting the 433 wireless communication method, the instruction receiving thread can obtain the operator's instructions in real time and accurately, ensuring that the robot can perform the cleaning operation according to the predetermined trajectory. At the same time, this thread also needs to have certain error handling and instruction verification functions to ensure that the received instructions are valid and reliable, thereby improving the efficiency and safety of the entire cleaning operation.

[0112] such as Figure 13 as shown, the method for the remote controller instruction receiving thread to verify the remote control instruction is:

[0113] A10, set the read timeout parameter according to the configuration file;

[0114] A20, judge whether the remote controller reception times out according to the read timeout parameter. If so, clear the instruction reception buffer. If not, read the received data and perform CRC verification;

[0115] A30, if the CRC verification result is correct, analyze the valid instructions in the remote control instruction in groups and count them separately; otherwise, clear the instruction reception buffer, reset the read timeout parameter, and jump to A20;

[0116] A40, determine whether the count of the valid instructions exceeds a set count threshold. If so, jump to A50; otherwise, reset the read timeout parameter according to the configuration file and jump to A20;

[0117] A50, if it is an incorrect combination instruction, discard it, reset the read timeout parameter, and jump to A20; otherwise, output the valid instruction, normally execute the valid instruction in the remote control instruction, reset the read timeout parameter, and jump to A20.

[0118] 5) Reel control thread

[0119] The reel of the curtain wall cleaning robot is indeed one of its important components. It not only ensures the safe operation of the robot during the cleaning task, preventing accidental falls, but also undertakes the crucial task of providing power and water source for the robot. Through a precisely designed mechanical structure and control system, the reel ensures the orderly retraction and extension and stable traction of the robot's cables (including power cables, signal cables, and water pipes). When the robot moves up and down along the curtain wall, the reel can automatically adjust the length of the cables, keep the cables in a tensioned state, and avoid the risk of the robot getting out of control or falling due to cable slack or entanglement. At the same time, the cable management system integrated inside the reel can also ensure the smooth transmission of power lines and water pipes, providing continuous and stable power supply and the water flow required for cleaning for the robot. This design enables the robot to maintain efficient and stable cleaning performance at high altitudes far from the ground without worrying about power or water source interruption. The reel of the curtain wall cleaning robot can be controlled to wind and unwind the cable through the remote control. The control process of the reel control thread of the curtain wall cleaning robot is as Figure 14 shown, specifically: Parse the remote control instruction and control the reel to retract and extend the pipeline and traction; the pipeline includes a power cable, a signal cable, and a water pipe.

[0120] 6) System operation monitoring thread

[0121] The system operation monitoring thread of the curtain wall cleaning robot is a key background process to ensure the robot maintains high efficiency, stability, and safety during high-altitude cleaning tasks, and it can ensure that the robot does not crash during operation. This thread continuously monitors each thread of the robot. The monitoring thread can quickly detect any abnormality or potential fault and immediately send an alarm to the main controller for timely response measures. The specific operation process of the system operation monitoring thread is as Figure 15 shown, specifically:

[0122] Set the time of the watchdog timer, block and wait for the watchdog timer, and obtain the cumulative dog feeding flag;

[0123] Judge whether it is necessary to feed the dog according to the cumulative dog feeding identification. If it is necessary, perform the dog feeding operation and read the status data of each motor. Otherwise, directly read the status data of each motor;

[0124] Based on the status data, analyze the status of each motor. If the status is normal, reset the time of the watchdog timer and perform status judgment. Otherwise, send a warning alarm.

[0125] In addition, when controlling the curtain wall cleaning robot to complete the curtain wall cleaning operation, there are two control methods. One method is manual control, and the other is automatic control (that is, under relatively ideal environmental conditions, after setting the boundary parameters, the cleaning trolley automatically performs cleaning, and can automatically avoid obstacles when encountering obstacles during the cleaning process. In the manual control mode, an operator remotely controls the trolley to move forward, and the trolley independently performs cleaning and obstacle avoidance). The paths of the two cleaning operations are as Figure 16 shown. In this embodiment, automatic control is preferably used.

[0126] The specific process of the manual-controlled curtain wall cleaning operation is as follows:

[0127] Transport the trolley module and the hoist module to the top of the building and fix the hoist module;

[0133] Connect the power supply and water source interfaces of the hoist module to an external power supply (preferably 220V) and a water source;

[0128] Turn on the water source and power switch, and adsorb the trolley on the edge of the curtain wall of the top of the building, with the head facing down;

[0129] Remotely control the trolley to move downward, and the cleaning component of the trolley starts the cleaning action. During the downward movement, the trolley can actively perform obstacle avoidance actions through the detection sensor;

[0130] When the ground control personnel reach the bottom of the curtain wall, remotely control the trolley to turn (make a 180° U-turn in place), and the trolley moves upward without performing cleaning actions during the upward movement;

[0131] When the trolley reaches the top of the curtain wall, the ground control personnel control the trolley to turn 90°, move horizontally a distance equal to the width of the vehicle body, and then turn 90° again to make the trolley move downward;

[0132] The trolley moves downward and repeats the previous actions

[0133] The specific process of the automatic-controlled curtain wall cleaning operation is as follows:

[0134] Transport the cleaning trolley and the hoist module to the top of the building and fix the hoist module;

[0135] Connect the power supply and water source interfaces of the hoist module to an external power supply (preferably 220V) and a water source;

[0136] Turn on the switches of the water source and the power supply, and adsorb the cleaning cart on the edge of the curtain wall of the roof, with the head facing downwards;

[0137] Input the cleaning cart running trajectory data into the remote controller and click to start running; the running trajectory data includes the running height and the overall running width of the cleaning cart;

[0138] Control the cleaning cart to run according to the remote control instruction. When the cleaning cart moves downward, the cleaning component of the cleaning cart starts the cleaning action. During the downward movement, sensor data is obtained through the detection sensor to perform obstacle avoidance actions;

[0139] When the cleaning cart runs to the set running height, control the cleaning cart to turn 180° and move upward, and turn off the cleaning action during the upward movement;

[0140] When the cleaning cart reaches the top of the curtain wall, control the cart to turn 90°, move horizontally a distance equal to the width of one vehicle body, turn 90° again, and the cart moves downward;

[0141] The cleaning cart moves up and down reciprocally. When the horizontal movement distance reaches the set overall running width, perform the downward movement for the last time. After the downward movement is completed, control the cleaning cart to stop moving, and complete the curtain wall cleaning operation.

[0142] The specific planned path is described in detail as follows in the implementation manner:

[0143] Environmental perception module: After the robot is started, it first collects images and data of the curtain wall and its surrounding environment through the camera and the ranging module sensor. These data are transmitted to the environmental perception module for processing and analysis to generate an environmental map or model.

[0144] Obstacle Recognition Module: In the environmental map or model, the obstacle recognition module uses deep learning or image processing techniques to identify obstacles that may hinder the movement of the robot and determine their positions and sizes. This information is stored in the robot's memory for subsequent path planning. Path Planning Module: Based on the results of environmental perception and obstacle recognition, the path planning module adopts a full-coverage path planning method. It needs to find the shortest collision-free path from the starting position to the end point. More importantly, it needs to scan and traverse the entire obstacle-free working area, and finally form a continuous path that passes through all obstacle-free areas in the working area from the starting position. At the same time, considering the kinematic constraints and dynamic characteristics of the robot, the path is further optimized and adjusted. The Priority A* algorithm defines the heuristic rule as the priority order of the upper grid, lower grid, and left grid. The rule is simple and feasible, but the robot turns frequently and consumes a large amount of energy. The Confidence Direction Function considers the robot's turning and uses the direction function strategy to make the planned path keep going straight as much as possible, reducing energy consumption. However, this is similar to the random movement method of the robot, and the robot is prone to getting into a deadlock state. Edge Following is the process in which the robot moves horizontally and vertically along the environmental boundary or the obstacles close to the environmental boundary, and detects the relative distance between the environmental or obstacle boundary and the robot through the ranging sensors on the left and right sides of the robot body, realizing the learning of the contour of the working environment and laying a foundation for the subsequent grid-based environmental modeling. For a complex environment, the mobile robot may need to cross the interior of the environment to fully learn the contour of the entire working environment. After obtaining the environmental map through environmental contour learning, it is divided into several grids of a fixed size. The grids are divided into obstacle grids, covered grids, and uncovered grids according to whether they contain obstacles and whether they are covered. For the convenience of information storage and computational processing, the state of each grid is described by (x, y, f), where (x, y) represents the position of the grid in the map, and f is the attribute value of the grid. The attribute value f of each grid is set according to the following formula.

[0145] A dynamic grid means that the attribute value f of the grid is variable. For example, to increase the "attraction" of the uncovered grid to the robot and avoid the robot repeatedly wiping the same grid, it is agreed that each time the grid is covered, its attribute value decreases by -1. At the same time, the robot only updates the attribute values of the grids that have been covered during the working process, without the need to perform iterative calculations on all neurons or grids according to the shunt equation like the bio-inspired neural network environmental model, greatly reducing the computational amount and improving the working efficiency of the robot. Suppose the robot grid map is as Figure 18 shown.

[0146] For the convenience of research and simulation experiments, the following assumptions are made: (1) During the movement of the robot, the obstacles and the environment remain unchanged all the time, that is, the robot works in a static environment and the obstacles are in regular shapes. (2) The boundaries of the environment and the obstacles have been "puffed up" according to the actual physical size of the robot. The boundary of the obstacle is the safe area and the robot is represented by a mass point. At the same time, when there are no obstacles and environmental boundaries, the robot can move to the grids in 8 directions of the neighborhood. (3) After the robot covers a grid, it is considered that it has completed the work as required, such as cleaning or wiping work. When the same grid is covered again, it belongs to repeated traversal.

[0147] Based on the environmental model, the robot continuously selects the grid to be moved according to the established priority rules. When the robot falls into a dead zone, an escape algorithm is used to plan a path to quickly escape from the dead zone and continue the established task, ultimately achieving the goal of full coverage. Based on the environmental modeling of the dynamic grid method, the priority rules and the escape algorithm are directly related to the area coverage rate and the trajectory repetition rate of the robot, and are the key points in the research of full coverage path planning.

[0148] The priority A* algorithm defines the heuristic rule as the priority order of the upper grid, the lower grid, and the left grid. The rule is simple and feasible, but the robot turns frequently and consumes a large amount of energy. The confidence direction function considers the robot's turning and uses the direction function strategy to make the planned path as straight as possible, reducing energy consumption. However, this is similar to the random movement method of the robot, and the robot is prone to falling into a deadlock state. For the requirements of full coverage work, the robot often falls into a dead zone during the movement in the environmental map constructed by the dynamic grid method. The robot falling into a dead zone means that its surrounding adjacent grids are a combination of boundaries, obstacle grids, and covered grids, and there are no uncovered grids. Only by escaping from the dead zone can the full coverage task continue, and the path to escape from the dead zone affects the path repetition rate of full coverage. When the robot falls into a dead zone using the priority heuristic rule, this paper uses the ant colony algorithm to quickly plan the optimal escape path and solves this problem well. Search for a temporary target point When the robot falls into a dead zone, before using the ant colony algorithm to escape from the dead zone, a temporary search point must be established as the escape target point of the robot. It requires that the grid is not covered and the distance from the current position of the robot is the shortest. At the same time, the selection of the target point affects the level of the repetition rate of the robot. Based on the full consideration of obstacles, this paper selects the point closest to the actual distance of the robot as the temporary target point, thus greatly reducing the repetition rate. Assume that the current position of the robot is P r (x c ,y c ), and the position of the grid to be selected is P i (x i ,y i), i = 1, ..., m, where m is the total number of candidate target points that meet the conditions. When there is no obstacle between the robot and the candidate point, the distance d between the robot's position and the candidate grid i is given by the formula:

[0149] When there is an obstacle between the robot and the candidate point, the obstacle position can be described as O ri ={A(x 1i , y 1i ), B(x 1i , y 2i ), C(x 2i , y 1i ), D(x 2i , y 2i )}. First, the vertex of the obstacle that is closest to the robot's current position and has the shortest perpendicular distance to the line P r P i is used as the transition point. For example, in the following figure, point B is the transition point when there is an obstacle. As shown in Figure 19 , then the distance d is calculated according to the formula i :

[0150] where: θ c represents the angle between the robot's current position, the transition point, and the coordinate axis, and θ i represents the angle between the candidate grid, the transition point, and the coordinate axis.

[0151] After obtaining the distances d i between all candidate grids and the robot's position, the grid with the minimum d i is found and its position coordinates are recorded as p0(x0, y0). This grid is used as the temporary search point when the robot escapes from the dead zone. When the temporary search point is determined, the process of the robot escaping from the dead zone is transformed into a point-to-point path planning problem. The ant colony algorithm can quickly plan the optimal escape path and solve this problem well. In the traditional ant colony algorithm, when ant k selects the next position j at time t, it tends to choose the short path with a high pheromone concentration as the moving direction, ignoring the requirement of global optimization in path planning. Therefore, a term representing the distance between the position to be moved and the target position is introduced into the state transition rule to make the search more directional. When an ant makes a state transition, it first generates a random number 0 ≤ q ≤ 1. If q ≤ q0 (q0 represents the initial position point), then the next position is selected according to the following state transition rule formula, otherwise it is selected according to the following roulette wheel method.

[0152] where: S represents the next position selected according to the roulette wheel method.

[0153] where: j ∈ allowed k , representing the positions where ant k is allowed to move next; τ ij represents the pheromone concentration on edge (i, j); η ij represents a heuristic information, usually η ij = 1 / d ij ; y jg represents the Euclidean distance between the allowed transfer position and the target position; α, β, and λ respectively reflect the relative weights of pheromone concentration, heuristic information, and target position information in the path selection of ants. For pheromone update, to effectively prevent ants from converging to the same path, local pheromone update is required when an ant moves from city i to city j, and the rule is shown in the following formula: τ ij (t + 1) = (1 - μ)τ ij (t) + μτ0

[0154] where μ represents the pheromone evaporation coefficient, 0 < μ < 1; τ0 represents a small positive constant.

[0155] After all ants complete one iteration, global pheromone update is performed. According to the principle of the Max - Min Ant System, only the pheromone on the currently optimal path of the iteration is enhanced. At the same time, to avoid stagnation, upper and lower limits τ min ≤ τ ij (t) ≤ τ max are set for the pheromone concentration in each iteration. The global pheromone update rule is shown in the following formula: τ ij (t + n) = (1 - ρ)τ ij (t) + ρΔτ ij

[0156] where ρ represents the pheromone evaporation coefficient, 0 < ρ < 1; L ib represents the currently optimal path of the iteration.

[0157] Aiming at the problems existing in the traditional algorithm, on the basis of considering grid attributes and robot turning, a neighborhood grid distance term and an uncovered area size term are introduced, and a priority heuristic rule as shown in Equation 1 is proposed for the robot to select the next moving grid position in the non - dead - zone state.

[0158] where r j represents the next moving grid position, f jRepresents the attribute value of grid j. Grid j belongs to the neighborhood grid of the current grid i of the robot. c represents the uncovered area coefficient, indicating the influence degree of the side with the least number of uncovered grids on both sides of the original movement direction on the robot's selection of the next grid. Take c = 0.5, G c Represents the sum of the number of uncovered grids on one side of the direction of the proposed moving grid and the number of uncovered grids in the original movement direction, G max Represents the total number of uncovered grids when the robot is at the current grid, w ij Represents the distance between the current grid i and the proposed moving grid j, s j Represents the turning function.

[0159] Flowchart of the improved priority ant colony algorithm (i.e., path planning flowchart), as Figure 17 shown below:

[0160] S10, Environment modeling. Use the sensors on the robot body to perform edge learning on the environment to obtain the environment contour, divide it into equal-sized grids, and set attribute values according to the status of each grid, and finally form a grid map.

[0161] S20, Path selection. The robot searches for the next feasible grid according to the improved priority rule and modifies the attribute values of the covered grids. This cycle continues until there are no uncovered grids in the neighborhood grid of the robot's current position.

[0162] S30, If the curtain wall cleaning robot falls into a dead zone and there is a grid with a status attribute value of 1 among the next feasible grids, then jump to S40. If the curtain wall cleaning robot falls into a dead zone and there is no grid with a status attribute value of 1 among the next feasible grids, jump to S60. Otherwise, loop and execute S20;

[0163] S40, Temporary target point. When there is no obstacle between the robot and the candidate point, calculate the distance between them according to the formula to find the distance between them, and finally select the grid with the shortest distance as the temporary search point (that is, calculate the distance between the grid where the curtain wall cleaning robot is currently located and the next feasible grid with a status attribute value of 1, and select the grid with the shortest distance as the temporary search point).

[0164] S50, Escape from the dead zone. Based on the position of the temporary search point and the grid where the curtain wall cleaning robot is currently located, plan the optimal route to escape from the dead zone through the improved ant colony algorithm, and then transfer to S20 to continue the path full-coverage work.

[0165] S60, End of search. If there is no grid with an attribute value of 1 in the grid map, it means that the full-coverage work has been completed and the algorithm ends.

[0166] Real-time adjustment module: During the cleaning process, the robot continuously monitors changes in the surrounding environment through sensors. When new obstacles or changes in lighting conditions are detected, the real-time adjustment module adjusts the cleaning path and speed according to these changes to ensure the smooth progress of the cleaning task.

[0167] It should be noted that the curtain wall cleaning robot control system provided in the above embodiments is only illustrated by the division of the above function modules. In actual applications, the above functions can be allocated to different function modules according to needs, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiments can be combined into one module, or further split into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing each module or step, and are not regarded as an improper limitation of the present invention.

[0168] Those skilled in the art should be able to realize that the modules and method steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computer software, or a combination of the two. The programs corresponding to the software modules and method steps can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the technical field. To clearly illustrate the interchangeability of electronic hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in the form of electronic 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 to exceed the scope of the present invention.

[0169] The terms "first", "second", "third", etc. are used to distinguish similar objects, rather than to describe or represent a specific order or sequence. So far, the technical solutions of the present invention have been described in combination with the preferred embodiments shown in the drawings. However, those skilled in the art can easily understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the protection scope of the present invention.

Claims

1. A control system for a curtain wall cleaning robot, the curtain wall cleaning robot comprising a cleaning trolley and a hoisting module, characterized in that, The control system of the curtain wall cleaning robot includes: a main controller; The main controller is configured to receive data from various sensors of the curtain wall cleaning robot and remote control instructions sent by a remote controller; Based on the data from the sensors and the remote control instructions, multi-threaded programming is adopted to process multiple tasks in parallel, and then control the curtain wall cleaning robot to complete the curtain wall cleaning operation; Among them, the threads in multi-threaded programming include a motor control thread, a remote controller instruction receiving thread, a cable winder control thread, a sensor reading thread, an air pump and water valve control thread, and a system operation monitoring thread.

2. The control system of a curtain wall cleaning robot according to claim 1, wherein, The motors equipped on the curtain wall cleaning robot include a lifting motor, a traveling motor, and a brush rolling motor; The motor control thread is configured to determine whether the lifting motor is at the initial point. If so, the motor control thread is started; otherwise, the lifting motor is controlled to return to the initial point, and then the motor control thread is started. After the motor control thread is started, if the remote control instruction is received, the remote control instruction is parsed to obtain the type of the remote control instruction, and then the corresponding motor is controlled to perform the corresponding action according to the type.

3. The control system of a curtain wall cleaning robot according to claim 1, characterized in that, The sensors equipped on the curtain wall cleaning robot include an edge sensor, a ranging sensor, and an infrared sensor; The sensor reading thread is configured to start a timer and determine whether the time set by the timer has arrived. If so, a reading program is started to read and integrate the data of each sensor, and the integrated data is sent to a message queue.

4. The curtain wall cleaning robot control system according to claim 1, wherein, The air pump and water valve control thread is configured to parse the remote control instruction. If the remote control instruction is a water pump control instruction, the curtain wall cleaning robot is controlled by the water pump to complete the control of the water flow rate during the curtain wall cleaning operation; If the remote control instruction parsed is an air pump control instruction, the air pump motor is controlled to act, and then it is determined whether the air pump control is started. If so, it is determined whether the air pump pressure reaches the set pressure. If the set pressure is not reached, the air pump is controlled to pressurize until the set pressure is reached.

5. The control system of a curtain wall cleaning robot according to claim 1, characterized in that, The remote controller instruction receiving thread is configured to receive the remote control instruction from the remote controller and verify the remote control instruction; The method for verifying the remote control instruction is as follows: A10, set a read timeout parameter according to the configuration file; A20, determine whether the remote controller reception times out according to the read timeout parameter. If so, clear the instruction reception buffer; if not, read the received data and perform a CRC check; A30, if the CRC check result is correct, analyze the valid instructions in the remote control instruction by grouping and count them separately; Otherwise, clear the instruction reception buffer, reset the read timeout parameter, and jump to A20; A40, determine whether the count of the valid instructions exceeds the set count threshold. If so, jump to A50; otherwise, reset the read timeout parameter according to the configuration file and jump to A20; A50, if it is an incorrect combination instruction, eliminate it, reset the read timeout parameter, and jump to A20; otherwise, output the valid instruction, normally execute the valid instruction in the remote control instruction, reset the read timeout parameter, and jump to A20.

6. The control system of a curtain wall cleaning robot according to claim 2, characterized in that, The coiler control thread is configured to parse the remote control instruction and control the coiler to wind and unwind and tow the pipeline; the pipeline includes a power line, a signal line, and a water pipe.

7. The control system of a curtain wall cleaning robot according to claim 6, characterized in that, The system operation monitoring thread is configured to set the time of the watchdog timer, block and wait for the watchdog timer, and obtain the cumulative dog feeding flag; Judge whether dog feeding is required according to the cumulative dog feeding flag. If so, perform the dog feeding operation and read the status data of each motor. Otherwise, directly read the status data of each motor; Based on the status data, analyze the status of each motor. If the status is normal, reset the time of the watchdog timer and perform status judgment. Otherwise, send a warning alarm.

8. The control system of a curtain wall cleaning robot according to claim 1, characterized in that, Control the curtain wall cleaning robot to complete the curtain wall cleaning operation. The method is as follows: Transport the cleaning trolley and the hoisting module to the roof of the building and fix the hoisting module; Connect the power supply and water source interfaces of the hoisting module to the external power supply and water source; Turn on the switches of the water source and the power supply, and adsorb the cleaning trolley on the edge of the curtain wall of the roof, with the head facing down; Input the running track data of the cleaning trolley into the remote controller and click to start running; the running track data includes the running height and the overall running width of the cleaning trolley; Control the cleaning trolley to run according to the remote control instruction. When the cleaning trolley moves downward, the cleaning component of the cleaning trolley starts the cleaning action. During the downward movement, sensor data is obtained through the detection sensor to perform obstacle avoidance actions; When the cleaning trolley runs to the set running height, control the cleaning trolley to turn 180°, move upward, and turn off the cleaning action during the upward movement; When the cleaning trolley reaches the top of the curtain wall, control the trolley to turn 90°, move horizontally by a distance equal to the width of one vehicle body, turn 90° again, and the trolley moves downward; The cleaning trolley moves up and down reciprocally. When the horizontal movement distance reaches the set overall running width, perform the downward movement for the last time. After the downward movement is completed, control the cleaning trolley to stop moving, and complete the curtain wall cleaning operation.

9. The control system of a curtain wall cleaning robot according to claim 8, characterized in that, When the curtain wall cleaning robot performs the curtain wall cleaning operation, it also includes autonomous path planning, specifically: After the curtain wall cleaning robot is started, images and data of the curtain wall and the surrounding environment are collected as input data; According to the input data, an environmental map is generated, and then obstacles that may hinder the movement of the curtain wall cleaning robot are identified, and the information of the obstacles is determined; the information of the obstacles includes the position and size; Combined with the obstacle information, an optimal cleaning path is generated through an improved path planning algorithm: S10, combined with the obstacle information, divide the environmental map into grids of equal size, and set the state attribute values of each grid, thereby forming a grid map; S20, according to the grid where the curtain wall cleaning robot is currently located, modify the state attribute values of the covered grids, and combine the improved A* algorithm to obtain the next feasible grid; The heuristic rule of the improved A* algorithm is: where r j represents the next moving grid position, f j represents the attribute value of grid j, where grid j belongs to the neighborhood grid of the current grid i of the robot, c represents the uncovered area coefficient, indicating the influence degree of the side with the least number of uncovered grids on both sides of the original movement direction on the robot's selection of the next grid, G c represents the sum of the number of uncovered grids on the side of the direction of the proposed moving grid and the number of uncovered grids in the original movement direction, G max represents the total number of uncovered grids when the robot is at the current grid, w ij represents the distance between the current grid i and the proposed moving grid j, s j represents the steering function; S30. If the curtain wall cleaning robot gets stuck in a dead zone and there is a grid with a state attribute value of 1 among the next feasible grids, jump to S40. If the curtain wall cleaning robot gets stuck in a dead zone and there is no grid with a state attribute value of 1 among the next feasible grids, jump to S60. Otherwise, loop and execute S20; S40. Calculate the distance between the grid where the curtain wall cleaning robot is currently located and the next feasible grid with a state attribute value of 1, and select the grid with the shortest distance as the temporary search point; S50. Based on the temporary search point and the position of the grid where the curtain wall cleaning robot is currently located, plan the optimal escape route from the dead zone through the improved ant colony algorithm, and jump to S20; S60. End the search and obtain the optimal cleaning path.

10. The control system of a curtain wall cleaning robot according to claim 9, characterized in that, The method for planning the optimal escape route from the dead zone through the improved ant colony algorithm is as follows: When making a state transition, first generate a random number q, where 0 ≤ q ≤ 1, and select the next position according to the following rules: Among them, S represents the selection of the next position through roulette, and τ ij represents the pheromone concentration on the edge, and η ij represents the heuristic information, and y jg represents the Euclidean distance between the allowed transfer position and the target position. α, β, and λ respectively represent τ ij , η ij , and y jg 's relative weights in the ant's path selection. q0 represents the initial position point, t represents time, and allowed k represents the positions allowed for the next transfer of ant k; When moving from i to j, the local pheromone update process is: τ ij (t + 1) = (1 - μ)τ ij (t) + μτ0 where μ represents the pheromone evaporation coefficient during local update, and τ0 represents a small positive constant; The global pheromone update process is: τ ij (t + n) = (1 - ρ)τ ij (t) + ρΔτ ij Among them, ρ represents the pheromone evaporation coefficient during global update, and L ib represents the optimal path of the current iteration.

Images