Well site safety automated patrol method and robot

Abstract

The application discloses a kind of well site safety automatic inspection method and robot.Therein, the method includes: information acquisition module is configured to collect the surrounding environment information of the robot in the well site;Path planning module is configured to select an optimal path for the robot in the well site based on the collected surrounding environment information, and the obstacle on the optimal path is identified and dynamically avoided, to carry out the automatic cruise of the robot.The application solves the technical problem that a large number of manpower is needed for inspection due to the complex environment of the well site.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of robot control and detection technology, and in particular, to a well site safety automatic inspection method and robot. BACKGROUND

[0002] The automatic inspection robot is used for detection and rescue operation in oil well blowout site, can replace the rescue personnel to enter the dangerous accident site with high temperature, high noise, toxic, oxygen deficiency or smoke to carry out detection, fire extinguishing and other operations, can effectively solve the safety threat of rescue personnel in harsh environment, low fire extinguishing efficiency, insufficient data information collection and other problems, has important significance for improving well site detection efficiency, rescue safety and reducing labor consumption.

[0003] At present, the inspection robot is applied to multiple fields, and the product function is also increasingly updated with demand, but in the process of inspection, most of the inspection robots are single sensor for collecting a part of data, which will cause the defects of missing or inaccurate data collection.

[0004] In view of the above problems, no effective solution has been proposed at present. SUMMARY

[0005] The embodiments of the present application provide a well site safety automatic inspection method and robot to at least solve the technical problem that a large amount of manpower is needed for inspection due to the complex environment of the well site.

[0006] According to one aspect of the embodiments of the present application, a well site safety automatic inspection robot is provided, comprising: an information collection module configured to collect surrounding environment information of the robot in the well site; a path planning module configured to select an optimal path for the robot to travel in the well site based on the collected surrounding environment information, and identify and dynamically avoid obstacles on the optimal path to perform automatic cruising of the robot.

[0007] According to another aspect of the embodiments of the present application, a well site safety automatic inspection method is also provided, comprising: collecting surrounding environment information of a robot for well site safety automatic inspection in the well site; selecting an optimal path for the robot to travel in the well site based on the collected surrounding environment information, and identifying and dynamically avoiding obstacles on the optimal path to perform automatic cruising of the robot.

[0008] In the embodiment of the present application, the robot comprises an information collection module and a path planning module. The information collection module is configured to collect surrounding environment information of the robot in the well site; the path planning module is configured to select an optimal path for the robot to travel in the well site based on the collected surrounding environment information, and identify and dynamically avoid obstacles on the optimal path, so as to realize automatic cruising of the robot, and solve the technical problem that a large amount of manpower is required for inspection due to the complex environment of the well site. BRIEF DESCRIPTION OF DRAWINGS

[0009] The accompanying drawings, which are included to provide a further understanding of the present application and constitute a part of this application, illustrate certain illustrative embodiments of the present application and together with the description serve to explain the present application. In the drawings:

[0010] Figure 1 FIG. 1 is a structural schematic diagram of a well site safety automatic inspection robot according to an embodiment of the present application;

[0011] Figure 2 FIG. 2 is a structural schematic diagram of another well site safety automatic inspection robot according to an embodiment of the present application;

[0012] Figure 3 FIG. 3 is a flowchart of fire detection by a fire detection module according to an embodiment of the present application;

[0013] Figure 4 FIG. 4 is a flowchart of a path planning method according to an embodiment of the present application;

[0014] Figure 5 FIG. 5 is a flowchart of a well site safety automatic inspection method according to an embodiment of the present application;

[0015] Figure 6 FIG. 6 is a flowchart of a method for planning a robot travel path according to an embodiment of the present application;

[0016] Figure 7 FIG. 7 is a flowchart of another method for planning a robot travel path according to an embodiment of the present application;

[0017] Figure 8 FIG. 8 is a flowchart of a method for planning a global path by using an improved A* algorithm according to an embodiment of the present application;

[0018] Figure 9 FIG. 9 is a schematic diagram of a 5x5 extended search neighborhood graph according to an embodiment of the present application;

[0019] Figure 10 FIG. 10 is a corresponding relationship diagram of an included angle Y and reserved and discarded directions according to an embodiment of the present application;

[0020] Figure 11This is a flowchart of a method for planning local paths using a dynamic window algorithm according to an embodiment of the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] Example 1

[0024] According to embodiments of the present invention, an automated well site safety inspection robot is provided, such as... Figure 1 As shown, the robot includes an information acquisition module and a path planning module.

[0025] An information acquisition module is configured to acquire information about the robot's surrounding environment at the well site. The information acquisition module includes: a gas sensor configured to detect toxic and flammable gases at the well site; a temperature sensor configured to detect the ambient temperature of the well site; and a high-definition image acquisition device, including a panoramic camera and a binocular vision camera, configured to acquire image data of the well site.

[0026] The path planning module is configured to select an optimal path for the robot to travel in the well site based on the collected surrounding environmental information, and to identify and dynamically avoid obstacles on the optimal path in order to enable the robot to cruise automatically.

[0027] In one example, the path planning module comprises: a GPS positioning system for automatically locating the position of the robot; a global path planning system for planning the starting point and the ending point of the robot cruise, selecting the optimal path; an obstacle identification system comprising a laser radar and a binocular vision camera, using the information collected by the laser radar and the binocular vision camera to determine the pose of the obstacle, to identify the obstacle on the optimal path; a local path planning system for avoiding the identified obstacle.

[0028] In one example, the robot further comprises a fire detection module and a fire extinguishing tracking module. The fire detection module is configured to identify the fire in the well site using a fusion identification method; the fire extinguishing tracking module is configured to control the fire water cannon to track the fire source in real time and take a snapshot of the fire occurrence site for accident analysis when the fire detection module identifies that a fire has occurred.

[0029] In one example, the fire detection module comprises an external thermal imager, a binocular vision camera and a fusion identification system. The external thermal imager is configured to capture the position of temperature anomaly in the well site and determine the suspected fire occurrence point; the binocular vision camera is configured to identify smoke and flame respectively based on deep learning and machine learning methods, extract the depth information of the smoke and / or flame, and obtain the information data of the suspected fire occurrence point; the fusion identification system is configured to fuse the information data of the suspected fire occurrence point, and determine the area with overlapping suspected fire occurrence points as the fire occurrence point.

[0030] In one example, the fire extinguishing tracking module comprises: an infrared thermal imager for obtaining a temperature distribution thermal map of the fire site; a binocular vision camera for collecting depth information of the fire site and obtaining the distance from the area with the highest temperature to the robot; a temperature field reconstruction system for constructing a three-dimensional temperature field based on the temperature distribution thermal map by image processing method; a fire extinguishing real-time tracking system for adjusting the spray angle of the fire water cannon and the position of the robot based on the constructed three-dimensional temperature field to track and extinguish the fire source of the fire site in real time.

[0031] In one example, the robot also includes a communication interruption autonomous control module, configured to autonomously control the robot after communication with the control backend is interrupted, and to travel to a predetermined destination before the communication interruption through path planning. This communication interruption autonomous control module includes: an industrial control computer configured to generate control commands based on the robot's current state when communication is interrupted during movement or firefighting operations; and an autonomous positioning system, including a fiber optic gyroscope and an odometer, used to perform real-time positioning of the robot after the communication interruption, using the real-time positioning as a new starting point to replan the path.

[0032] This application embodiment describes a fire detection system based on the fusion of an infrared thermal imager and a binocular vision camera. It employs detection methods based on location capture of temperature anomalies, machine learning based on flame features, and deep learning based on smoke recognition to detect fires more accurately and effectively, enabling timely and effective fire suppression.

[0033] Example 2

[0034] Figure 2 This is a main structural block diagram of an automated well site safety inspection robot system according to an embodiment of the present invention.

[0035] like Figure 2 As shown, the well site safety automatic inspection robot system in this embodiment of the invention mainly includes: an information acquisition module, a path planning module, a fire detection module, a fire extinguishing tracking module, a communication interruption autonomous control module, and a central control system.

[0036] The information acquisition module includes: a gas sensor, a temperature sensor, and an image acquisition system.

[0037] The gas sensor mainly collects information on toxic and flammable gases in the well site air, such as sulfur dioxide and hydrogen sulfide, and stores the collected gas concentration and type data in the storage module.

[0038] Temperature sensors collect the ambient temperature at the well site and store the data in the storage module.

[0039] The image acquisition system includes a panoramic camera and a binocular vision camera. The panoramic camera is used to collect environmental information around the robot and can rotate 360° for shooting without blind spots. The binocular vision camera is used in the obstacle recognition and fire extinguishing tracking modules described below, and can collect the location information of obstacles and the depth information of the fire location. After acquiring the data, the image acquisition system stores the acquired data in the storage module.

[0040] The data stored in the storage module by the aforementioned information collection module is sent to the back-end terminal through the MESH wireless digital transmission system, and the back-end terminal screen can display the transmitted data information in real time.

[0041] The path planning module includes a positioning system and a global path planning system, wherein the global path planning system includes an obstacle recognition system and a local path planning system.

[0042] The positioning system uses GPS, which can display the robot's coordinates in real time on the back-end terminal. More accurate positioning can be achieved by adding sensors.

[0043] The global path planning system establishes a grid map, determines the starting and ending points, and then plans an optimal path on the grid map to complete the inspection task. When the robot encounters insurmountable obstacles during the inspection process, the obstacle recognition system can collect the location information of both static and dynamic obstacles. The obstacle recognition system includes a binocular vision camera and a LiDAR device, employing multi-sensor information fusion technology. It combines data collected by the LiDAR and binocular camera using fusion algorithms to better enable the robot to identify obstacles at the well site. When the robot travels along the route planned globally, and encounters and identifies insurmountable obstacles, the local path planning system uses obstacle avoidance algorithms to perform a detour around the obstacles.

[0044] Fire detection module: As the robot moves, it detects the surrounding environment and uses a fusion recognition system to identify fires. It utilizes infrared thermal imagers to capture areas of abnormal temperature, machine learning methods based on flame features, and deep learning methods based on smoke recognition to determine suspected fire locations. These suspected fire locations are then cooled until their temperature falls below a certain threshold. A panoramic camera captures images, allowing back-end operators to monitor the situation in real time.

[0045] After identifying suspected fire areas, such as areas A, B, and C, using the three methods described above, the intersecting areas AB, AC, BC, and ABC are determined as the actual fire area. Simultaneously, the coordinates of this area are sent to the central control system. The direction of the line connecting the fire point's coordinates to the robot's current coordinates is the direction the robot needs to travel. The robot's endpoint is determined based on the spray distance of its fire monitor. After determining the starting and ending points and direction, path planning is performed to reach the designated location and extinguish the fire at the identified fire point until the temperature drops below a certain threshold.

[0046] The fire extinguishing and tracking module includes an infrared thermal imager and a binocular vision camera. The infrared thermal imager acquires three-dimensional temperature distribution information of the fire scene and uses image processing methods to reconstruct the three-dimensional temperature field. The binocular camera acquires depth information of the fire scene and establishes a spatial coordinate system. The coordinate values ​​of the three-dimensional fire scene are obtained through three-dimensional reconstruction, and the three-dimensional temperature field is fused with the spatial coordinate system so that the temperature and coordinate values ​​of the three-dimensional temperature field and the three-dimensional fire scene correspond one-to-one with the temperature and coordinate values ​​in the actual fire scene. A crosshair can be set to move arbitrarily in the spatial coordinate system, and the position moved to provides the temperature value and the actual coordinate position in the fire scene.

[0047] The robot prioritizes the area with the highest temperature when extinguishing fires. It moves its crosshair coordinates to the corresponding spatial coordinate point with the highest temperature. Based on the spatial coordinate point, it determines its position in the actual fire scene. The binocular camera obtains the distance between the area with the highest temperature and the robot. Combined with theoretical calculations of the spray angle and spray distance of the fire monitor, the robot's position is adjusted so that the water sprayed by the water monitor can accurately land at the position with the highest temperature in the spatial coordinate system corresponding to the crosshair coordinates in the fire scene.

[0048] The area with the highest temperature in the fire scene changes in real time, and the crosshairs also change in the spatial coordinate system. The spray angle of the fire monitor and the position of the robot are adjusted in time to enable the fire monitor to track the area with the highest temperature in real time. The binocular camera captures images of the fire scene to assist the back-end operators in accident analysis.

[0049] The communication interruption autonomous control module includes an industrial control computer and an autonomous positioning system. The back-end terminal receives and transmits gas, temperature, humidity, and surrounding environmental information collected by the robot's front end. When the back-end terminal cannot see the information collected by the robot's front end information acquisition module or cannot control the robot, the robot does not receive feedback signals, the GPS positioning system stops working, the robot switches to autonomous control mode, and the industrial control computer immediately sends a command to the central control system to stop all robot operations.

[0050] The autonomous positioning system includes an odometer and a fiber optic gyroscope. When communication is good, data from the odometer and gyroscope are collected in real time. When communication is interrupted, the system calculates the robot's position based on the odometer data, introduces a single-axis fiber optic gyroscope to correct axial angle errors, and the industrial control computer obtains the robot's pose parameters based on the data collected by the odometer and gyroscope. Based on the robot's state before the communication interruption, control commands are generated and sent to the central control system to complete the remaining inspection tasks before the communication interruption.

[0051] The automatic well site safety inspection robot provided in this embodiment of the invention enables the automatic inspection robot to replace emergency personnel in entering well sites with harsh environments such as high temperature and high noise to carry out operations such as detection and fire extinguishing, making the inspection robot more autonomous, efficient and intelligent.

[0052] Example 3

[0053] This invention provides an automated well site safety inspection robot. This robot is capable of detecting fires, autonomously locating its location, and autonomously extinguishing fires.

[0054] Detecting fires: such as Figure 3 As shown, a fusion recognition system is used to identify fires. An infrared thermal imager acquires temperature information at the well site, capturing locations with abnormally high temperatures and defining them as suspected fire points. A binocular camera acquires images of the surrounding area, preprocesses the images, and uses a machine learning method based on flame features to determine suspected fire points. Since a large amount of smoke is generated during a fire, the acquired images are preprocessed, and a deep learning method based on smoke recognition is used to determine suspected fire points. The coordinate information of the suspected fire points determined by the above three methods is acquired: coordinates determined by temperature information are defined as region A, coordinates determined by flame features are defined as region B, and coordinates determined by smoke recognition are defined as region C. The intersections of regions A, B, and C are extracted and defined as fire points AB, AC, ABC, and BC. The intersection area can be identified as the location of the fire. The coordinates of the intersection area are converted into coordinate values ​​and sent to the central control system. The direction of the line connecting the coordinates of the fire point and the robot's current coordinates is the direction the robot needs to move. The robot's endpoint is determined based on the spray distance of the robot's fire monitor, and path planning is performed to reach the designated location for firefighting operations.

[0055] Autonomous Localization: When communication is good, the robot cruises according to the path planning module, collecting data from the odometer and fiber optic gyroscope in real time. When communication is interrupted, it calculates its position based on the odometer data, using a single-axis fiber optic gyroscope to correct steering angle errors, thereby obtaining the robot's pose parameters. The previously set endpoint when communication is good is used as the new path endpoint, and the robot's position when communication is interrupted is set as the starting point. The industrial control computer generates control commands based on the robot's pose parameters and sends them to the central control system to complete the new path planning.

[0056] Firefighting tracking: Infrared thermal imagers determine temperature by receiving the radiant energy of objects, store the data in the form of a color index table, and represent temperature distribution information in the form of images. The three-dimensional temperature field is reconstructed through image processing technology. A binocular camera acquires depth information from the actual fire scene, establishing a spatial coordinate system. 3D reconstruction yields the coordinates of the 3D fire scene, and the 3D temperature field is fused with the spatial coordinate system, ensuring a one-to-one correspondence between the temperature in the 3D temperature field, the coordinates in the spatial coordinate system, and the actual temperature and coordinates in the fire scene. A crosshair is set to move freely within the spatial coordinate system, displaying the temperature value and actual fire scene coordinates at the corresponding location. The robot prioritizes the area with the highest average temperature, moving the crosshair to the corresponding point with the highest temperature in the spatial coordinate system, and determining its position within the fire scene based on this point. The binocular camera acquires the distance between the area with the highest temperature and the robot. Combined with theoretical calculations of the fire monitor's spray angle and distance, the robot's distance from the area with the highest temperature is adjusted, ensuring the water sprayed by the monitor accurately lands at the location with the highest temperature corresponding to the crosshair in the spatial coordinate system. The area with the highest temperature in the fire scene changes in real time; through the above operations, the fire monitor's spray angle and the robot's position are adjusted accordingly, enabling the fire monitor to track the area with the highest temperature in real time.

[0057] The automated well site safety inspection robot in this embodiment includes a central control system, an information acquisition module, a path planning module, a fire detection module, a fire extinguishing tracking module, and a communication interruption autonomous control module.

[0058] The central control system is used to control the above four modules and execute relevant operation instructions.

[0059] The information acquisition module is used to collect environmental information around the robot and feed the information back to the backend terminal in the form of images and numbers. The information acquisition module includes: a gas sensor, a temperature sensor, and a high-definition image acquisition system. The gas sensor detects toxic and flammable gases at the well site; the temperature sensor detects the ambient temperature of the well site; the high-definition image acquisition system consists of a panoramic camera and a binocular vision camera.

[0060] The information acquisition module transmits the collected information to the back-end terminal via a wireless transmission system. This wireless transmission system uses a MESH wireless digital transmission system. The back-end terminal receives the information transmitted by the robot's MESH wireless digital transmission system and can send control commands to the robot's front end.

[0061] The path planning module is used to enable the robot to select an optimal path to complete its patrol during its movement in the well site, and to identify and dynamically avoid obstacles along the path. The path planning module includes: a GPS positioning system, a global path planning system, an obstacle recognition system, and a local path planning system.

[0062] The GPS positioning system is used for the automatic patrol robot's location positioning; the global path planning system is used for planning the robot's starting and ending points for patrol and selecting an optimal path to complete the patrol task; the obstacle recognition system includes a LiDAR and a binocular camera device, which fuses information collected by the LiDAR and the binocular vision camera to determine the pose of obstacles, and is used to identify obstacles on the globally planned route. The local path planning system is used to perform obstacle avoidance processing on the identified obstacles.

[0063] A fire detection module is used by the robot to detect fire locations in the well site environment. The fire detection module includes: an infrared thermal imager, a binocular vision camera, and a fusion recognition system. The infrared thermal imager is used to capture locations of abnormal temperatures to identify suspected fire locations; the binocular vision camera uses deep learning and machine learning methods to identify dense smoke and flames, extracting depth information of the smoke and flames to obtain suspected fire locations; the fusion recognition system fuses the information data of suspected fire locations, and identifies the actual fire location by identifying overlapping areas between different suspected fire locations.

[0064] The fire extinguishing tracking module is used to track the fire source in real time after a fire is detected by the fire monitor, and to capture images of the fire location to assist back-end operators in accident analysis and improve the efficiency of fire extinguishing operations.

[0065] The fire extinguishing tracking module includes: an infrared thermal imager, a binocular vision camera, a temperature field reconstruction system, and a real-time fire extinguishing tracking system. The infrared thermal imager acquires a thermal map of the temperature distribution at the fire location; the binocular vision camera collects depth information of the actual fire scene, determining the distance of the hottest area from the robot, thus preparing the robot for real-time fire extinguishing operations; the temperature field reconstruction system constructs a three-dimensional temperature field from the thermal map using image processing methods; and the real-time fire extinguishing tracking system adjusts the spray angle of the fire monitor and the robot's position based on the depth information collected by the binocular camera and the information from the three-dimensional temperature field reconstructed by the infrared thermal imager to track and extinguish the fire source in real time.

[0066] Specifically, the fire extinguishing tracking module uses information from the three-dimensional temperature field constructed by the infrared thermal imager and coordinate information from the three-dimensional fire scene reconstruction based on the binocular camera to prioritize precise cooling of the areas with the highest temperature. It also adjusts the spray angle of the fire monitor and the distance between the fire monitor and the fire point in real time based on the changing temperature field information and the coordinate information extracted by the binocular camera, thereby achieving real-time tracking of the fire monitor.

[0067] The communication interruption autonomous control module is used to perform autonomous control after the communication between the robot's front end and the control back end is interrupted, and to travel to the destination set before the communication interruption through path planning.

[0068] The communication interruption autonomous control module includes an industrial control computer and an autonomous positioning system. The industrial control computer executes the program after a communication signal interruption. When the robot loses communication with the handheld device during movement or firefighting operations (i.e., the backend terminal cannot see the video transmitted from the robot's front end or manually control the robot), it immediately sends a command to the robot to stop working and generates control commands based on the robot's current state, sending them to the robot's central control system to control the robot to continue its patrol mission. Once the communication signal is restored, it stops generating control commands. The autonomous positioning system includes a fiber optic gyroscope and an odometer. After a communication interruption, the robot uses the gyroscope and odometer for real-time positioning. The location of the communication interruption is set as the new starting point, and the pre-set endpoint from the path planning module is used as the endpoint for replanning the path. For example, the autonomous positioning system can enable the robot to accurately reposition itself after a communication interruption or when GPS positioning is lost. The odometer calculates the trajectory for positioning, and the fiber optic gyroscope corrects for steering angle errors.

[0069] The central control system receives control commands from the industrial control computer and completes the cruise missions that were not completed before the communication interruption.

[0070] The robot provided in this embodiment can automatically inspect well sites. Its multi-sensor system can collect environmental information, detect suspected fires, capture images of locations requiring firefighting operations, and track and extinguish fires. Furthermore, it can switch modes for autonomous control after communication interruptions. This application meets the inspection needs of complex well site environments and reduces manual labor.

[0071] Example 4

[0072] This invention provides an automated well site safety inspection robot. The robot's structure is similar to that of the robot described in the previous embodiments, with the main difference being the path planning module. Therefore, this embodiment will focus on describing the robot's path planning module, while other components or modules of the robot will not be described in detail.

[0073] The path planning module in this embodiment implements the following path planning methods, such as... Figure 4 As shown, this path planning method includes the following steps:

[0074] Step S402: Establish an environmental model of the well site.

[0075] In this embodiment, the grid mapping method is mainly used to construct the environment model. The grid method divides the robot's safety inspection area into multiple grid units containing binary information. For example, the grid method is used to divide the robot's safety inspection area into multiple uniform grids containing binary information, and a grid model is obtained.

[0076] In one example, the grid model is

[0077] M (i，j) ={(x a y b )|i×m≤a≤(i+1)×m, j×n≤b≤(j+1)×n}

[0078] Among them, M (i，j) Represents a raster cell, where i represents the x-coordinate and j represents the y-coordinate; (x a y b ) represents the pixel coordinates of the original image, a represents the x-coordinate of the original image, and b represents the y-coordinate of the original image. Each grid cell is typically represented by a value of 0 or 1 to indicate environmental information, i.e., 0 means there are no obstacles in the environment and the robot can move freely in the area; 1 means there are obstacles in the environment and the robot cannot pass through the area.

[0079] The least squares method is used to fit the plane, that is, to obtain the optimal fit by minimizing the sum of squared errors. Assume the expression for the fitted plane equation is:

[0080] z = a0x + a1y + a2

[0081] Where x, y, and z represent the x-axis, y-axis, and z-axis coordinates, respectively, and a0, a1, and a2 are the parameters of the equation, which can be obtained by the following formula.

[0082] Therefore, it is necessary to make Minimum, where n represents the number of choices, and s is the sum of the equations.

[0083] Therefore, the conditions that need to be met are:

[0084]

[0085] Therefore:

[0086]

[0087] Right now

[0088]

[0089] Where, x i y i , z i These represent the horizontal coordinate, the axis coordinate, and the vertical coordinate, respectively.

[0090] Ultimately, we can obtain:

[0091]

[0092] in, Z = [z1 z2 … z] n ] T

[0093] Step S404, global path planning.

[0094] Based on a pre-established environment model, a path planning algorithm is used to perform global path planning for the robot's movement.

[0095] When performing global path planning, it is necessary to consider the slope, roughness, and undulation of the route. Therefore, based on the grid and the robot's motion capabilities, it is necessary to establish the robot's slope cost function, roughness cost function, and undulation cost function.

[0096] In one example, the slope cost function is established as follows: The least squares method is used to fit the plane, obtaining the fitted plane equation. Based on the plane equation, the slope of the grid is calculated, and the slope cost function is established based on the grid slope. In another example, the roughness cost function is established as follows: The distance from all elevation points within the grid to the fitted plane is calculated. Based on the distance, the roughness of the grid is calculated, and the roughness cost function is established based on the roughness. In yet another example, the undulation cost function is established as follows: The standard deviation of the elevation of all discrete points within the grid is calculated; the undulation cost function is established based on the standard deviation of the elevation.

[0097] Specifically, in one example, the slope of grid e:

[0098]

[0099] To prevent the robot from being unable to pass through the area due to excessive grid slope, the slope of the robot's traversable area is limited based on the robot's climbing ability, thus establishing the following slope cost function:

[0100]

[0101] Where k1 is a constant value, which can be learned from a neural network, and θ maxThis indicates the robot's maximum climbing ability.

[0102] The roughness of the terrain affects a robot's mobility. Excessive roughness can cause the robot to have difficulty moving, or even become stuck and unable to move. Roughness R is described by the average deviation of the fitted surface.

[0103] The formula for calculating roughness is as follows:

[0104]

[0105] in, The distance from each elevation point to the fitted plane is denoted by , and n represents the number of selected coordinate points.

[0106] Similarly, to prevent the robot from being unable to pass through the area due to excessive grid roughness, the roughness of the robot's traversable area is limited based on the robot's driving capability, thus establishing a roughness cost function:

[0107]

[0108] Where k2 is a constant value, which can be determined according to actual needs or empirical knowledge, and R max This indicates the maximum driving capability of the robot.

[0109] The undulation of the terrain affects a robot's climbing ability; excessive undulation can cause tipping over or collisions. The undulation degree H is described by the standard deviation of the elevations of all discrete points within the corresponding grid. This method is more objective and realistic than simply using the difference between the highest and lowest elevation points.

[0110]

[0111] in, It is the arithmetic mean of all elevation values ​​within grid e.

[0112] Similarly, to prevent the robot from being unable to pass through the area due to excessive grid undulation, the undulation of the robot's drivable area is limited based on the robot's obstacle-crossing ability, thus establishing an undulation cost function.

[0113]

[0114] Where k3 is a constant value that can be determined based on actual needs or experience, and h represents the height of the robot.

[0115] Based on these established cost functions, path planning is performed. The specific planning method is as follows:

[0116] First, each search position is evaluated to obtain the best position, and then the search continues from this position until the target is reached. This eliminates a large number of unnecessary search paths, improving efficiency. In heuristic search, position estimation is crucial. In this embodiment, the estimation function is improved by adjusting the weighting coefficients of the actual cost and the estimated cost. Then, based on the improved estimation function, global path planning is performed for the robot's movement.

[0117] In a preferred embodiment, the evaluation function is:

[0118] F(n)=V(n)G(n)α+W(n)H(n)B

[0119] In the formula, V(n) represents the weight coefficient of the actual cost, and W(n) represents the weight coefficient of the estimated cost H(n). The improved algorithm adds weight coefficients V(n) and W(n), and adds modulation factor α and correction factor β obtained through the neural network. By adjusting the coefficients of the weight factors, the influence on the algorithm during path search can be changed. The larger W(n) is, the closer it is to the BFS algorithm; the larger V(n) is, the closer it is to the Dijkstra algorithm.

[0120] The robot's starting position is added to the open list OL. Based on the improved evaluation function, path planning is performed iteratively. It is determined whether there are expandable points in the planned path and whether the target point has not been found. If the result is yes, the optimal candidate in OL is added to the closed list CL and path backtracking is performed. Otherwise, path backtracking is performed directly to determine the optimal global planning path.

[0121] Step S406, obstacle avoidance.

[0122] Typically, based on the coordinates of the obstacle edge points returned by the robot, a safety distance is added to the obstacle's size, thereby changing the grid cell size to m×n and the number of grid cells. After the raster cells are changed, the path of the local region is recalculated.

[0123] Step S408, corner processing.

[0124] The route planned by the improved algorithm is only the mathematically optimal path. For the actual operation of the robot, the ultimate goal is to obtain the optimal route for the robot. A 90-degree turn is very detrimental to actual operation, so it is necessary to further optimize the route and turn angles to minimize the number of turns without increasing the path cost.

[0125] In one example, within the global path, the index values ​​of the expansion point and the target point are obtained. Based on these index values, corner optimization is performed. For instance, the index value of the parent node of the expansion point is calculated using the position information stored in the domain pointer. It is then determined whether the index value of the parent node is equal to the starting point. If it is, it indicates that the point is an expansion from the starting point, and optimization is skipped. Otherwise, the index value of the parent node is used to check the position information of the parent node in the cell array domain pointer, and the index value of the expected next point is calculated using this position information.

[0126] In another example, the steps of the corner optimization algorithm are as follows:

[0127] First, calculate the position information stored in the pointer based on the point that was originally to be expanded. If the position information stored in the pointer is L, it means that the current expansion point is expanded from the point on the left. That is, the index value of its parent node is the index value of the point minus n (n is the length of each row or column). Similarly, the processing methods for the other three position information can be obtained.

[0128] Secondly, before calculating the index value of the expected next point, it is determined whether the index value of the parent node calculated in the previous step is equal to the starting point. If it is equal to the starting point, it means that this point is an extension of the starting point, and the direction is a straight line, so the optimization is skipped; otherwise, proceed to the next step.

[0129] Finally, using the index value of the parent node obtained in the first step, we can check the position information of the parent node in the cell array pointer, and use this position information to calculate the index value of the point to be moved next.

[0130] This application proposes a global path planning scheme based on an improved algorithm. A weight ratio W(n) / V(n) is set to influence the objective function F(n), thereby finding the optimal weight ratio, reducing runtime, and improving the algorithm's computational efficiency. Furthermore, local path corner optimization effectively reduces the number of turns the robot makes during operation, ensuring the robot's stability during well site operation.

[0131] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0132] Example 5

[0133] According to embodiments of the present invention, an automatic well site safety inspection method is also provided, such as... Figure 5 As shown, the method includes:

[0134] Step S502: Collect information about the surrounding environment of the robot used for automatic inspection of well site safety in the well site.

[0135] Step S504: Based on the collected surrounding environment information, select an optimal path for the robot to travel in the well site, and identify and dynamically avoid obstacles on the optimal path to enable the robot to cruise automatically.

[0136] The method in this embodiment can achieve all the functions described in the above device embodiments, therefore, it will not be repeated here.

[0137] Example 6

[0138] This invention provides an automated well site safety inspection robot. The robot's structure is similar to that of the robot described in the previous embodiments, with the main difference being the path planning module. Therefore, this embodiment will focus on describing the robot's path planning module, while other components or modules of the robot will not be described in detail.

[0139] The path planning module in this embodiment implements the following path planning methods, such as... Figure 6 As shown, the method includes:

[0140] Step S602: Obtain the robot's starting position and target position. Based on the starting position and target position, use the A* algorithm to plan the robot's global path.

[0141] For example, the search neighborhood and search direction of the A* algorithm are expanded; the evaluation function of the A* algorithm is determined based on the actual cost from the starting position to the target position, the cost function from the starting position to the target position, and the weight coefficients; and the global path of the robot is planned based on the expanded search neighborhood and search direction, the evaluation function, and the pre-acquired static obstacle information.

[0142] In some examples, the weighting coefficient can be set by the following steps: determining the robot's current position; setting a larger weighting coefficient as the current position is farther from the target position; and setting a smaller weighting coefficient as the current position is closer to the target position.

[0143] In other examples, the weighting coefficients can be set by the following steps: determining the distance between the robot's current position and the target position; and determining the weighting coefficients based on the estimated cost between the current position and the target position and the actual cost from the current position to the target position.

[0144] After planning the robot's global path, you can traverse the nodes in the global path and delete collinear nodes. In this case, the node before and after the current node is collinear if the previous node and the next node are in the same search direction. You can also traverse the nodes in the global path and delete redundant inflection points.

[0145] Step S604: The robot acquires real-time information on dynamic obstacles around it during its movement, and plans the robot's local path using a dynamic window algorithm based on the dynamic obstacle information.

[0146] First, the robot's speed range is determined based on its speed constraints. For example, the robot's maximum and minimum speed constraints are determined based on its minimum and maximum linear velocities, as well as its minimum and maximum angular velocities; the robot's motor acceleration and deceleration constraints are determined based on its maximum linear acceleration and deceleration, maximum angular acceleration and deceleration, and its current linear and angular velocities; and the robot's safety constraints are determined based on its maximum safe deceleration distance.

[0147] Next, the robot's speed range is determined based on the maximum and minimum speed constraints, motor acceleration and deceleration constraints, and safety constraints.

[0148] Subsequently, the evaluation function of the dynamic window algorithm is optimized based on dynamic obstacle information and speed range. For example, the angle difference between the endpoint orientation and the target orientation of the robot's predicted trajectory, as well as the distance from the endpoint of the predicted trajectory to the nearest obstacle, are determined. Here, the endpoint orientation is the direction of the robot's predicted trajectory endpoint at the current speed, and the target orientation is the orientation of the target location. The evaluation function of the dynamic window algorithm is optimized based on the angle difference between the endpoint orientation and the target orientation, the distance from the endpoint of the predicted trajectory to the nearest obstacle, and the current speed.

[0149] Finally, the robot's local path is planned based on the optimized evaluation function.

[0150] Step S606: Merge the global path and the local path to determine the robot's travel path.

[0151] This embodiment provides a robot global path planning method based on the improved A* algorithm and dynamic window method. By optimizing the evaluation function and search direction of the A* algorithm and adopting a redundant point removal strategy, it improves path search efficiency and path smoothness. Furthermore, this embodiment also incorporates the dynamic window method, achieving real-time dynamic obstacle avoidance while ensuring global path optimization.

[0152] Example 7

[0153] This invention provides an automated well site safety inspection robot. The robot's structure is similar to that of the robot described in the previous embodiments, with the main difference being the path planning module. Therefore, this embodiment will focus on describing the robot's path planning module, while other components or modules of the robot will not be described in detail.

[0154] The path planning module in this embodiment implements the following robot global path planning method based on the fusion of the improved A* algorithm and the dynamic window method, such as... Figure 7 As shown, the method includes:

[0155] Step S702: Plan the global path of the robot using the improved A* algorithm.

[0156] The A* algorithm is essentially a heuristic search algorithm that can perform global path calculations. The evaluation function for the A* algorithm is:

[0157] F(n) = G(n) + H(n)

[0158] The traditional A* algorithm has the following problems: 1) It has fewer search directions and more search nodes, which reduces the search efficiency of the algorithm; 2) The global path planned by the A* algorithm has redundant collinear nodes and redundant inflection points, which affects the continuity of the robot's movement path.

[0159] Therefore, this embodiment improves the A* algorithm, and the specific improvement method is as follows: Figure 8 As shown, the procedure includes the following steps S7022 to S7026.

[0160] Step S7022: Determine the search neighborhood and direction.

[0161] First, the 3×3 search neighborhood is optimized to a 5×5 search neighborhood, and the number of search directions increases from 8 to 16. Figure 9 As shown, the outermost black grid represents the nodes to be deleted. To further improve computational efficiency, the 16 search directions are evaluated. Based on the positional relationship between the current node and the target node, 9 search directions are retained, and 7 are discarded. Let γ be the angle between the line connecting the current node and the target node and the n5 direction. The correspondence between angle γ and the 9 retained directions and the 7 discarded directions is as follows:Figure 10 As shown.

[0162] The traditional A* algorithm uses a 3×3 search neighborhood with 8 search directions. The limited search neighborhood and directions result in low search efficiency, and the path planning trajectory lacks smoothness due to numerous turning points. This embodiment, through optimization of the search neighborhood and search directions, improves both the smoothness of the path trajectory and computational efficiency.

[0163] Step S7024: Optimize the heuristic function.

[0164] The appropriate selection of heuristic functions plays a crucial role in global path planning. Ideally, the actual cost between the current node and the target node should be estimated in one step, but this is difficult to achieve. During path search, when the heuristic function's estimate is smaller than the actual cost, it leads to more redundant nodes and low computational efficiency; when the heuristic function's estimate is larger than the actual cost, redundant nodes are reduced and computational efficiency is high, but the planned path is not optimal.

[0165] In this embodiment, the heuristic function H(n) is determined based on the Euclidean distance between the starting node and the target node and the weight coefficients. The specific formula for the heuristic function is as follows:

[0166] F(n) = G(n) + ε*H(n)

[0167]

[0168] Where F(n) is the evaluation function from the starting node to the target node, G(n) is the actual cost from the starting node to the current node, ε is the weight coefficient of H(n), and R is the distance between the starting node and the target node.

[0169] In this embodiment, the value of H(n) is always less than the actual distance between the current node and the target node. When the current node is far from the target node, the estimated value of H(n) is less than the actual value, and there are many nodes to search. In this case, the value of ε is increased to improve the computational efficiency. When the current node is close to the target node, the estimated value of H(n) gradually approaches the actual value, but when the estimated value is greater than the actual value, the optimal solution of the path cannot be obtained. In this case, the value of ε is decreased to obtain the optimal path.

[0170] The existing A* algorithm searches for redundant nodes, affecting search efficiency, mainly due to the design of the heuristic function H(n). In this embodiment, a weight coefficient ε is introduced when designing the heuristic function. By adjusting the value of ε, computational efficiency can be improved and the optimal path can be obtained.

[0171] Step S7026: Delete redundant nodes.

[0172] Although the search efficiency is improved after the above steps S7022 and S7026, there are still redundant nodes in the search path, which affects the robot's following ability. Therefore, this embodiment introduces a redundant point deletion strategy to delete some redundant nodes and retain only the necessary inflection points.

[0173] First, traverse the nodes in the search path and delete nodes that are collinear. If the current node and the node following it are in the same search direction, delete the following node. Figure 9 The black grid around the center represents the nodes that need to be deleted.

[0174] Next, remove redundant inflection points. Let the nodes in the route be {X}. k |k=1,2,…n},Connect points X1 and X k Point, if X1X k If the distance to the obstacle is less than the set safety distance, then connect X1X. K-1 Delete nodes from 1 to K-1, and then repeat the above steps starting from point X2 until all nodes in the segment have been traversed.

[0175] Step S704: Plan the local path of the robot using a dynamic window algorithm.

[0176] The implementation process of the dynamic window algorithm in this embodiment mainly includes three parts: velocity sampling, trajectory prediction, and trajectory evaluation. The robot's speed is within a certain speed range, within which many path trajectories can be generated. This embodiment uses the dynamic window algorithm to select the optimal path.

[0177] Figure 11 This is a method for planning the local path of the robot using a dynamic window algorithm according to an embodiment of this application, such as... Figure 11 As shown, the method includes:

[0178] Step S7042: Establish the motion model of the robot.

[0179] Establish a motion model for the robot to simulate its trajectory. The robot's initial pose is (x0, y0, θ0), and its velocity is (ν0, ω0). Assuming the robot moves at a constant velocity (ν′, ω′) over the next time interval Δt, the dynamic window model of the robot is:

[0180]

[0181] Among them, (x n y n θ m Let θ represent the robot's pose at time n, v' represent the linear velocity at the next time step, ω' represent the angular velocity at the next time step, and θ represent the position of the robot at time n. nThis represents the angle between the robot's direction of motion and the horizontal direction after an interval of n, where n represents time n.

[0182] Step S7044: Perform velocity sampling.

[0183] The velocity space contains multiple sets of linear and angular velocities. Due to certain constraints, the range of velocity generation for the robot is limited. These constraints are as follows:

[0184] Maximum and minimum speed constraints for the robot:

[0185] V m ={(v, ω)|v∈[v min v max ]∩ω∈[ω min ω max ]}

[0186] Among them, v min v max ω represents the robot's minimum and maximum linear velocities. min ω max Let vm represent the robot's minimum and maximum angular velocities, v represent the robot's maximum and minimum velocity constraints, v represent the robot's linear velocity, and ω represent the robot's angular velocity.

[0187] Robot motor acceleration / deceleration constraints:

[0188]

[0189] Among them, v c ω c These are the robot's current linear velocity and angular velocity; The maximum linear acceleration and deceleration of the robot; Let vd represent the robot's maximum angular acceleration and deceleration, and vd represent the robot motor acceleration and deceleration speed constraints.

[0190] Robot safety constraints: The following speeds are required for the robot to safely brake before colliding with an obstacle:

[0191]

[0192] Among them, v a This indicates the safety constraints for the robot.

[0193] In summary, the robot's speed range should meet the following conditions:

[0194] ν r =ν m ∩v d ∩ν a

[0195] Among them, vr This indicates the speed constraint range for the robot.

[0196] Step S7046: Optimize the evaluation function.

[0197] To overcome the tendency of the dynamic window method to get trapped in local optima, this embodiment optimizes the evaluation function. The optimized evaluation function is:

[0198] G(ν,ω)=σ[α·head(v,ω)+β·dist(v,ω)+γ·νel(ν,ω)]

[0199] Where head(v, ω) represents the angle difference between the predicted trajectory endpoint and the target trajectory; dist(v, ω) represents the distance from the trajectory to the nearest obstacle; vel(v, ω) represents the evaluation function of the current velocity magnitude; σ is the smoothing function; α, β, and γ are the weights of the above three terms head(v, ω), dist(v, ω), and vel(v, ω).

[0200] Finally, the optimized evaluation function is used to plan local paths.

[0201] Step S706: Merge the global path and the local path to determine the robot's travel path.

[0202] This application expands the search neighborhood of the A* algorithm to 5×5, optimizes the number of path search directions from 16 to 9, optimizes the heuristic function, introduces a redundant point deletion method, improves the search efficiency of the algorithm, and optimizes the search path of the algorithm; it achieves local path planning through the dynamic window method, overcoming the defect that the robot cannot dynamically avoid obstacles in global path planning.

[0203] Example 8

[0204] Embodiments of the present invention also provide a storage medium. The storage medium is configured to store program code for performing the above methods. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0205] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An automated well site safety inspection robot, characterized in that, include: The information acquisition module is configured to acquire information about the robot's surrounding environment in the well site; The path planning module is configured to select an optimal path for the robot to travel in the well site based on the collected surrounding environmental information, and to identify and dynamically avoid obstacles on the optimal path in order to enable the robot to cruise automatically. The path planning module is further configured as follows: The search neighborhood and search direction of the A* algorithm are expanded; the evaluation function of the A* algorithm is determined based on the actual cost from the starting position to the target position, the cost function from the starting position to the target position, and the weight coefficients; the global path of the robot is planned based on the expanded search neighborhood and search direction, the evaluation function, and the pre-acquired static obstacle information. The robot acquires real-time information on dynamic obstacles around it during its movement, and plans the robot's local path using a dynamic window algorithm based on this information. By integrating global and local paths, the robot's travel path is determined; The cost function is determined by at least one of the following: The least squares method is used to fit the plane to obtain the fitted plane equation. Based on the plane equation, the slope of the grid is calculated, and based on the slope of the grid, the slope cost function is established. Calculate the distance from all elevation points within the grid to the fitted plane, calculate the roughness of the grid based on the distance, and establish a roughness cost function based on the roughness. Calculate the standard deviation of elevation for all discrete points within the grid, and establish an undulation cost function based on the standard deviation of elevation. After determining the cost function, the robot is further configured to: optimize the cost function based on a heuristic function, wherein the value of the heuristic function is always less than the actual distance between the current node and the target node, and when the current node is far from the target node, the value of the weight factor of the heuristic function is increased; when the current node is close to the target node, the value of the weight factor of the heuristic function is decreased.

2. The robot according to claim 1, characterized in that, Also includes: The fire detection module is configured to identify fires at the well site using a fusion recognition method; The fire extinguishing tracking module is configured to control the fire monitor to track the fire source in real time and capture images of the fire location when the fire detection module detects a fire, for use in accident analysis.

3. The robot according to claim 1, characterized in that, Also includes: The communication interruption autonomous control module is configured to perform autonomous control after the robot loses communication with the control backend, and to travel to the destination set before the communication interruption through path planning.

4. The robot according to claim 1, characterized in that, The information collection module includes: A gas sensor is configured to detect toxic and flammable gases at the well site; A temperature sensor is configured to detect the ambient temperature of the well site; A high-definition image acquisition device, including a panoramic camera and a binocular vision camera, is configured to acquire image data of the well site.

5. The robot according to claim 2, characterized in that, The fire detection module includes: An external thermal imager is configured to capture locations of temperature anomalies in the well site and identify suspected fire locations; The binocular vision camera is configured to: identify dense smoke and flames respectively based on deep learning and machine learning methods, extract depth information of dense smoke and / or flames, and obtain information data of the suspected fire location; The fusion identification system is configured to fuse information data of the suspected fire locations and identify overlapping areas of different suspected fire locations as the fire locations.

6. The robot according to claim 2, characterized in that, The fire extinguishing tracking module includes: Infrared thermal imagers are used to acquire thermal maps of temperature distribution at fire sites. A binocular vision camera is used to collect depth information of the fire scene and obtain the distance between the area with the highest temperature and the robot. A temperature field reconstruction system is used to construct a three-dimensional temperature field from the temperature distribution heatmap using image processing methods. The real-time fire extinguishing tracking system, based on the constructed three-dimensional temperature field, adjusts the spray angle of the fire monitor and the position of the robot to track and extinguish the fire source in the fire scene in real time.

7. The robot according to claim 3, characterized in that, The communication interruption autonomous control module includes: The industrial control computer is configured to generate control commands based on the robot's current state when communication is interrupted during the robot's movement or firefighting operations. An autonomous positioning system, including a fiber optic gyroscope and an odometer, is used to locate the robot in real time after a communication interruption, and to use the real-time location as a new starting point to replan the path.

8. A method for automatic inspection of well site safety, characterized in that, include: The robot used for automated safety inspection of the well site collects information about the surrounding environment in the well site; Based on the collected surrounding environment information, an optimal path is selected for the robot to travel in the well site, and obstacles on the optimal path are identified and dynamically avoided to enable the robot to cruise automatically. The process includes selecting an optimal path for the robot to travel in the well site, identifying and dynamically avoiding obstacles along the optimal path to enable the robot's automatic navigation, including: The search neighborhood and search direction of the A* algorithm are expanded; the evaluation function of the A* algorithm is determined based on the actual cost from the starting position to the target position, the cost function from the starting position to the target position, and the weight coefficients; the global path of the robot is planned based on the expanded search neighborhood and search direction, the evaluation function, and the pre-acquired static obstacle information. The robot acquires real-time information on dynamic obstacles around it during its movement, and plans the robot's local path using a dynamic window algorithm based on this information. By integrating global and local paths, the robot's travel path is determined; The cost function is determined by at least one of the following: The least squares method is used to fit the plane to obtain the fitted plane equation. Based on the plane equation, the slope of the grid is calculated, and based on the slope of the grid, the slope cost function is established. Calculate the distance from all elevation points within the grid to the fitted plane, calculate the roughness of the grid based on the distance, and establish a roughness cost function based on the roughness. Calculate the standard deviation of elevation for all discrete points within the grid, and establish an undulation cost function based on the standard deviation of elevation. After determining the cost function, the method further includes: optimizing the cost function based on a heuristic function, wherein the value of the heuristic function is always less than the actual distance between the current node and the target node, and when the current node is far from the target node, the value of the weight factor of the heuristic function is increased; when the current node is close to the target node, the value of the weight factor of the heuristic function is decreased.

9. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed, the computer performs the method as described in claim 8.

Images