Robot inspection data management system and method based on machine learning

Abstract

The present invention discloses a robot inspection data management system and method based on machine learning, which belongs to the field of machine learning technology. It includes: planning the robot's moving route according to the checkpoints, and obtaining the robot's inspection range during the movement; monitoring and acquiring the moving object, judging whether the moving object enters the inspection range, and if it enters the inspection range, obtaining the position, moving direction and moving speed of the moving object and the robot when they just enter the inspection range, obtaining the robot detection area, and judging whether there are calibrated corner points in the robot detection area, and executing different action instructions to the robot according to the position of the calibrated corner points and the moving speed of the moving object. The present invention reasonably plans the robot's operating instructions to slow down the robot, stop the robot or optimize the inspection route, and realize intelligent inspection of moving objects such as personnel to prevent personnel from being frightened or injured due to collision with the robot.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning, and in particular to a robot inspection data management system and method based on machine learning. Background Art

[0002] With the development of science and technology, there are many advantages to using robots to inspect cable tunnels. On the one hand, robots can complete inspection tasks more efficiently and accurately, transmit data in real time, and facilitate timely detection of problems; on the other hand, they can replace manual work in harsh environments and reduce personnel risks. Robots can adapt to various special situations during the inspection of cable tunnels, including: smoothly passing through narrow places such as fire doors and drainage pipes, actively bypassing obstacles such as fire boxes, and smoothly passing through the leakage net of the sewer well, etc., which brings great convenience and safety to the construction.

[0003] However, when the robot performs inspection tasks, it needs to follow instructions to travel in the cable tunnel. During the driving process, there will inevitably be moving obstacles such as moving workers, carts and other robots around it. If the robot is small or the cable tunnel is dark, it will be difficult for workers to find the robot. In addition, for working workers and other groups whose attention is not on the robot, if the robot cannot avoid them in time during the working process, it will cause the robot to collide with workers, carts or goods, causing the workers to be frightened, or have minor pain or abrasions, or even cause sprains, fractures, and damage to the robot equipment. Therefore, predicting the movement trajectory of obstacles in advance and taking correct countermeasures can effectively avoid collisions, ensure the safety and experience of workers, and reduce possible injuries and adverse consequences. Summary of the invention

[0004] The purpose of the present invention is to provide a robot inspection data management system and method based on machine learning to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] The robot inspection data management method based on machine learning includes the following steps:

[0007] Step S1: obtaining each checkpoint in a venue, wherein the venue includes a robot for inspection, and the robot plans a moving route according to map information of the venue and each checkpoint in the venue;

[0008] Step S2: While moving along the moving route, the position of the robot is obtained in real time, and the inspection range of the robot is obtained according to the position of the robot and the moving route;

[0009] Step S3: Acquire the image of the moving object in real time, and detect in real time whether the moving object enters the inspection range. If the moving object enters the inspection range, the time when the moving object just enters the inspection range is recorded as T, and the motion trajectory of the moving object within t seconds before the T moment is acquired, and the position, moving direction and moving speed of the moving object at the T moment are obtained;

[0010] Step S4: Obtain the position, moving direction and moving speed of the robot at time T, and obtain the robot detection area according to the position, moving direction and moving speed of the moving object at time T, and determine whether there is a calibrated corner point in the robot detection area. The calibrated corner point is a point at the corner of the venue that is easy to block the line of sight;

[0011] Step S5: If there is no calibrated corner point in the robot detection area, different action instructions are executed on the robot according to the angle between the moving direction of the moving object and the moving direction of the robot;

[0012] Step S6: If there are calibrated corner points in the robot detection area, the moving object detection area is obtained according to the positions, moving directions and moving speeds of the robot and the moving object at time T, and different action instructions are executed on the robot according to the calibrated corner points in the robot detection area and the moving object detection area;

[0013] Step S7: Based on the results of data analysis, generate warning information and optimize the robot inspection route.

[0014] Furthermore, planning the moving route in step S1 includes: acquiring all checkpoints in the venue, inspecting all checkpoints once, and finding the inspection route with the shortest inspection path as the planned moving route.

[0015] Further, step S2 includes the following steps:

[0016] Step S2-1: During the movement of the robot, an area with a radius of r meters and a current position of the robot as the center is obtained, and the area is used as the first inspection range;

[0017] Step S2-2: during the movement of the robot, a route starting from the current position of the robot and moving forward s meters along the moving route is obtained as a forward inspection route, and an area within k meters on both sides of the forward inspection route is taken as a second inspection range;

[0018] Step S2-3: The area where the first inspection range and the second inspection range overlap is used as the third inspection range.

[0019] Furthermore, the moving speed of the moving object at time T obtained in step S3 includes: if the position of the moving object is within the third inspection range, the robot stops moving immediately; if the moving object enters the first inspection range or the second inspection range, but does not enter the third inspection range, obtain the moving distance L of the moving object within t seconds before time T, and obtain the moving speed v=L / t of the moving object at time T.

[0020] Furthermore, the step of obtaining the robot detection area in step S4 includes:

[0021] Step S4-1: Based on the positions of the robot and the moving object at time T, the direction starting from the position of the robot and ending at the position of the moving object is taken as the first target direction, and the angle between the first target direction and the moving direction of the robot is recorded as θ b , 0°≤θ b ≤180°;

[0022] Step S4-2: Take the robot's position at time T as the center of the circle and the angle as θ b , with a radius of h b , obtain the robot detection fan-shaped area, and use the robot detection fan-shaped area as the robot detection area; where the radius of the robot detection fan-shaped area h b The calculation formula is:

[0023]

[0024] Among them, h b is the radius of the fan-shaped area detected by the robot, r is the radius of the first inspection range, θ b is the angle between the first target direction and the robot's moving direction, cosθ b is the angle θ b The cosine value, P T ∈γ1 means that at time T the position of the mobile object is within the first inspection range, L is the distance between the robot position and the mobile object position, k is the length of the forward inspection route from the left and right sides of the second inspection range in step S2-2, P T ∈γ2 indicates that the location of the mobile object at time T is within the second inspection range.

[0025] It should be noted that the use of The purpose is: Since 0°≤θ b ≤180°, then (0.5 is to prevent the radius from being too short), and The closer it is to 1, the closer the moving direction of the moving object is to the moving direction of the robot, and the more likely it is to collide. In order to detect more accurately, the area of ​​the robot's detection fan-shaped area should be larger, so the radius h bThe value should be larger; the closer it is to 0.5, it indicates that the moving direction of the moving object deviates more from the moving direction of the robot, and it is less likely to collide. Therefore, the radius h b should be smaller;

[0026] The purpose of using k(1 - cosθ b ) is that since 0° ≤ θ b ≤ 180°, in order to prevent the detection deviation from being too large due to the short distance L between the position of the robot and the position of the moving object, the present invention considers adding the value of k into the radius h b of the detection sector area of the robot. At the same time, when the value of θ b is larger, the influence degree of the value of k on the radius h b of the detection sector area of the robot should also be greater;

[0027] Furthermore, if there are no calibrated corner points in the detection area of the robot, different action instructions for the robot include:

[0028] If the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies A1 < A0 ≤ 180°, the robot slows down its moving speed; if the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies 0° ≤ A0 < A2, the robot slows down its moving speed or moves at a normal speed; if the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies A2 ≤ A0 ≤ A1, the robot immediately stops moving or moves at a normal speed; where 0° < A2 < A1 < 180° and 0° ≤ A0 ≤ 180°.

[0029] Furthermore, step S6 includes the following steps:

[0030] Step S6 - 1: If there are calibrated corner points in the detection area of the robot, then perform feature marking on the calibrated corner point closest to the robot in the detection area of the robot;

[0031] Step S6 - 2: According to the positions of the moving object and the robot at time T, take the direction from the position of the moving object as the starting point and the position of the robot as the ending point as the second target direction, and denote the included angle between the second target direction and the moving direction of the moving object as θ p , 0° ≤ θ p ≤ 180°;

[0032] Step S6 - 3: Take the position of the moving object at time T as the center of the circle, with an included angle of θ p , and a radius of h p , to obtain the detection sector area of the moving object, and use the detection sector area of the moving object as the detection area of the moving object; where the radius h p of the detection sector area of the moving object is calculated by the formula:

[0033]

[0034] Among them, h p is the radius of the moving object detection sector area, r is the radius of the first inspection range, θ p is the angle between the second target direction and the moving direction of the moving object, cosθ p is the angle θ p The cosine value, P T ∈γ1 means that at time T the position of the mobile object is within the first inspection range, L is the distance between the position of the mobile object and the position of the robot, k is the length of the forward inspection route from the left and right sides of the second inspection range in step S2-2, P T ∈γ2 means that the location of the mobile object at time T is within the second inspection range;

[0035] It should be noted that the use of The purpose is: Since 0°≤θ p ≤180°, θ p The smaller it is, the closer the moving object is to the robot. p The smaller the radius h of the moving object detection sector area is, p should be larger, and θ p The larger the value, the more the moving object and the robot deviate. At this time, the radius of the moving object detection fan-shaped area is h. p The smaller it should be;

[0036] Using kcosθ p The purpose is: p The smaller the value, the closer the moving object is to the robot. The value of k has an influence on the radius h of the moving object detection fan-shaped area. p The influence of θ should also be greater, while p The larger the value, the farther the moving object is from the robot. The value of k has an influence on the radius h of the fan-shaped area where the moving object is detected. p The impact should also be smaller;

[0037] Step S6-4: If there is no calibrated corner point in the moving object detection area, the robot's moving speed remains unchanged; otherwise, the calibrated corner point closest to the moving object in the moving object detection area is marked with a second feature, and if the calibrated corner point of the second feature mark is not the same as the calibrated corner point of the first feature mark, the robot slows down its moving speed.

[0038] Further optimization of the robot inspection route includes:

[0039] Step S7-1: If the calibrated corner point of the second feature mark and the first feature mark in step S6-4 is the same, compare the moving speed v of the moving object at time T with the speed threshold v0, and if v≤v0, the robot slows down the moving speed;

[0040] Step S7-2: If v>v0, the robot stops moving immediately; if the moving object is detected to leave the inspection range within ta seconds after time T, the robot slows down its movement speed; otherwise, the robot turns around and takes the calibrated corner position of the first feature mark as the center of the circle with a radius of R d The area is taken as the calibration area, and all uninspected checkpoints in the venue are obtained. All the uninspected checkpoints are inspected once, and the inspection path does not pass through the calibration area. The inspection route with the shortest inspection path is found as the optimized inspection route of the robot.

[0041] The robot inspection data management system based on machine learning includes the following modules:

[0042] Mobile route module: used to obtain each checkpoint in the venue, the venue includes a robot for inspection, and the robot plans the mobile route according to the map information of the venue and each checkpoint in the venue;

[0043] Inspection range module: used to obtain the robot's position in real time while it is moving along the moving route, and obtain the robot's inspection range based on the robot's position and moving route;

[0044] Moving object module: used to obtain the image of the moving object in real time, and detect in real time whether the moving object enters the inspection range. If the moving object enters the inspection range, the time when the moving object just enters the inspection range is recorded as T, and the motion trajectory of the moving object within t seconds before the T moment is obtained, and the position, moving direction and moving speed of the moving object at the T moment are obtained;

[0045] Robot detection area module: used to obtain the position, moving direction and moving speed of the robot at time T, and obtain the robot detection area according to the position, moving direction and moving speed of the moving object at time T, and determine whether there is a calibrated corner point in the robot detection area. The calibrated corner point is a point at the corner of the venue that is easy to block the line of sight;

[0046] Action execution module: used to execute different action instructions for the robot according to the angle between the moving direction of the moving object and the moving direction of the robot if there is no calibrated corner point in the robot detection area;

[0047] Mobile object detection area module: used to execute the robot detection area. If there are calibrated corner points in the robot detection area, the mobile object detection area is obtained according to the position, moving direction and moving speed of the robot and the mobile object at time T, and different action instructions are executed on the robot according to the calibrated corner points in the robot detection area and the mobile object detection area;

[0048] Inspection route optimization module: used to generate early warning information and optimize robot inspection routes based on data analysis results.

[0049] Furthermore, the inspection range module includes the following units:

[0050] The first inspection range unit is used to obtain an area with a radius of r meters with the current position of the robot as the center during the movement of the robot, and use the area as the first inspection range;

[0051] The second inspection range unit is used to obtain a moving route of s meters forward from the current position of the robot as a forward inspection route during the movement of the robot, and to take an area within k meters on both sides of the forward inspection route as a second inspection range;

[0052] The third inspection range unit is used to take the area where the first inspection range and the second inspection range overlap as the third inspection range.

[0053] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention provides a robot inspection data management system and method based on machine learning, including: planning the robot's moving route according to the checkpoints, and obtaining the robot's inspection range during movement; monitoring and acquiring the moving object, judging whether the moving object enters the inspection range, and if it enters the inspection range, obtaining the position, moving direction and moving speed of the moving object and the robot when they just enter the inspection range, obtaining the robot's detection area, and judging whether there are calibrated corner points in the robot's detection area, and executing different action instructions on the robot according to the position of the calibrated corner points and the moving speed of the moving object. The present invention reasonably plans the robot's operating instructions to slow down the robot, stop the robot, or optimize the inspection route, and realize intelligent inspection of moving objects such as personnel to prevent personnel from being frightened or injured due to collision with the robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0055] Figure 1 It is a flow chart of a robot inspection data management method based on machine learning of the present invention;

[0056] Figure 2 It is a structural diagram of the robot inspection data management system based on machine learning of the present invention;

[0057] Figure 3 It is a schematic diagram of the robot inspection range structure of the robot inspection data management system and method based on machine learning of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] See also Figure 1-Figure 2 The present invention provides a robot inspection data management system and method based on machine learning, and the corresponding flowchart of the robot inspection data management method based on machine learning is as follows: Figure 1 As shown. The following steps are included:

[0060] Step S1: Acquire each checkpoint in the venue, which includes a robot for inspection. The robot plans a moving route according to the map information of the venue and each checkpoint in the venue.

[0061] The location in this embodiment is a cable tunnel. The inspection work of the robot in the cable tunnel is diverse, including facility inspection, equipment status inspection, dangerous point inspection, etc., and these inspection contents are often concerned by workers and frequently used. According to these inspection contents, multiple checkpoints are formed, and the location of each checkpoint is obtained. The cable tunnel also includes robots for inspecting these checkpoints. According to the distribution of these checkpoints and the map information of the cable tunnel, the robot's moving route is planned. The robot's moving route should meet the requirements of inspecting all checkpoints, and each checkpoint should be inspected once, and the inspection route with the shortest inspection path should be found as the robot's moving route. In this implementation, the robot's moving route can be planned using a genetic algorithm or an ant colony algorithm to find the inspection route with the shortest inspection path. Since genetic algorithms and ant colony algorithms are both existing technologies, the specific implementation steps can refer to existing materials and will not be repeated here.

[0062] Step S2: During the movement along the moving route, the robot's position is obtained in real time, and the robot's inspection range is obtained according to the robot's position and the moving route. The inspection range includes a first inspection range, a second inspection range, and a third inspection range.

[0063] Step S2-1: During the movement of the robot, an area with a radius of r meters and a current position of the robot as the center is obtained, and the area is used as the first inspection range.

[0064] As attached Figure 3 As shown, according to the planned movement route of the robot, during the movement, a circular area with a radius of r meters and a current position of the robot as the center is used as the first inspection range, wherein in this embodiment, the radius r=1.

[0065] Step S2-2: During the movement of the robot, a route starting from the current position of the robot and extending s meters forward along the moving route is obtained as a forward inspection route, and an area within k meters on both sides of the forward inspection route is taken as a second inspection range.

[0066] As attached Figure 3 As shown, according to the planned movement route of the robot, during the movement process, the robot's current position is taken as the starting point, and the route s meters forward along the movement route is used as the forward inspection route, and the area within k meters on both sides of the forward inspection route is used as the second inspection range. Among them, the shape of the second inspection range is not all regular, and the area is less than or equal to 2sk.

[0067] Step S2-3: As shown in the attached Figure 3 As shown, the area where the first inspection range and the second inspection range overlap is taken as the third inspection range. It should be noted that the reason for obtaining the third inspection range is that the moving object in the third inspection range is more likely to collide with the robot. When the moving object enters the third inspection range, the robot will be directly commanded to stop moving. The third inspection range does not change the first inspection range and the second inspection range.

[0068] Step S3: Acquire the image of the moving object in real time, and detect in real time whether the moving object enters the inspection range. If the moving object enters the inspection range, the moment when the moving object just enters the inspection range is recorded as T, and the motion trajectory of the moving object within t seconds before the T moment is obtained, and the position, moving direction and moving speed of the moving object at the T moment are obtained.

[0069] The monitoring obtains the image of the moving object in real time. If the position of the moving object is detected to be within the third inspection range during the operation of the robot, the robot stops moving immediately. If the moving object enters the first inspection range or the second inspection range, but does not enter the third inspection range, obtain the moving distance L of the moving object in t seconds before time T, and obtain the moving speed v=L / t of the moving object at time T. In this embodiment, t=1 in the first t seconds; and determine the moving direction of the moving object according to the direction of the moving object when it enters the inspection range.

[0070] Step S4: Obtain the position, moving direction and moving speed of the robot at time T, and obtain the robot detection area based on the position, moving direction and moving speed of the moving object at time T, and determine whether there are calibrated corner points in the robot detection area. The calibrated corner points are points at the corners in the venue that are easy to block the line of sight, such as the turning points of the walls in the cable tunnel.

[0071] Step S4-1: Based on the positions of the robot and the moving object at time T, the direction starting from the position of the robot and ending at the position of the moving object is taken as the first target direction, and the angle between the first target direction and the moving direction of the robot is recorded as θ b , 0°≤θ b ≤180°.

[0072] Among them, the position, moving direction and moving speed of the robot at time T can be obtained through the sensors carried by the robot itself, such as position sensors, speed sensors and angle sensors, etc. The obtaining process is existing technology and will not be repeated here.

[0073] Step S4-2: According to the obtained positions, moving directions and moving speeds of the robot and the moving object at time T, the robot detection area is obtained. Specifically, the robot position at time T is taken as the center of the circle, and the angle is θ b , with a radius of h b , obtain the robot detection fan-shaped area, and use the robot detection fan-shaped area as the robot detection area; where the radius of the robot detection fan-shaped area h b The calculation formula is:

[0074]

[0075] Among them, h b is the radius of the fan-shaped area detected by the robot, r is the radius of the first inspection range, θ b is the angle between the first target direction and the robot's moving direction, cosθ b is the angle θ b The cosine value, P T ∈γ1 means that at time T the position of the mobile object is within the first inspection range, L is the distance between the robot position and the mobile object position, k is the length of the forward inspection route from the left and right sides of the second inspection range in step S2-2, P T ∈γ2 indicates that the location of the mobile object at time T is within the second inspection range.

[0076] It should be noted that the use of The purpose is: Since 0°≤θ b ≤180°, then (0.5 is to prevent the radius from being too short), and The closer it is to 1, the closer the moving direction of the moving object is to the moving direction of the robot, and the more likely a collision is to occur. To detect more accurately, the area of the fan-shaped area detected by the robot should be larger, so the radius h b should be larger; the closer it is to 0.5, the more deviated the moving direction of the moving object is from the moving direction of the robot, and the less likely a collision is to occur. Therefore, the radius h b should be smaller.

[0077] The purpose of using k(1 - cosθ b ) is that since 0° ≤ θ b ≤ 180°, in order to prevent the detection deviation from being too large due to the short distance L between the position of the robot and the position of the moving object, the present invention considers adding the value of k into the radius h b of the fan-shaped area detected by the robot. At the same time, when the value of θ b is larger, the influence degree of the value of k on the radius h b of the fan-shaped area detected by the robot should also be greater.

[0078] Step S5: If there are no calibrated corner points in the detection area of the robot, different action instructions are executed for the robot according to the included angle between the moving direction of the moving object and the moving direction of the robot.

[0079] If there are no calibrated corner points in the detection area of the robot, it indicates that the robot and the moving object do not meet at a corner. According to such a situation, different encounter situations are divided between the robot and the moving object, and different encounter situations have different response measures and warning levels for the robot. Specifically, they are divided into the following types:

[0080] If the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies A1 < A0 ≤ 180°, the robot slows down its moving speed. This situation indicates that they meet in the same direction. In a cable tunnel, if a worker sees a robot coming head-on and meets directly in a narrow passage or other places. Since the worker can directly observe the robot in this situation, the danger level is relatively low, and the robot slows down its action speed.

[0081] If the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies 0° ≤ A0 < A2, the robot slows down its moving speed or moves at a normal speed; this situation indicates that they meet head-on. The robot and the moving object are moving in the same direction. It may be that the robot approaches the moving object from behind, in which case the robot slows down its moving speed, or the moving object approaches the robot from behind, in which case the robot moves at a normal speed.

[0082] If the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies A2 ≤ A0 ≤ A1, the robot immediately stops moving or moves at a normal speed; this situation indicates that the travel routes of both parties cross at a certain angle. It may be that the robot is behind the moving object, in which case the robot immediately stops moving, or the moving object is behind the robot, in which case the robot moves at a normal speed. Among them, 0° < A2 < A1 < 180°, 0° ≤ A0 ≤ 180°. In this embodiment, A2 = 40°, A1 = 140°, and the specific values can be adjusted according to the actual situation.

[0083] Step S6: If there are calibration corner points in the robot detection area, obtain the moving object detection area based on the positions, moving directions, and moving speeds of the robot and the moving object at time T, and execute different action instructions on the robot according to the calibration corner point conditions in the robot detection area and the moving object detection area.

[0084] It should be noted that when the robot and the moving object accidentally meet at a corner, the degree of danger is the highest and the judgment is the most complex. If there are groups such as workers who are working and their attention is not on the robot, collisions are very likely to occur. Therefore, the present invention judges the encounter at the corner according to the following steps:

[0085] Step S6-1: If there are calibration corner points in the robot detection area, perform feature marking on the calibration corner point closest to the robot in the robot detection area.

[0086] Step S6-2: According to the positions of the moving object and the robot at time T, use the direction from the position of the moving object as the starting point to the position of the robot as the ending point as the second target direction, and denote the included angle between the second target direction and the moving direction of the moving object as θ p , 0° ≤ θ p ≤ 180°.

[0087] Step S6-3: According to the obtained positions, moving directions, and moving speeds of the robot and the moving object at time T, obtain the moving object detection area, specifically: with the position of the moving object at time T as the center of the circle, the included angle is θ p , and the radius is h p , obtain the moving object detection sector area, and use the moving object detection sector area as the moving object detection area; among them, the radius h p of the moving object detection sector area is calculated by the formula:

[0088]

[0089] Among them, h p is the radius of the moving object detection sector area, r is the radius of the first inspection range, θ pis the angle between the second target direction and the moving direction of the moving object, cosθ p is the angle θ p The cosine value, P T ∈γ1 means that at time T the position of the mobile object is within the first inspection range, L is the distance between the position of the mobile object and the position of the robot, k is the length of the forward inspection route from the left and right sides of the second inspection range in step S2-2, P T ∈γ2 indicates that the location of the mobile object at time T is within the second inspection range.

[0090] It should be noted that the use of The purpose is: Since 0°≤θ p ≤180°, θ p The smaller it is, the closer the moving object is to the robot. p The smaller the radius h of the moving object detection sector area is, p should be larger, and θ p The larger the value, the more the moving object and the robot deviate. At this time, the radius of the moving object detection fan-shaped area is h. p Should be smaller.

[0091] Using kcosθ p The purpose is: p The smaller the value, the closer the moving object is to the robot. The value of k has an influence on the radius h of the moving object detection fan-shaped area. p The influence of θ should also be greater, while p The larger the value, the farther the moving object is from the robot. The value of k has an influence on the radius h of the fan-shaped area where the moving object is detected. p The impact should also be smaller.

[0092] Step S6-4: If there is no calibrated corner point in the moving object detection area, the robot's moving speed remains unchanged; otherwise, the calibrated corner point closest to the moving object in the moving object detection area is marked with a second feature, and if the calibrated corner point of the second feature mark is not the same as the calibrated corner point of the first feature mark, the robot slows down its moving speed.

[0093] Step S7: Based on the results of data analysis, generate warning information and optimize the robot inspection route.

[0094] Step S7-1: If the calibrated corner point of the second characteristic mark and the first characteristic mark in step S6-4 is the same, compare the moving speed v of the moving object at time T with the speed threshold v0, if v≤v0, the robot slows down. In this embodiment, the speed threshold v0 is 5km / h.

[0095] Step S7-2: If v>v0, the robot stops moving immediately; if the moving object is detected to leave the inspection range within ta seconds after time T, the robot slows down its movement speed; otherwise, the robot turns around and takes the calibrated corner position of the first feature mark as the center of the circle with a radius of R d The area is taken as the calibration area, all uninspected checkpoints in the place are obtained, and all the uninspected checkpoints are inspected once, and the inspection path does not pass through the calibration area, and the inspection route with the shortest inspection path is found as the optimized inspection route of the robot. In this embodiment, t α =1.

[0096] The present invention plans the robot's moving route according to the checkpoints, and obtains the robot's inspection range during movement; monitors and obtains the moving object, determines whether the moving object enters the inspection range, and if it enters the inspection range, obtains the position, moving direction and moving speed of the moving object and the robot when they just enter the inspection range, obtains the robot's detection area, and determines whether there are calibrated corner points in the robot's detection area, and according to the position of the calibrated corner points and the moving speed of the moving object, executes different action instructions to the robot. By rationally planning the robot's operating instructions, or slowing down the robot, or stopping the robot, or optimizing the robot's inspection route, timely and intelligent inspections of moving objects such as personnel can be achieved to prevent personnel from being frightened or injured due to collisions with the robot.

[0097] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0098] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A robot inspection data management method based on machine learning, characterized in that: Including the following steps: Step S1: Obtain each checkpoint within the venue. There is a robot for patrol inspection within the venue. The robot plans a moving route according to the map information of the venue and each checkpoint within the venue; Step S2: During the movement according to the moving route, obtain the position of the robot in real time, and obtain the patrol inspection range of the robot based on the position of the robot and the moving route; Step S3: Obtain the moving object image in real time, and detect in real time whether the moving object enters the patrol inspection range. If the moving object enters the patrol inspection range, record the moment when the moving object just enters the patrol inspection range as T, obtain the movement trajectory of the moving object within t seconds before the moment T, and obtain the position, moving direction and moving speed of the moving object at the moment T; Step S4: Obtain the position, moving direction and moving speed of the robot at the moment T, and obtain the detection area of the robot based on the position, moving direction and moving speed of the moving object at the moment T, and judge whether there is a calibration corner point within the detection area of the robot. The calibration corner point is a point at the corner within the venue where the line of sight is easily blocked; Step S5: If there is no calibration corner point within the detection area of the robot, execute different action instructions on the robot according to the included angle between the moving direction of the moving object and the moving direction of the robot; If the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies A1 < A0 ≤ 180°, the robot slows down its moving speed; if the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies 0° ≤ A0 < A2, the robot slows down its moving speed or moves at a normal speed; if the included angle A0 between the moving direction of the moving object and the moving direction of the robot satisfies A2 ≤ A0 ≤ A1, the robot immediately stops moving or moves at a normal speed; where 0° < A2 < A1 < 180°, 0° ≤ A0 ≤ 180°; Step S6: If there is a calibration corner point within the detection area of the robot, obtain the detection area of the moving object based on the positions, moving directions and moving speeds of the robot and the moving object at the moment T, and execute different action instructions on the robot according to the calibration corner point situation within the detection area of the robot and the detection area of the moving object; Step S6-1: If there is a calibration corner point within the detection area of the robot, perform a first feature marking on the calibration corner point closest to the robot within the detection area of the robot; Step S6-2: Based on the positions of the moving object and the robot at time T, the direction with the position of the moving object as the starting point and the position of the robot as the end point is taken as the second target direction, and the angle between the second target direction and the moving direction of the moving object is recorded as θ p , 0°≤θ p ≤180°; Step S6-3: Take the position of the moving object at time T as the center of the circle and the angle as θ p , with a radius of h p , obtain a moving object detection sector area, and use the moving object detection sector area as the moving object detection area; wherein the radius h of the moving object detection sector area p The calculation formula is: Among them, h p is the radius of the moving object detection sector area, r is the radius of the first inspection range, θ p is the angle between the second target direction and the moving direction of the moving object, cosθ p is the angle θ p The cosine value, P T ∈γ1 means that at time T the position of the mobile object is within the first inspection range, L is the distance between the position of the mobile object and the position of the robot, k is the length of the forward inspection route from the left and right sides of the second inspection range in step S2-2, P T ∈γ2 means that the location of the mobile object at time T is within the second inspection range; Step S6-4: If there is no calibration corner point within the detection area of the moving object, the moving speed of the robot remains unchanged; otherwise, perform a second feature marking on the calibration corner point closest to the moving object within the detection area of the moving object. If the calibration corner points of the second feature marking and the first feature marking are not the same, the robot slows down its moving speed; Step S7: Generate a warning message based on the result of data analysis and optimize the patrol inspection route of the robot.

2. The robot inspection data management method based on machine learning according to claim 1 is characterized in that: The moving route planning in Step S1 includes: obtaining all checkpoints within the venue, performing patrol inspections on all the checkpoints, and each checkpoint is inspected once, and finding the patrol inspection route with the shortest patrol path as the planned moving route.

3. The robot inspection data management method based on machine learning according to claim 1 is characterized in that: Step S2 includes the following steps: Step S2-1: During the movement of the robot, an area with a radius of r meters and a current position of the robot as the center is obtained, and the area is used as a first inspection range; Step S2-2: during the movement of the robot, a route starting from the current position of the robot and moving s meters forward along the moving route is obtained as a forward inspection route, and an area within k meters on both sides of the forward inspection route is taken as a second inspection range; Step S2-3: The area where the first inspection range and the second inspection range overlap is used as the third inspection range.

4. The robot inspection data management method based on machine learning according to claim 1 is characterized in that: The moving speed of the moving object at time T obtained in step S3 includes: if the position of the moving object is within the third inspection range, the robot stops moving immediately; if the moving object enters the first inspection range or the second inspection range, but does not enter the third inspection range, obtain the moving distance L of the moving object within t seconds before time T, and obtain the moving speed v=L / t of the moving object at time T.

5. The robot inspection data management method based on machine learning according to claim 1 is characterized in that: The step of obtaining the robot detection area in step S4 includes: Step S4-1: Based on the positions of the robot and the moving object at time T, the direction starting from the position of the robot and ending at the position of the moving object is taken as the first target direction, and the angle between the first target direction and the moving direction of the robot is recorded as θ b , 0°≤θ b ≤180°; Step S4-2: Take the robot's position at time T as the center of the circle and the angle as θ b , with a radius of h b , obtain the robot detection fan-shaped area, and use the robot detection fan-shaped area as the robot detection area; wherein the radius h of the robot detection fan-shaped area b The calculation formula is: Among them, h b is the radius of the fan-shaped area detected by the robot, r is the radius of the first inspection range, θ b is the angle between the first target direction and the robot's moving direction, cosθ b is the angle θ b The cosine value, P T ∈γ1 means that at time T the position of the mobile object is within the first inspection range, L is the distance between the robot position and the mobile object position, k is the length of the forward inspection route from the left and right sides of the second inspection range in step S2-2, P T ∈γ2 indicates that the location of the mobile object at time T is within the second inspection range.

6. The robot inspection data management method based on machine learning according to claim 1 is characterized in that: Optimizing robot inspection routes includes: Step S7-1: If the calibrated corner point of the second feature mark and the first feature mark in step S6-4 is the same, compare the moving speed v of the moving object at time T with the speed threshold v0, and if v≤v0, the robot slows down the moving speed; Step S7-2: If v>v0, the robot stops moving immediately; if it moves t after time T a If the moving object leaves the inspection range within seconds, the robot will slow down its movement speed. Otherwise, the robot will turn around and take the calibrated corner position of the first feature mark as the center of the circle with a radius of R. d The area is taken as the calibration area, and all uninspected checkpoints in the venue are obtained. All the uninspected checkpoints are inspected once, and the inspection path does not pass through the calibration area. The inspection route with the shortest inspection path is found as the optimized inspection route of the robot.

7. A robot inspection data management system for executing the robot inspection data management method based on machine learning as described in any one of claims 1 to 6, characterized in that: Includes the following modules: Mobile route module: used to obtain each checkpoint in the venue, the venue includes a robot for inspection, and the robot plans the mobile route according to the map information of the venue and each checkpoint in the venue; Inspection range module: used to obtain the robot's position in real time while it is moving along the moving route, and obtain the robot's inspection range based on the robot's position and moving route; Moving object module: used to obtain the image of the moving object in real time, and detect in real time whether the moving object enters the inspection range. If the moving object enters the inspection range, the time when the moving object just enters the inspection range is recorded as T, and the motion trajectory of the moving object within t seconds before the T moment is obtained, and the position, moving direction and moving speed of the moving object at the T moment are obtained; Robot detection area module: used to obtain the position, moving direction and moving speed of the robot at time T, and obtain the robot detection area according to the position, moving direction and moving speed of the moving object at time T, and determine whether there is a calibrated corner point in the robot detection area. The calibrated corner point is a point at the corner of the venue that is easy to block the line of sight; Action execution module: used to execute different action instructions for the robot according to the angle between the moving direction of the moving object and the moving direction of the robot if there is no calibrated corner point in the robot detection area; Mobile object detection area module: used to execute the robot detection area. If there are calibrated corner points in the robot detection area, the mobile object detection area is obtained according to the position, moving direction and moving speed of the robot and the mobile object at time T, and different action instructions are executed on the robot according to the calibrated corner points in the robot detection area and the mobile object detection area; Inspection route optimization module: used to generate early warning information and optimize robot inspection routes based on data analysis results.

8. The robot inspection data management system according to claim 7, characterized in that: The inspection range module includes the following units: The first inspection range unit is used to obtain an area with a radius of r meters with the current position of the robot as the center during the movement of the robot, and use the area as the first inspection range; The second inspection range unit is used to obtain, during the movement of the robot, a moving route of s meters forward starting from the current position of the robot as a forward inspection route, and an area within k meters on both sides of the forward inspection route as a second inspection range; The third inspection range unit is used to take the area where the first inspection range and the second inspection range overlap as the third inspection range.

Images