A positioning control method and system for automobile body detection

Abstract

The present invention provides a positioning control method and system for automobile body detection, wherein the method includes: monitoring the key point distribution when the robot group autonomously performs preliminary body detection on the automobile in a first time period; deciding the depth detection scheme based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model diagram of the automobile; based on the depth detection scheme, controlling the robot group to perform deep body detection on the automobile in a second time period; wherein the second time period is after the first time period and adjacent to the first time period. The robot of the present invention can perform deep detection on the detected body anomaly without the need for manual completion by staff, which reduces labor costs and improves the comprehensiveness and detection efficiency of automobile body detection.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile body detection positioning control, and in particular to a positioning control method and system for automobile body detection. Background Art

[0002] Before a car leaves the factory, it needs to be inspected to ensure the quality of the car body. With the continuous development of automation technology, the current car body inspection is mostly performed by robots. However, most robots are set with a fixed control program for car body inspection, that is, the robot can only perform car body inspection according to fixed inspection steps. When the robot detects an abnormality in the car body, it can only output an alarm and cannot perform in-depth inspection on the detected abnormality. However, the in-depth inspection work is mostly performed by the staff who receive the alarm, which has a high labor cost and reduces the comprehensiveness and efficiency of car body inspection.

[0003] Therefore, a solution is urgently needed. Summary of the invention

[0004] One of the purposes of the present invention is to provide a positioning control method for automobile body inspection, which monitors the distribution of key points located by a robot group when it autonomously performs preliminary body inspection on a car within a first time period, decides on a depth inspection plan based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model diagram of the car, and controls the robot group to perform deep body inspection on the car within a second time period based on the depth inspection plan. The robot can perform deep inspection on detected body abnormalities without the need for manual completion by staff, thereby reducing labor costs and improving the comprehensiveness and efficiency of automobile body inspection.

[0005] An embodiment of the present invention provides a positioning control method for automobile body detection, comprising:

[0006] The distribution of key points located when the monitoring robot group autonomously performs preliminary body inspection of the car in the first time period;

[0007] Determine the depth detection solution based on the key point distribution, individual position distribution of the robot group, and the 3D model of the car;

[0008] Based on the deep detection scheme, the robot group is controlled to perform deep body detection on the car in the second time period;

[0009] The second time period is after the first time period and adjacent to the first time period.

[0010] Preferably, the depth detection scheme is determined based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model diagram of the car, including:

[0011] Based on the point cluster division conditions, multiple point clusters are divided from the key point distribution;

[0012] Traverse each point cluster in turn;

[0013] During each traversal, the traversed point cluster is characterized to obtain a first characteristic vector;

[0014] Determining detection knowledge corresponding to the first characterization vector from a detection knowledge base;

[0015] Based on detection knowledge, determine the area search rules and detection constraints;

[0016] Based on the area search rules, the area location distribution of the area to be inspected is searched on the three-dimensional model map and associated with the detection constraint conditions;

[0017] After traversing each point cluster, the detection posture route is planned based on the detection posture route planning system according to the regional location distribution and individual location distribution;

[0018] A depth detection scheme is generated based on the detection constraints associated with the detection posture route and the regional position distribution.

[0019] Preferably, the point cluster division conditions include:

[0020] The same point cluster contains at least two key points in the key point distribution;

[0021] and / or,

[0022] Each key point in the same point cluster has a standard point relationship between each other;

[0023] and / or,

[0024] The matching degree between the multimodal information corresponding to each key point in the same point cluster in the three-dimensional model image and the standard multimodal information is greater than or equal to the matching degree threshold.

[0025] Preferably, the positioning control method for automobile body detection further includes:

[0026] When a target robot in the robot group detects an abnormal vehicle body event, the severity of the abnormal vehicle body event is analyzed;

[0027] When the severity is greater than or equal to the severity threshold, the online expert corresponding to the abnormal vehicle body event is connected to the detection perspective of the target robot;

[0028] Based on the depth detection scheme, determine the safe parking space of the target robot within a preset time period in the future;

[0029] Based on the perspective planning template corresponding to the abnormal events of the vehicle body, multiple observation perspectives are planned according to the safe parking space;

[0030] Based on each observation perspective, generate an observation perspective list;

[0031] Displaying a list of observation perspectives to online experts;

[0032] When the online expert selects an observation perspective from the observation perspective list, the target robot is controlled to move so that the online expert accesses the selected observation perspective.

[0033] Preferably, the positioning control method for automobile body detection further includes:

[0034] Based on the perspective screening condition, the target observation perspective is screened out from various observation perspectives;

[0035] Performing characterization processing on the visible picture under the target observation angle to obtain a second characterization vector;

[0036] Determining a detection intention corresponding to the second characterization vector from a detection intention library;

[0037] Based on the detection intention, the deep detection plan is modified;

[0038] Based on the revised deep detection scheme, the robot group is relay-controlled to perform deep body detection on the car in the second time period;

[0039] The viewing angle screening conditions include:

[0040] The total number of historical times that online experts select the target observation perspective is greater than or equal to the number threshold;

[0041] and / or,

[0042] The total duration of the historical perspective of the target observation perspective is greater than or equal to the first duration threshold;

[0043] and / or,

[0044] The online expert has selected the target observation perspective in the third time period and the fourth time period, and the perspective maintenance duration after selecting other observation perspectives in the fifth time period between the third time period and the fourth time period is less than or equal to the second time threshold; the fourth event segment is after the third time period.

[0045] An embodiment of the present invention provides a positioning control system for automobile body detection, comprising:

[0046] A monitoring module, used to monitor the distribution of key points located when the robot group autonomously performs preliminary body inspection on the car in the first time period;

[0047] A decision module is used to decide the depth detection scheme based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model of the car;

[0048] A detection module, used to control the robot group to perform a deep body detection on the car within a second time period based on the deep detection solution;

[0049] The second time period is after the first time period and adjacent to the first time period.

[0050] Preferably, the decision module decides on a depth detection scheme based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model diagram of the car, including:

[0051] Based on the point cluster division conditions, multiple point clusters are divided from the key point distribution;

[0052] Traverse each point cluster in turn;

[0053] During each traversal, the traversed point cluster is characterized to obtain a first characteristic vector;

[0054] Determining detection knowledge corresponding to the first characterization vector from a detection knowledge base;

[0055] Based on detection knowledge, determine the area search rules and detection constraints;

[0056] Based on the area search rules, the area location distribution of the area to be inspected is searched on the three-dimensional model map and associated with the detection constraint conditions;

[0057] After traversing each point cluster, the detection posture route is planned based on the detection posture route planning system according to the regional location distribution and individual location distribution;

[0058] A depth detection scheme is generated based on the detection constraints associated with the detection posture route and the regional position distribution.

[0059] Preferably, the point cluster division conditions include:

[0060] The same point cluster contains at least two key points in the key point distribution;

[0061] and / or,

[0062] Each key point in the same point cluster has a standard point relationship between each other;

[0063] and / or,

[0064] The matching degree between the multimodal information corresponding to each key point in the same point cluster in the three-dimensional model image and the standard multimodal information is greater than or equal to the matching degree threshold.

[0065] Preferably, the positioning control system for automobile body detection further includes:

[0066] Interaction modules for:

[0067] When a target robot in the robot group detects an abnormal vehicle body event, the severity of the abnormal vehicle body event is analyzed;

[0068] When the severity is greater than or equal to the severity threshold, the online expert corresponding to the abnormal vehicle body event is connected to the detection perspective of the target robot;

[0069] Based on the depth detection scheme, determine the safe parking space of the target robot within a preset time period in the future;

[0070] Based on the perspective planning template corresponding to the abnormal events of the vehicle body, multiple observation perspectives are planned according to the safe parking space;

[0071] Based on each observation perspective, generate an observation perspective list;

[0072] Displaying a list of observation perspectives to online experts;

[0073] When the online expert selects an observation perspective from the observation perspective list, the target robot is controlled to move so that the online expert accesses the selected observation perspective.

[0074] Preferably, the positioning control system for automobile body detection further includes:

[0075] Correction modules for:

[0076] Based on the perspective screening condition, the target observation perspective is screened out from various observation perspectives;

[0077] Performing characterization processing on the visible picture under the target observation angle to obtain a second characterization vector;

[0078] Determining a detection intention corresponding to the second characterization vector from a detection intention library;

[0079] Based on the detection intention, the deep detection plan is modified;

[0080] Based on the revised deep detection scheme, the robot group is relay-controlled to perform deep body detection on the car in the second time period;

[0081] The viewing angle screening conditions include:

[0082] The total number of historical times that online experts select the target observation perspective is greater than or equal to the number threshold;

[0083] and / or,

[0084] The total duration of the historical perspective of the target observation perspective is greater than or equal to the first duration threshold;

[0085] and / or,

[0086] The online expert has selected the target observation perspective in the third time period and the fourth time period, and the perspective maintenance duration after selecting other observation perspectives in the fifth time period between the third time period and the fourth time period is less than or equal to the second time threshold; the fourth event segment is after the third time period.

[0087] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0088] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] 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:

[0090] Figure 1 A schematic diagram of a positioning control method for automobile body detection in an embodiment of the present invention;

[0091] Figure 2 The figure is a schematic diagram of a positioning control system for automobile body detection in an embodiment of the present invention. DETAILED DESCRIPTION

[0092] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0093] The embodiment of the present invention provides a positioning control method for automobile body detection, such as Figure 1 As shown, including:

[0094] S1, the distribution of key points located when the monitoring robot group autonomously performs preliminary body inspection of the car in the first time period;

[0095] S2, decide the depth detection scheme based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model of the car;

[0096] S3, based on the depth detection scheme, controlling the robot group to perform a depth body detection on the car in the second time period;

[0097] The second time period is after the first time period and adjacent to the first time period.

[0098] There are multiple robots in the robot group that perform automobile body inspection; the starting time of the first time period is the time when the robot group starts to inspect the automobile body, and the length of the first time period is the average time it takes for the robot group to complete the automobile body inspection; the robot group performs preliminary automobile body inspection according to a preset program, for example: detects pits, scratches, uneven painting and other problems on the automobile body, and locates the key point distribution, which is the position of the automobile body that produces these pits, scratches, uneven painting and other problems; the individual position distribution of the robot group is the current position of each robot in the robot group; the three-dimensional model diagram of the automobile is a 1:1 simulated three-dimensional model of the automobile body; based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model diagram of the automobile, a depth detection plan is decided, and the depth detection plan is a control plan for controlling the robot group to perform deep body detection on the automobile; based on the depth detection plan, the robot group is controlled to perform deep body detection on the automobile in the second time period; the length of the second time period can be set in advance by the technician according to actual needs, for example: 3 minutes.

[0099] The present application monitors the distribution of key points located by a robot group when it autonomously performs a preliminary body inspection of a car in a first time period, decides on a depth detection plan based on the distribution of key points, the individual position distribution of the robot group, and the three-dimensional model diagram of the car, and controls the robot group to perform a deep body inspection of the car in a second time period based on the depth detection plan. The robot can perform a deep inspection of the detected body abnormalities without the need for manual completion by staff, thereby reducing labor costs and improving the comprehensiveness and efficiency of car body inspection.

[0100] In one embodiment, the depth detection scheme is determined based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model of the car, including:

[0101] Based on the point cluster division conditions, multiple point clusters are divided from the key point distribution;

[0102] Traverse each point cluster in turn;

[0103] During each traversal, the traversed point cluster is characterized to obtain a first characteristic vector;

[0104] Determining detection knowledge corresponding to the first characterization vector from a detection knowledge base;

[0105] Based on detection knowledge, determine the area search rules and detection constraints;

[0106] Based on the area search rules, the area location distribution of the area to be inspected is searched on the three-dimensional model map and associated with the detection constraint conditions;

[0107] After traversing each point cluster, the detection posture route is planned based on the detection posture route planning system according to the regional location distribution and individual location distribution;

[0108] Generate a deep detection solution based on the detection constraints associated with the detection posture route and regional location distribution;

[0109] The point cluster division conditions include:

[0110] The same point cluster contains at least two key points in the key point distribution;

[0111] and / or,

[0112] Each key point in the same point cluster has a standard point relationship between each other;

[0113] and / or,

[0114] The matching degree between the multimodal information corresponding to each key point in the same point cluster in the three-dimensional model image and the standard multimodal information is greater than or equal to the matching degree threshold.

[0115] Based on the point cluster division conditions, multiple point clusters are divided from the key point distribution. The key points in the same point cluster need to jointly and simultaneously reflect the situation of a decision-making depth detection plan. Each point cluster is traversed in turn. Each time the traversal is performed, the traversed point cluster is characterized to obtain a first characterization vector in vector form, and the first characterization vector reflects the situation. There is detection knowledge corresponding to different first characterization vectors in the detection knowledge base. The detection knowledge indicates how to make a decision on the depth detection plan in this situation. For example: the key points in the point cluster that forms the first characterization vector are the positions of the three pits on the vehicle hood, and the corresponding detection knowledge is to scan and determine the depth of each pit, and scan to determine whether there are other pits on the vehicle hood; based on the detection knowledge, the regional search rules and detection constraints can be determined. The regional search rules are the rules for searching the area to be inspected that needs to be subjected to deep vehicle body detection. The detection constraints are the constraints required for deep vehicle body detection in the area to be inspected, for example: the regional search rules The hood area on the car body is searched out as the area to be inspected, and the detection constraint condition is to scan to determine the depth of each pit, and scan to determine whether there are other pits on the hood of the car body; based on the regional search rule, the regional position distribution of the area to be inspected is searched out on the three-dimensional model map, and associated with the detection constraint condition; after traversing each point cluster, based on the detection posture route planning system, according to the regional position distribution and the individual position distribution, the detection posture route is planned. The detection posture route is the posture change movement route of each robot in the robot group for deep body detection of the area to be inspected. The detection posture route planning system can be a route planning algorithm, which belongs to the existing technical scope of robot control and is not elaborated here; based on the detection constraint condition associated with the detection posture route and the regional position distribution, a deep detection plan is generated; in actual control, each robot in the robot group moves according to the detection posture route, and when it is close to the area to be inspected, the area to be inspected is subjected to deep body detection according to the detection constraint condition associated with the position distribution of the area to be inspected.

[0116] In the point cluster division conditions, firstly, it is ensured that the same point cluster contains at least two key points to avoid meaningless situations caused by a single key point forming a point; secondly, it is ensured that each key point in the same point cluster has a standard point relationship between each other. The standard point relationship can be that they belong to the same body part and the straight-line distance between the key points is less than 3 meters, etc. In this way, the key points in the same point cluster can reasonably reflect the situation of a decision depth detection scheme; finally, the multimodal information corresponding to each key point in the same point cluster in the three-dimensional model diagram refers to the structural information between the key points. The standard multimodal information is the information indicating that the key points in the same point cluster can reasonably reflect the situation of a decision depth detection scheme, for example: indicating the relationship of mutual support between the structures of the key points. When the matching degree of the multimodal information and the multimodal information is greater than or equal to the matching degree threshold, it means that the key points in the same point cluster can reasonably reflect the situation of a decision depth detection scheme; the matching degree threshold can be, for example, 85%.

[0117] The embodiment of the present invention introduces point cluster division conditions when deciding on a depth detection scheme based on the distribution of key points, the individual position distribution of the robot group and the three-dimensional model diagram of the car, and performs rough characterization processing on a unit separately to obtain the corresponding detection knowledge with the first characterization vector, thereby improving the pertinence and rationality of the detection knowledge determination, improving the system work efficiency, using the detection knowledge to determine the area search rules and detection constraints, and completing the generation of the depth detection scheme, thereby improving the efficiency of the depth detection scheme decision-making.

[0118] In one embodiment, the positioning control method for automobile body detection further includes:

[0119] When the target robot in the robot group detects an abnormal vehicle body event, the severity of the abnormal vehicle body event is analyzed; the abnormal vehicle body event is an event in which an abnormal vehicle body occurs, and the severity of the abnormal vehicle body event represents the severity of the abnormal vehicle body event;

[0120] When the severity is greater than or equal to the severity threshold, the online expert corresponding to the vehicle body abnormal event is connected to the detection view of the target robot; the severity threshold may be, for example, 8; when the severity is greater than or equal to the severity threshold, it means that the severity of the vehicle body abnormal event is relatively large, and the online expert corresponding to the vehicle body abnormal event is connected to the detection view of the target robot for online expert viewing. The online expert corresponding to the vehicle body abnormal event refers to an automotive expert who is good at solving vehicle body abnormal events;

[0121] Based on the depth detection scheme, determine the safe parking space of the target robot within a preset time length in the future; the depth detection scheme contains the future detection posture routes of different robots in the robot group, so the safe parking space of the target robot within a preset time length in the future can be determined. The preset time length can be 5 minutes. The safe parking space refers to the space where the target robot can move automatically without colliding with other robots;

[0122] Based on the perspective planning template corresponding to the abnormal event of the vehicle body, multiple observation perspectives are planned according to the safe parking space; the perspective planning template is an execution template for planning multiple observation perspectives for observing the vehicle body in the safe parking space in the safe parking space;

[0123] Based on each observation perspective, generate an observation perspective list;

[0124] Displaying a list of observation perspectives to online experts;

[0125] When the online expert selects an observation perspective from the observation perspective list, the target robot is controlled to move so that the online expert accesses the selected observation perspective.

[0126] When the target robot detects a serious vehicle body abnormality event in the embodiment of the present invention, it directly remotely accesses the online expert and plans and determines the safe parking space and observation angle, which greatly improves the timeliness and convenience of the online expert in understanding the situation of the vehicle body abnormality event and other processing.

[0127] In one embodiment, the positioning control method for automobile body detection further includes:

[0128] Based on the perspective screening condition, the target observation perspective is screened out from various observation perspectives;

[0129] Performing characterization processing on the visible picture under the target observation angle to obtain a second characterization vector;

[0130] Determining a detection intention corresponding to the second characterization vector from a detection intention library;

[0131] Based on the detection intention, the deep detection plan is modified;

[0132] Based on the revised deep detection scheme, the robot group is relay-controlled to perform deep body detection on the car in the second time period;

[0133] The viewing angle screening conditions include:

[0134] The total number of historical times that online experts select the target observation perspective is greater than or equal to the number threshold;

[0135] and / or,

[0136] The total duration of the historical perspective of the target observation perspective is greater than or equal to the first duration threshold;

[0137] and / or,

[0138] The online expert has selected the target observation perspective in the third time period and the fourth time period, and the perspective maintenance duration after selecting other observation perspectives in the fifth time period between the third time period and the fourth time period is less than or equal to the second time threshold; the fourth event segment is after the third time period.

[0139] The target observation perspective is an observation perspective that reflects the online expert's intention to inspect other parts of the car body; the visual image under the target observation perspective is characterized to obtain a second characterization vector in vector form; the visual image under the target observation perspective reflects how the online expert wants to inspect other parts of the car body, so the second characterization vector represents the reflected situation; there are detection intentions corresponding to different second characterization vectors in the detection intention library, for example: if the visible image is the image of the left rearview mirror of the car, it means that the online expert wants to inspect the right rearview mirror of the car, then the detection intention corresponding to the second characterization vector formed by the visible image is to perform depth detection on the right rearview mirror of the car; based on the detection intention, the depth detection scheme is corrected; based on the corrected depth detection scheme, the machine is relayed to control in the second time period. A group of people conducts in-depth body detection of a car; in the perspective screening conditions, the number threshold may be, for example, 3 times; the first duration threshold may be, for example, 10 seconds; the second duration threshold may be, for example, 5 seconds; when the total number of historical times that the online expert selects the target observation perspective is greater than or equal to the number threshold and / or the total historical perspective retention time of the target observation perspective is greater than or equal to the first duration threshold and or the online expert has selected the target observation perspective in both the third time period and the fourth time period and the perspective retention time after selecting other observation perspectives in the fifth time period between the third time period and the fourth time period is less than or equal to the second duration threshold, it means that the visualization picture under the target observation perspective reflects that the online expert has the intention to detect other parts of the car body; the third time period may be 10 seconds; the fourth time period may be 15 seconds.

[0140] The embodiment of the present invention automatically identifies the detection intention of the online expert, and based on the detection intention, modifies the depth detection scheme, thereby improving the humanization, and introduces the perspective screening condition to improve the accuracy and rationality of determining the target observation perspective.

[0141] The embodiment of the present invention provides a positioning control system for automobile body detection, such as Figure 2 As shown, including:

[0142] Monitoring module 1, used to monitor the distribution of key points located when the robot group autonomously performs preliminary body inspection on the car in the first time period;

[0143] Decision module 2, used to decide the depth detection scheme based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model of the car;

[0144] Detection module 3, used to control the robot group to perform deep body detection on the car within a second time period based on the deep detection solution;

[0145] The second time period is after the first time period and adjacent to the first time period.

[0146] The decision module 2 decides the depth detection scheme based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model of the car, including:

[0147] Based on the point cluster division conditions, multiple point clusters are divided from the key point distribution;

[0148] Traverse each point cluster in turn;

[0149] During each traversal, the traversed point cluster is characterized to obtain a first characteristic vector;

[0150] Determining detection knowledge corresponding to the first characterization vector from a detection knowledge base;

[0151] Based on detection knowledge, determine the area search rules and detection constraints;

[0152] Based on the area search rules, the area location distribution of the area to be inspected is searched on the three-dimensional model map and associated with the detection constraint conditions;

[0153] After traversing each point cluster, the detection posture route is planned based on the detection posture route planning system according to the regional location distribution and individual location distribution;

[0154] A depth detection scheme is generated based on the detection constraints associated with the detection posture route and the regional position distribution.

[0155] The point cluster division conditions include:

[0156] The same point cluster contains at least two key points in the key point distribution;

[0157] and / or,

[0158] Each key point in the same point cluster has a standard point relationship between each other;

[0159] and / or,

[0160] The matching degree between the multimodal information corresponding to each key point in the same point cluster in the three-dimensional model image and the standard multimodal information is greater than or equal to the matching degree threshold.

[0161] The positioning control system for automobile body detection also includes:

[0162] Interaction modules for:

[0163] When a target robot in the robot group detects an abnormal vehicle body event, the severity of the abnormal vehicle body event is analyzed;

[0164] When the severity is greater than or equal to the severity threshold, the online expert corresponding to the abnormal vehicle body event is connected to the detection perspective of the target robot;

[0165] Based on the depth detection scheme, determine the safe parking space of the target robot within a preset time period in the future;

[0166] Based on the perspective planning template corresponding to the abnormal events of the vehicle body, multiple observation perspectives are planned according to the safe parking space;

[0167] Based on each observation perspective, generate an observation perspective list;

[0168] Displaying a list of observation perspectives to online experts;

[0169] When the online expert selects an observation perspective from the observation perspective list, the target robot is controlled to move so that the online expert accesses the selected observation perspective.

[0170] The positioning control system for automobile body detection also includes:

[0171] Correction modules for:

[0172] Based on the perspective screening condition, the target observation perspective is screened out from various observation perspectives;

[0173] Performing characterization processing on the visible picture under the target observation angle to obtain a second characterization vector;

[0174] Determining a detection intention corresponding to the second characterization vector from a detection intention library;

[0175] Based on the detection intention, the deep detection plan is modified;

[0176] Based on the revised deep detection scheme, the robot group is relay-controlled to perform deep body detection on the car in the second time period;

[0177] The viewing angle screening conditions include:

[0178] The total number of historical times that online experts select the target observation perspective is greater than or equal to the number threshold;

[0179] and / or,

[0180] The total duration of the historical perspective of the target observation perspective is greater than or equal to the first duration threshold;

[0181] and / or,

[0182] The online expert has selected the target observation perspective in the third time period and the fourth time period, and the perspective maintenance duration after selecting other observation perspectives in the fifth time period between the third time period and the fourth time period is less than or equal to the second time threshold; the fourth event segment is after the third time period.

[0183] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A positioning control method for automobile body detection, characterized in that: include: The distribution of key points located when the monitoring robot group autonomously performs preliminary body inspection of the car in the first time period; Determine the depth detection solution based on the key point distribution, individual position distribution of the robot group, and the 3D model of the car; Based on the deep detection scheme, the robot group is controlled to perform deep body detection on the car in the second time period; The second time period is after the first time period and adjacent to the first time period; The depth detection scheme is determined based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model of the car, including: Based on the point cluster division conditions, multiple point clusters are divided from the key point distribution; Traverse each point cluster in turn; During each traversal, the traversed point cluster is characterized to obtain a first characteristic vector; Determining detection knowledge corresponding to the first characterization vector from a detection knowledge base; Based on detection knowledge, determine the area search rules and detection constraints; Based on the area search rules, the area location distribution of the area to be inspected is searched on the three-dimensional model map and associated with the detection constraint conditions; After traversing each point cluster, the detection posture route is planned based on the detection posture route planning system according to the regional location distribution and individual location distribution; A depth detection scheme is generated based on the detection constraints associated with the detection posture route and the regional position distribution.

2. The positioning control method for automobile body detection according to claim 1, characterized in that: The point cluster division conditions include: The same point cluster contains at least two key points in the key point distribution; and / or, Each key point in the same point cluster has a standard point relationship between each other; and / or, The matching degree between the multimodal information corresponding to each key point in the same point cluster in the three-dimensional model image and the standard multimodal information is greater than or equal to the matching degree threshold.

3. The positioning control method for automobile body detection according to claim 1, characterized in that: Also includes: When a target robot in the robot group detects an abnormal vehicle body event, the severity of the abnormal vehicle body event is analyzed; When the severity is greater than or equal to the severity threshold, the online expert corresponding to the abnormal vehicle body event is connected to the detection perspective of the target robot; Based on the depth detection scheme, determine the safe parking space of the target robot within a preset time period in the future; Based on the perspective planning template corresponding to the abnormal events of the vehicle body, multiple observation perspectives are planned according to the safe parking space; Based on each observation perspective, generate an observation perspective list; Displaying a list of observation perspectives to online experts; When the online expert selects an observation perspective from the observation perspective list, the target robot is controlled to move so that the online expert accesses the selected observation perspective.

4. The positioning control method for automobile body detection according to claim 3, characterized in that: Also includes: Based on the perspective screening condition, the target observation perspective is screened out from various observation perspectives; Performing characterization processing on the visible picture under the target observation angle to obtain a second characterization vector; Determining a detection intention corresponding to the second characterization vector from a detection intention library; Based on the detection intention, the deep detection plan is modified; Based on the revised deep detection scheme, the robot group is relay-controlled to perform deep body detection on the car in the second time period; The viewing angle screening conditions include: The total number of historical times that online experts select the target observation perspective is greater than or equal to the number threshold; and / or, The total duration of the historical perspective of the target observation perspective is greater than or equal to the first duration threshold; and / or, The online expert has selected the target observation perspective in the third time period and the fourth time period, and the perspective retention time after selecting other observation perspectives in the fifth time period between the third time period and the fourth time period is less than or equal to the second time period threshold; the fourth time period is after the third time period.

5. A positioning control system for automobile body detection, characterized in that: include: A monitoring module, used to monitor the distribution of key points located when the robot group autonomously performs preliminary body inspection on the car in the first time period; A decision module is used to decide the depth detection scheme based on the key point distribution, the individual position distribution of the robot group and the three-dimensional model of the car; A detection module, used to control the robot group to perform a deep body detection on the car within a second time period based on the deep detection solution; The second time period is after the first time period and adjacent to the first time period; The decision module decides on a depth detection solution based on the key point distribution, the individual position distribution of the robot group, and the three-dimensional model of the car, including: Based on the point cluster division conditions, multiple point clusters are divided from the key point distribution; Traverse each point cluster in turn; During each traversal, the traversed point cluster is characterized to obtain a first characteristic vector; Determining detection knowledge corresponding to the first characterization vector from a detection knowledge base; Based on detection knowledge, determine the area search rules and detection constraints; Based on the area search rules, the area location distribution of the area to be inspected is searched on the three-dimensional model map and associated with the detection constraint conditions; After traversing each point cluster, the detection posture route is planned based on the detection posture route planning system according to the regional location distribution and individual location distribution; A depth detection scheme is generated based on the detection constraints associated with the detection posture route and the regional position distribution.

6. The positioning control system for automobile body detection according to claim 5, characterized in that: The point cluster division conditions include: The same point cluster contains at least two key points in the key point distribution; and / or, Each key point in the same point cluster has a standard point relationship between each other; and / or, The matching degree between the multimodal information corresponding to each key point in the same point cluster in the three-dimensional model image and the standard multimodal information is greater than or equal to the matching degree threshold.

7. The positioning control system for automobile body detection according to claim 5, characterized in that: Also includes: Interaction modules for: When a target robot in the robot group detects an abnormal vehicle body event, the severity of the abnormal vehicle body event is analyzed; When the severity is greater than or equal to the severity threshold, the online expert corresponding to the abnormal vehicle body event is connected to the detection perspective of the target robot; Based on the depth detection scheme, determine the safe parking space of the target robot within a preset time period in the future; Based on the perspective planning template corresponding to the abnormal events of the vehicle body, multiple observation perspectives are planned according to the safe parking space; Based on each observation perspective, generate an observation perspective list; Displaying a list of observation perspectives to online experts; When the online expert selects an observation perspective from the observation perspective list, the target robot is controlled to move so that the online expert accesses the selected observation perspective.

8. The positioning control system for automobile body detection according to claim 7, characterized in that: Also includes: Correction modules for: Based on the perspective screening condition, the target observation perspective is screened out from various observation perspectives; Performing characterization processing on the visible picture under the target observation angle to obtain a second characterization vector; Determining a detection intention corresponding to the second characterization vector from a detection intention library; Based on the detection intention, the deep detection plan is modified; Based on the revised deep detection scheme, the robot group is relay-controlled to perform deep body detection on the car in the second time period; The viewing angle screening conditions include: The total number of historical times that online experts select the target observation perspective is greater than or equal to the number threshold; and / or, The total duration of the historical perspective of the target observation perspective is greater than or equal to the first duration threshold; and / or, The online expert has selected the target observation perspective in the third time period and the fourth time period, and the perspective retention time after selecting other observation perspectives in the fifth time period between the third time period and the fourth time period is less than or equal to the second time period threshold; the fourth time period is after the third time period.

Images