Self-adaptive visual perception and autonomous navigation method for dynamic industrial environment

Abstract

The invention relates to the technical field of autonomous navigation, in particular to a dynamic industrial environment-oriented self-adaptive visual perception and autonomous navigation method, which comprises the following steps of: acquiring a three-frame skeleton joint point position sequence, judging node stability and establishing region correspondence, labeling an unstable node track, calculating shielding probability density, and labeling a shielding region. Combining the action record of the path segment with the event frequency to judge the stability, identifying the interruption segment, analyzing the path reduction trend, and marking the frozen state information. According to the method, joint point space displacement and direction change in personnel actions are extracted, a shielding trend track in a navigation area is predicted, the area in a navigation picture is labeled in advance, and the navigation picture is obtained by capturing continuous path section response state change and locking a section with a continuous deviation trend. The active identification of the path instability condition is realized, the path in the abnormal state is frozen in time, and the situation that the mobile platform falls into the abnormal stagnation state due to the abnormal narrowing of the local structure of the path is avoided.

Description

Technical Field

[0001] The present invention relates to the field of autonomous navigation technology, and in particular to an adaptive visual perception and autonomous navigation method for dynamic industrial environments. Background Art

[0002] The field of autonomous navigation technology focuses on the entire process of path planning, state estimation, obstacle avoidance control, and target guidance for mobile objects based on acquired environmental information, without external intervention or human control. This field integrates multiple technologies, including computer vision, sensor fusion, path planning algorithms, control theory, and machine learning, and is dedicated to improving the positioning accuracy, navigation stability, and response speed of autonomous systems in complex, dynamic, and unstructured environments. Typical application scenarios include service robots, unmanned vehicles, intelligent manufacturing equipment, and inspection robots. Key challenges in this field include the uncertainty of environmental mapping and perception, the fusion processing of multi-source sensor data, efficient path generation mechanisms, and the integration of dynamic obstacle avoidance and real-time control strategies.

[0003] Among them, the adaptive visual perception and autonomous navigation method for dynamic industrial environments aims to develop an autonomous navigation method with visual perception capabilities that can dynamically adapt to complex changes in industrial sites. This method dynamically models environmental states using visual information and combines it with autonomous navigation algorithms to implement path replanning and behavior adjustments. This allows robots or mobile platforms to safely and stably execute mobile tasks in industrial environments characterized by frequent human activity and frequently changing obstacles. This method's primary applications include industrial production line inspections, dynamic material handling, and autonomous scheduling in flexible manufacturing scenarios, enhancing the flexibility and intelligence of automation systems.

[0004] Traditional navigation methods fail to deeply identify the detailed changes in the continuous impact of human and environmental factors on the navigation area during continuous motion. Environmental modeling and navigation path planning rely on single-moment or short-term environmental information, failing to consider the specific impact of human motion direction, velocity fluctuations, and continuous changes in spatial position on path response stability. This leads to slow and inaccurate assessments of occlusion formation and path instability. Traditional methods only initiate path replanning or stop movement when an obstacle actually appears or the path is blocked. They fail to anticipate potential occlusion and path contraction risks caused by human motion or changes in motion, resulting in frequent emergency braking and replanning of mobile vehicles, increasing wear and tear on the mobile platform and reducing mission efficiency. Due to the lack of accurate identification and tracking of consistent responses between path segments, they are unable to proactively detect abnormal path narrowing caused by continuous instability, making it difficult for mobile platforms to recover autonomously after straying into confined spaces. This seriously impacts the operational safety and stability of industrial field navigation systems. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an adaptive visual perception and autonomous navigation method for dynamic industrial environments.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an adaptive visual perception and autonomous navigation method for dynamic industrial environments, comprising the following steps:

[0007] S1: Obtain machine vision images of industrial scenes and extract the spatial position sequence of the human skeleton joints for three consecutive frames. Classify the joints with sudden changes in direction and fluctuations in movement amplitude as unstable nodes to obtain information on the stability of the person's movements.

[0008] S2: Based on the stability information of the personnel's movements, corresponding areas in the image screen used for navigation and recognition in the industrial workshop are called, the occlusion probability density value is calculated, and areas with occlusion trends are screened and layer-labeled to obtain image occlusion prediction labeling information;

[0009] S3: Obtaining the image occlusion prediction annotation information, judging the response stability of the path segment in a dynamic scene based on the change in the unit time action trigger frequency between adjacent path segments, and classifying the path segments into stable segments and fluctuating segments based on the change trend, thereby obtaining the path segment response consistency distribution information;

[0010] S4: Call the path segment response consistency distribution information, and based on the identified fluctuation segment, when the state of the path segment continuously deviates from the stable segment and no longer recovers to the original level, it is determined to be an execution interruption segment, and the path execution state is locked to obtain a navigation path instability lock list.

[0011] As a further solution of the present invention, the personnel motion stability information includes node fluctuation distribution, skeleton structure mapping relationship, regional motion discreteness index, position sequence overlap identification and spatial stability trend type; the image occlusion prediction annotation information is specifically the occlusion position tile index, node-region interaction relationship group, occlusion occurrence time period label, visual conflict direction type and texture fill call mark; the path segment response consistency distribution information includes path segment number identification, event triggering concentration trend, action feedback offset mode, path segment dynamic level partition and action response connection relationship set; the navigation path instability lock list is specifically the path interruption segment position set, execution state deviation sequence, continuous abnormality duration interval, control instruction freeze label and time period state inconsistency evidence group.

[0012] As a further solution of the present invention, the step of obtaining the stability information of the personnel movement is specifically as follows:

[0013] S101: Acquire a machine vision image of an industrial scene, extract the three-dimensional coordinate information of the skeleton joint points from each frame, perform structural alignment on the three frames of data according to the node numbers and construct a frame sequence set, identify the three consecutive positions of the same frame under the same number, and establish a skeleton node time sequence position group;

[0014] S102: Based on the skeleton node temporal position group, the displacement vector between two adjacent frames is called to calculate the direction angle, and whether the angle meets the direction mutation threshold condition is determined. The displacement vector modulus between three frames is then differed to determine whether it meets the amplitude fluctuation range. Joint points that meet the two conditions are selected as unstable nodes to obtain a joint point dynamic fluctuation index table;

[0015] S103: Call the joint point dynamic fluctuation index table, map the physical area position of the node in the industrial scene image according to the trunk, limbs, and head partition categories to which the node belongs in the standard skeleton structure, and combine the actual positioning coordinates of the node in the navigation mapping area in the picture to establish the motion stability mapping information of the corresponding area and generate the personnel motion stability information.

[0016] As a further solution of the present invention, the step of obtaining the image occlusion prediction annotation information is specifically as follows:

[0017] S201: Based on the stability information of the human motion, a coordinate set in consecutive image frames is sequentially superimposed in a time series order, and a trajectory curve set in consecutive frames is established for each unstable node. The trajectory curve set is mapped to the corresponding navigation area index structure in the workshop image coordinate plane to obtain a node trajectory offset mapping set;

[0018] S202: Based on the node trajectory offset mapping set, obtain the intersection segment with the edge of the image structure in the two-dimensional plane, filter the trajectory set whose influence value is within the preset occlusion sensitive interval, calculate and obtain the occlusion probability density value of each trajectory, retain the trajectory in the area where the occlusion probability density value is greater than the occlusion risk judgment threshold, and generate an occlusion trend probability distribution map;

[0019] S203: Based on the occlusion trend probability distribution map, locate the actual area boundary in the image pixel coordinate block, call the layer index information of the area to assign the occlusion label, and register the current timestamp and the index number corresponding to the node trajectory in the layer cache, establish the visual occlusion identification block relationship of the area, and generate image occlusion prediction annotation information.

[0020] As a further solution of the present invention, the formula for calculating the occlusion probability density value of each trajectory is specifically:

[0021]

[0022] Among them, ρ represents the occlusion probability density value corresponding to the trajectory sequence, n represents the number of image frames involved in the calculation, represents the final trajectory direction vector of the unstable node in the i-th frame, Represents the tangent vector of the edge of the image structure corresponding to the node in the i-th frame, represents the modulus of the node trajectory vector in the i-th frame, represents the modulus of the image edge vector in the i-th frame, represents the absolute value of the dot product of the node trajectory vector and the edge vector in the i-th frame, Indicates the distance value from the node to the edge in the i-th frame, Indicates the maximum distance between the node and the edge within three frames in the i-th frame.

[0023] As a further solution of the present invention, the step of obtaining the path segment response consistency distribution information is specifically as follows:

[0024] S301: Obtain the image occlusion prediction annotation information, collect the action trigger log of each path segment during the navigation process of the mobile platform in the current industrial scene, classify and count the number of actions in each path segment in time series based on the action execution records corresponding to each path segment per unit time in the log, establish a time mapping index table, and generate a path segment action frequency set;

[0025] S302: Based on the path segment action frequency set, the number of events occurring per unit time is counted and compared with the path segment action frequency. A response change intensity value of each path segment is calculated and a state classification of the path segment in the current dynamic scene is determined based on whether the change intensity exceeds a response stability threshold, thereby obtaining a response state label for the path segment.

[0026] S303: Based on the path segment response status label, a continuous numbering sequence is established for the path segments determined to be in a fluctuating response state. The time index and spatial location information are integrated and mapped, and spatial identifiers of stable segments and fluctuating segments are constructed on the path structure layer to obtain the path segment response consistency distribution information.

[0027] As a further solution of the present invention, the formula for calculating the response change intensity value of each path segment is specifically:

[0028]

[0029] Among them, f j represents the unit time action frequency of the j-th path, f j-1 represents the unit time action frequency of the j-1th path, represents the normalized mean of the path segment action frequency, s j Indicates the j-th segment personnel movement offset value, sj-1 Indicates the personnel movement offset value of the j-1 segment, represents the normalized mean of personnel deviation in the entire path segment, e j Indicates the number of device status updates in segment j, e j-1 represents the number of device status updates in the j-1th segment, m represents the total number of path segments, and δ is the response change intensity value of the path segment.

[0030] As a further solution of the present invention, the steps for obtaining the navigation path instability lock list are specifically as follows:

[0031] S401: Calling the path segment response consistency distribution information, combining the response state values ​​of adjacent path segments in chronological order, constructing the time series of the state values ​​into a path segment state evolution trajectory sequence, and marking the consistency of continuous offset directions in the sequence to generate a path state offset trend sequence;

[0032] S402: Based on the path state deviation trend sequence, determine whether the state value of each path segment remains consistent with the average state deviation direction of the stable segment within a continuous sliding time window, calculate the deviation interval length, select path segments with deviation lengths greater than the stability persistence threshold, extract the numbers and corresponding time slice indexes, and establish a path segment continuous deviation index group;

[0033] S403: Call the path segment continuous offset index group, extract the corresponding spatial coordinate sequence information in the navigation path according to the path segment number and time slice index, mark the path segment as an execution interruption segment, and uniformly record the spatial coordinates, time node and status lock identifier in the lock list mapping table to establish a navigation path instability lock list.

[0034] As a further embodiment of the present invention, the method further comprises the following steps:

[0035] S5: calling the navigation path instability lock list to identify whether the length change between consecutive segments shows a single direction of reduction, and comparing the change trend with the obstacle density change record in the scene. If the abnormal structure contraction feature is met, adding a pause instruction mark to the path segment to obtain navigation control freeze status information;

[0036] The navigation control freeze state information includes the path segment shrinkage trend type, the environment congestion section number, the pause trigger timestamp set, the path scheduling pause signal identifier and the structure constraint strengthening label.

[0037] As a further solution of the present invention, the step of obtaining the navigation control freeze state information is specifically as follows:

[0038] S501: calling the navigation path instability lock list, extracting the adjacent path segment indexes of the path segment in the navigation path planning record based on the marked path segment coordinate sequence, calculating the Euclidean distance between the start and end nodes of the path segment and establishing a length sequence, forming a distance change sequence of continuous path segments, and generating a path segment length change trend set;

[0039] S502: Based on the path segment length change trend set, determine whether the lengths of multiple consecutive path segments show a decreasing trend in a single direction, and sequence-mark the path segments that satisfy the decreasing relationship. Perform spatial overlap calculation on the corresponding positions of the marked sequence and the obstacle distribution segment in the scene coordinate system to obtain a set of path segment numbers that meet the contraction feature, thereby obtaining a set of abnormal structure aggregated path numbers.

[0040] S503: Call the abnormal structure aggregated path number set, bind the path segment execution status in the navigation control device, set the navigation signal freeze instruction identifier within the corresponding time period, synchronize the status to the path execution scheduling table and generate a control record, and establish navigation control freeze status information.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are:

[0042] The present invention extracts the spatial displacement and directional changes of joint points during human motion, accurately identifies the degree of motion stability, predicts the trend trajectory of occlusion within the navigation area, and enables pre-annotation of areas in the navigation image. Based on the regional occlusion trend and the spatial position of nodes, the response stability of path segments is determined in real time and refined into stable and fluctuating segments. This ensures that the mobile platform clearly understands the reliability of each path segment during path selection, avoiding frequent interruptions or misoperations caused by human movement and equipment status changes during navigation. By capturing the response status changes of continuous path segments and locking segments with a continuous deviation trend, path instability is actively identified and abnormal paths are promptly frozen. Based on the Euclidean length trend of the spatial coordinate sequence and the density of obstacles, abnormal path contraction is accurately detected and intervened in advance, avoiding abnormal stagnation caused by abnormal narrowing of the local path structure. This enables the mobile platform to actively perceive and predict risks in complex and dynamic industrial environments, significantly improving the navigation system's adaptive response speed and task execution stability in the face of sudden environmental changes, reducing navigation risks caused by on-site personnel interference and equipment emergencies, and enhancing overall autonomous navigation performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0045] Figure 2 This is a schematic diagram of the refinement of S1 of the present invention;

[0046] Figure 3 This is a schematic diagram of the refinement of S2 of the present invention;

[0047] Figure 4 This is a schematic diagram of the refinement of S3 of the present invention;

[0048] Figure 5 This is a schematic diagram of the refinement of S4 of the present invention;

[0049] Figure 6 This is a schematic diagram of the refinement of S5 of the present invention. DETAILED DESCRIPTION

[0050] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0051] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0052] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0053] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0054] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0055] See also Figure 1 The present invention provides a technical solution: an adaptive visual perception and autonomous navigation method for dynamic industrial environments, comprising the following steps:

[0056] S1: Acquire machine vision images of industrial scenes and extract the spatial position sequence of the human skeleton joints for three consecutive frames. Based on the changes in the displacement direction and position movement amplitude of each joint in the three frames, the stability of the joints in the continuous movement is determined. Joints with sudden changes in direction and fluctuations in movement amplitude are classified as unstable nodes. Based on their distribution in the skeleton structure, a corresponding relationship is established with the current visual modeling area to obtain information on the stability of the human movement.

[0057] The spatial position sequence of skeleton joints corresponds to the 2D or 3D coordinate set output by human pose estimation tools such as OpenPose and MediaPipe, in the form of (X, Y) or (X, Y, Z) coordinates of each joint in each frame. The change in displacement direction can be calculated using the cosine of the angle between three frames, and directional mutations can be understood as mutation features with angles greater than π / 2 (90 degrees).

[0058] S2: Based on the stability information of human motion, the spatial position set of unstable nodes is used to call the corresponding area in the image screen used for navigation and recognition in the industrial workshop. The movement trend trajectory of the node close to the navigation area in the time series is marked in the image. The occlusion probability density value is calculated based on the intersection range of the node movement direction and the edge of the image structure. The area with occlusion trend is selected and layered to obtain the image occlusion prediction annotation information;

[0059] S3: Obtain image occlusion prediction annotation information. Based on the action trigger records of each path segment when the mobile platform performs path navigation tasks in the current industrial scenario, the change in the unit time action trigger frequency between adjacent path segments is combined with the event frequency of personnel movement and equipment status update in the navigation environment within the same period. The response stability of the path segment in the dynamic scenario is determined, and the path segment is classified into stable segments and fluctuating segments based on the change trend to obtain the path segment response consistency distribution information.

[0060] The action trigger frequency per unit time refers to the number of navigation control actions performed by the robot or automatic navigation platform within a standard time window; the event frequency refers to the time-intensive distribution of perception events such as human detection events and equipment status changes, which are usually recorded in the status log table;

[0061] S4: Call the path segment response consistency distribution information. Based on the identified fluctuation segment, compare the segment's response consistency state change trend during the execution of continuous instructions. When the state of a path segment continuously deviates from the stable segment and no longer returns to the original level, it is determined to be an execution interruption segment. The spatial coordinate sequence and time node corresponding to the interruption segment are marked, and the path execution state is locked to obtain a navigation path instability lock list.

[0062] In response to the trend of consistency state changes, a sliding window method can be used to perform differential processing on the change of unit time action frequency of continuous path segments to monitor whether the state fluctuation segment persists for a period of time without recovery, thus forming a basis for interruption identification.

[0063] S5: Call the navigation path instability lock list, and perform trend analysis on the Euclidean length records of the path segments in adjacent segments based on the marked path segment coordinate sequence, identify whether the length change between consecutive segments shows a single-direction reduction, and correspond the change trend with the obstacle density change record in the scene. If the abnormal structure contraction feature is met, add a pause instruction mark to the path segment to obtain the navigation control freeze status information.

[0064] The stability information of personnel movements includes node fluctuation distribution, skeleton structure mapping relationship, regional movement discreteness index, position sequence overlap identification and spatial stability trend type. The image occlusion prediction annotation information specifically includes occlusion position tile index, node-region interaction relationship group, occlusion occurrence period label, visual conflict direction type and texture fill call mark. The path segment response consistency distribution information includes path segment number identification, event trigger concentration trend, action feedback offset mode, path segment dynamic level partition and action response connection relationship set. The navigation path instability lock list specifically includes path interruption segment position set, execution status deviation sequence, continuous abnormality duration interval, control instruction freeze label and status inconsistency evidence group within the time period. The navigation control freeze status information includes path segment contraction trend type, environmental congestion segment number, pause trigger timestamp set, path scheduling pause signal identification and structural constraint strengthening label.

[0065] See also Figure 2 ,The specific steps for obtaining personnel motion stability information are:

[0066] S101: Acquire a machine vision image of an industrial scene, collect three consecutive frames of a person in the image, extract the three-dimensional coordinate information of the skeleton joint points from each frame, perform structural alignment on the three frames of data based on the node number, and construct a frame sequence set. Identify the three consecutive positions of the same frame under the same number, and establish a skeleton node time sequence position group.

[0067] To obtain machine vision images of industrial scenes, it is necessary to configure an RGB-D camera or a visual device with depth perception capabilities, uniformly shoot the industrial site layout including personnel activity areas, fixed equipment areas, and navigation platform path areas, continuously collect three frames of image data and align the image timestamps, use the skeleton recognition module to recognize the human posture of the image, extract the 17 standard skeleton key point coordinate data in each frame of the image, and number each node based on the COCO skeleton definition, with the head as node 0, the left shoulder as node 5, the right shoulder as node 6, and so on. The numbering is used to identify and associate the same nodes between the three frames, construct a three-dimensional coordinate sequence, and record it as a three-frame time series skeleton data structure. For example: the coordinates of node 7 in the three frames are respectively The coordinates of the corresponding frames are (1.23, 2.45, 0.56), (1.38, 2.61, 0.58), and (1.65, 2.74, 0.59). The node numbered 7 records the coordinates of the corresponding frames in sequence, and the three-frame sequence is classified into a continuous structure array with node numbered 7. This operation is repeated to traverse 17 nodes to build a complete three-frame skeleton node structure mapping. During the process, the node information extracted from each frame image is archived in frames. When aligning the structure, the number consistency is used for horizontal splicing to ensure that each node has a complete position record in the three frames. At the same time, empty nodes and missing points are filtered, missing nodes are marked, and the construction failure point set is excluded. Finally, the node time series data group under the three-frame skeleton number is established to complete the construction of the skeleton node time series position group.

[0068] S102: Based on the skeleton node temporal position group, according to the spatial coordinate sequence of each joint point between three frames, the displacement vector between two adjacent frames is called to calculate the direction angle, and whether the angle meets the direction mutation threshold condition is determined. The displacement vector modulus between the three frames is then interpolated to determine whether it meets the amplitude fluctuation range. Joint points that meet the two conditions are selected as unstable nodes, and a joint point dynamic fluctuation index table is obtained;

[0069] According to the three-frame coordinate sequence of each node number recorded in the skeleton node temporal position group, the spatial direction change analysis of the inter-frame difference in the sequence is required. First, two continuous vector segments of each node in three frames are extracted based on the node number. The displacement vector directions between frame 1 to frame 2 and frame 2 to frame 3 are calculated respectively. The direction change is evaluated by the vector angle calculation method. The formula is: in The node displacement vectors are from frame 1 to frame 2 and from frame 2 to frame 3 respectively. The angle threshold is set to 60°. If the cosine value is less than 0.5, it is determined that the direction of the node has changed suddenly. At the same time, the difference in the length of the two vector segments in the three frames is calculated. If the change amplitude is greater than 0.15 meters, it is marked as an amplitude fluctuation node. For example, the node numbered 11 moves 0.12 meters and 0.31 meters between frames, and the difference is 0.19 meters, which meets the set fluctuation threshold. The node is included in the fluctuation set. If a node meets both the conditions of sudden change in direction and amplitude fluctuation, it is recorded as an unstable node. Finally, all node numbers are traversed, and the node index numbers that meet the dual conditions are written into the index table to establish a dynamic fluctuation index table of joint points.

[0070] S103: Calling the joint point dynamic fluctuation index table, extracting the marked unstable node number, mapping the node's physical area position in the industrial scene image according to the trunk, limbs, and head partition categories to which the node belongs in the standard skeleton structure, and combining the actual positioning coordinates of the node in the navigation mapping area in the image to establish the motion stability mapping information of the corresponding area, thereby generating the person's motion stability information;

[0071] The unstable node numbers recorded in the joint dynamic fluctuation index table are called. According to the partition definition of the standard skeleton structure, the region category of each unstable node is found and divided into six major regions: head, torso, left upper limb, right upper limb, left lower limb, and right lower limb. After determining the region attribution through number matching, the corresponding position of the node is called in the industrial scene image coordinates, and the navigation mapping area grid number of the position in the current frame is read. Spatial mapping is performed to associate the partition category of the node in the skeleton structure with the position index structure in the industrial scene. If node number 11 belongs to the right upper limb and is located in area B7 grid in the image, the mapping pair is written into the stability mapping structure table, and the node time index frame number is marked to indicate its position in the time series. Finally, the spatial region position, time frame number, and structural classification information of all unstable nodes are sorted and archived to form a multidimensional mapping set with node number as index, structural category as key, and image coordinate position as value to generate personnel motion stability information.

[0072] See also Figure 3 ,The specific steps for obtaining image occlusion prediction annotation information are:

[0073] S201: Based on the stability information of the human motion, the spatial positions of the joints marked as unstable in the image are called, and the coordinate sets in the consecutive image frames are sequentially superimposed in a time series order. A trajectory curve set in the consecutive frames is established for each unstable node, and the trajectory curve set is mapped to the corresponding navigation area index structure in the workshop image coordinate plane to obtain a node trajectory offset mapping set;

[0074] Based on the stability information of personnel movements, the unstable joint point number list recorded in the joint point dynamic fluctuation index table is first called, and the skeleton point position information in the three-frame image corresponding to each number is retrieved. The three-dimensional coordinate set of each number in the three frames is obtained by node number matching. The three-frame coordinates are arranged in sequence according to the frame timestamp order and stored as a set of time-series trajectories. Each set of trajectories is then mapped to the two-dimensional image coordinate system and converted into a plane point set corresponding to the image pixel position. An independent two-dimensional trajectory curve set is established for each unstable node. In this process, if the node number is 12, its three-frame positions are (1.2, 2.3, 0.4), ( 1.3,2.4,0.4), (1.6,2.8,0.5), the trajectory can be mapped to the image coordinate points (252,326), (258,332), (270,344), forming a curved path in the pixel plane. In the three-dimensional navigation scene, a regional index binding is established from the pixel point to the actual physical position, and the corresponding navigation area code is matched. For example, the image point (258,332) corresponds to the B3 grid area in the workshop, and the point is marked with a B3 label. Finally, the continuous trajectory of all unstable nodes in the time frame is generated, and the correspondence between its image and navigation area is marked to obtain the node trajectory offset mapping set;

[0075] S202: Based on the node trajectory offset mapping set, according to the direction vector of each trajectory, the intersection segment with the edge of the image structure in the two-dimensional plane is obtained, the intersection influence value is calculated by the angle between the direction of the node end vector and the edge direction vector, the trajectory set with the influence value within the preset occlusion sensitive range is selected, the occlusion probability density value of each trajectory is calculated, and the trajectory in the area where the occlusion probability density value is greater than the occlusion risk judgment threshold is retained to generate an occlusion trend probability distribution map;

[0076] The formula for calculating the occlusion probability density value of each trajectory is as follows:

[0077]

[0078] Among them, ρ represents the occlusion probability density value corresponding to the trajectory sequence, n represents the number of image frames involved in the calculation, represents the final trajectory direction vector of the unstable node in the i-th frame, Represents the tangent vector of the edge of the image structure corresponding to the node in the i-th frame, represents the modulus of the node trajectory vector in the i-th frame, represents the modulus of the image edge vector in the i-th frame, represents the absolute value of the dot product of the node trajectory vector and the edge vector in the i-th frame, Indicates the distance value from the node to the edge in the i-th frame, represents the maximum distance between the node and the edge within three frames in the i-th frame;

[0079] The occlusion probability density value measures the likelihood of occlusion risk arising from unstable motion nodes within a continuous trajectory in a dynamic industrial environment. This value is a probabilistic parameter representing occlusion trends within a spatiotemporal image. It is used to predict areas of the image likely to be affected by occlusion. This metric is essentially the mean density of occlusion behaviors projected across the trajectory in image space and is used to construct a layer occlusion risk map.

[0080] Calculation logic description:

[0081] The formula for the occlusion probability density value ρ is:

[0082]

[0083] The meaning of each element is as follows: The moving direction vector of the unstable node of the person in the i-th frame; The tangent vector of the edge of the image structure in the i-th frame; The actual distance between the frame node and the edge; The maximum distance from the node to the edge within three frames; The dot product of the trajectory direction and the edge direction indicates the degree of angle closeness; the weight term The closer the distance is to the edge (the closer it is to occlusion), the larger the factor is; finally, the probability density value is formed by accumulating and averaging all frames.

[0084] According to the trajectory number recorded in the node trajectory offset mapping set, the last segment vector coordinates of each trajectory in three consecutive frames are extracted, and the trajectory direction vector is constructed based on the last two points. If the point of the third frame is (268,340) and the point of the second frame is (258,332), then Calculate its modulus as After querying the image structure edge database, it is found that the end point of the trajectory is near the edge of the wall, and the corresponding edge direction vector is Its module length is The dot product is Taking the absolute value, we get Then calculate the cosine ratio term as:

[0085]

[0086] Next, we calculate the distance term. We set the actual pixel distance from the node to the edge in the third frame to be 38 pixels, and the maximum distance from the edge in the three frames to be 70 pixels. After normalization, The distance term is calculated as:

[0087]

[0088] Multiplying the two results together gives:

[0089] ρ i =0.9529·0.4571≈0.4359;

[0090] If the trajectory is the first trajectory and the total number of frames is 3, and the corresponding values ​​of the remaining two frames are set to 0.3121 and 0.6380, then the occlusion probability density corresponding to the complete trajectory of the node is:

[0091]

[0092] If the occlusion threshold is set to 0.45, then ρ = 0.462 > 0.45, and the track is determined to have an occlusion trend. Its track number and image location are recorded, and its occlusion risk label is added to the occlusion trend probability distribution map. Other tracks are fed into the calculation process in the same manner, filtered, and labeled. The remaining tracks are then used as the data for the occlusion trend probability distribution map.

[0093] S203: Based on the occlusion trend probability distribution map, according to the marked high occlusion probability area trajectory position, locate the actual area boundary in the image pixel coordinate block, call the area's layer index information to assign an occlusion label, and register the current timestamp and the index number corresponding to the node trajectory in the layer cache, establish the region's visual occlusion identification block relationship, and generate image occlusion prediction annotation information;

[0094] According to the list of high occlusion trajectory numbers and their spatial coordinate ranges in the occlusion trend probability distribution map, its pixel boundary in the image space is located, and its area number is used for layer identification by matching the image area label mapping table. The area index node in the layer management structure is read, and an occlusion risk identification is added to each number. For example, the occlusion path with node trajectory number 17 is located in image area R5, and the corresponding layer index is L3. Then the risk label number 17 is added to the layer index L3, and its timestamp and trajectory number are registered at the same time. An occlusion tile identification record consisting of timestamp, layer ID, and trajectory ID is constructed and stored in the layer cache dataset. Finally, the binding annotation of spatial occlusion information in the image hierarchy structure is completed, and image occlusion prediction annotation information is generated.

[0095] See also Figure 4 ,The specific steps for obtaining the path segment response consistency distribution information are:

[0096] S301: Obtain image occlusion prediction annotation information, collect action trigger logs for each path segment during the navigation process of the mobile platform in the current industrial scene, classify and count the number of actions for each path segment in time series based on the action execution records corresponding to each path segment per unit time in the logs, establish a time mapping index table, and generate a path segment action frequency set;

[0097] After obtaining the image occlusion prediction and annotation information, it is necessary to extract the action trigger data of each path segment from the navigation system action log in the industrial scene. The action trigger data uses the timestamp as the primary key and contains the control signal records of each path segment during the navigation process, including action types such as turning, deceleration, and stopping. Clustering is performed according to the path number, and a unit time action statistical sequence is constructed. For example, if the path segment P3 triggers the action sequence [forward, turn, stop] within 3 seconds, then its action frequency per unit time is 3 times. After collecting the action frequencies of all path segments, a data table structure sorted in chronological order is established. At the same time, the time index corresponding to each segment of data is obtained for subsequent state evolution analysis. The constructed path segment action frequency set is f1, f2, ..., f m ], where m is the total number of path segments. The action frequencies of P1 to P5 are set to 5, 4, 6, 3, and 2, respectively. Then this set is [5, 4, 6, 3, 2], which is used to support subsequent change trend analysis and response state judgment, and finally generate the path segment action frequency set;

[0098] S302: Based on the path segment motion frequency set, the motion frequency data of adjacent path segments is selected. The continuous position offset and device status update sequence in the personnel movement record are called. The number of events occurring per unit time is counted and compared with the path segment motion frequency. The response change intensity value of each path segment is calculated and calculated. Based on whether the change intensity exceeds the response stability threshold, the state classification of the path segment in the current dynamic scene is determined, and the response state label of the path segment is obtained.

[0099] The formula for calculating the response change intensity value of each path segment is as follows:

[0100]

[0101] Among them, f j represents the unit time action frequency of the j-th path, f j-1 represents the unit time action frequency of the j-1th path, represents the normalized mean of the path segment action frequency, s j Indicates the j-th segment personnel movement offset value, s j-1 Indicates the personnel movement offset value of the j-1 segment, represents the normalized mean of personnel deviation in the entire path segment, e j Indicates the number of device status updates in segment j, e j-1 represents the number of device status updates in the j-1th segment, m represents the total number of path segments, and δ is the response change intensity value of the path segment;

[0102] The response change intensity value is a temporal variability measurement parameter used in dynamic path analysis to measure the overall fluctuation amplitude of action feedback, personnel deviation behavior, and equipment status changes in different path segments per unit time. This value is a key intermediate quantity for evaluating the stability of path segments in dynamic industrial scenarios. It is used to divide path segments into stable segments and fluctuating segments. Its essence is to evaluate the fusion difference intensity of the changing trends of multi-source event characteristics, including changes in action trigger frequency, personnel deviation, and equipment response status.

[0103] Calculation logic description:

[0104] The formula for the response change intensity value δ is:

[0105]

[0106] The definitions of each item are as follows: j : The unit time action trigger frequency of the jth segment; s j : Personnel offset value of the jth segment; e j : The number of device status updates in segment j; The normalized average value of the corresponding indicator; each sub-item measures the intensity of three core behaviors between adjacent path segments: behavior frequency fluctuation, human movement stability fluctuation, and equipment state response fluctuation; all items are normalized and differentially expressed, which has the effect of high-pass filtering to amplify "mutations" and effectively capture path instability signals. The model integrates three common characteristic indicators (behavior frequency, displacement, and equipment events) to form a unified intensity expression, which is a reasonable and commonly used fusion metric in industrial dynamic path response models.

[0107] Based on the path segment action frequency set, continuous path segments are selected, and the action frequency values ​​of adjacent path segments are extracted as frequency change items. At the same time, the coordinate displacement difference in the personnel movement record is obtained as the offset item, and the statistical value of the equipment alarm, start and stop and other change behaviors in the equipment status log is obtained as the status update item. The three types of data are brought into the occlusion scene for joint analysis. For example, the action frequency from the second segment to the third segment is 4 to 6, the corresponding frequency difference is 2, the personnel offset is from (1.2, 0.9) to (1.9, 1.0), the Euclidean distance is 0.707 meters, and the equipment status change record is 2 items. The normalized mean of the path segment action frequency is set Normalized mean of personnel deviation Enter the formula to calculate:

[0108]

[0109] Calculated:

[0110] Finally, δ≈0.3846+0.3847+0.6667=1.436. If the response stability threshold is set to 1.1, the path segment state is classified as a fluctuating segment, marked as an abnormal state trajectory segment, and the path segment response state label is constructed;

[0111] S303: Based on the path segment response status labels, a continuous numbering sequence is established for the path segments determined to be in a fluctuating response state. Time indexes and spatial location information are integrated and mapped, and spatial identifiers of stable and fluctuating segments are constructed on the path structure layer to obtain path segment response consistency distribution information.

[0112] According to the list of path segment numbers marked as fluctuating states in the path segment response status label, they are consecutively numbered according to the timestamp and path segment number, and the start and end coordinates of the path segment corresponding to the number on the navigation path are extracted. The path segment position dataset in the scene coordinate system is established, and spatial structure layer mapping is performed in the order of numbers. The stable segment and the fluctuating segment are written into the layer mapping table with different label numbers. For example, the stable segment is marked as S, and the fluctuating segment is marked as W. Path segments P1-P2 are stable, and P3-P4 are fluctuating. Then, a label sequence [S, S, W, W, S] is generated and written into the corresponding coordinate structure layer to realize layer state partition mapping, and finally the path segment response consistency distribution information is obtained.

[0113] See also Figure 5 The specific steps for obtaining the navigation path instability lock list are as follows:

[0114] S401: Retrieving the path segment response consistency distribution information, combining the response state values ​​of adjacent path segments in chronological order according to the path segment numbers marked as fluctuating states, constructing the time series of state values ​​into a path segment state evolution trajectory sequence, and marking the consistency of continuous offset directions in the sequence to generate a path state offset trend sequence;

[0115] The path segment numbers marked as fluctuating in the path segment response consistency distribution information are used. First, a numbering sequence is established and the path segments are sorted in ascending timestamp order. The response state values ​​of each path segment during time evolution are extracted to construct a state time series of consecutive path segments. For example, if the path segments are numbered P3, P4, and P5, and their state values ​​are 0.6, 0.65, and 0.7, respectively, a state evolution sequence of [0.6, 0.65, 0.7] can be constructed. A shift direction consistency check is performed in the sequence to determine whether the signs of the differences between two adjacent state values ​​are the same. If the signs remain consistent, the shift directions are considered consistent. For example, in the above sequence, each incremental state value maintains a consistent direction, which is recorded as a positive shift. If 0.68, 0.64, and 0.60 appear subsequently, the direction changes continuously to a negative shift, which is marked as a reverse shift state. This logic is used to mark the continuous shift segments of the entire state time series and extract the segment start and end indexes. Finally, the path numbers and time slice information of the consecutive segments with consistent shift directions are written into the sequence structure to generate a path state shift trend sequence.

[0116] S402: Based on the path state deviation trend sequence, determine whether the state value of each path segment remains consistent with the average state deviation direction of the stable segment within the continuous sliding time window, calculate the deviation interval length, select path segments with deviation lengths greater than the stability persistence threshold, extract the numbers and corresponding time slice indexes, and establish a path segment continuous deviation index group;

[0117] Based on the path state deviation trend sequence, the fluctuation interval of each path segment state in the sliding window is judged. The sliding window width is set to 3. The direction of the difference between the current state of each path segment and the mean state of the stable segment in the sliding window is traversed to see if it is consistent. If the direction is consistent, it is judged that the deviation direction is maintained. The length of the continuous direction consistent interval is counted as the deviation interval length. The path segment state value is set to [0.42, 0.48, 0.55, 0.61, 0.64] and the mean of the stable segment is 0.40. The difference between the state value of each segment and the mean is [+0.02, +0.08, +0.15, +0.21, +0.24], all in the positive direction. The continuous deviation length is 5. The stability continuity threshold is set to 3. Since 5 is greater than the threshold, the path segment sequence is judged to be a continuous deviation state path segment. Its number, start frame index, end frame index, etc. are recorded in the structure list to construct a path segment continuous deviation index group;

[0118] S403: Calling the path segment continuous offset index group, extracting the corresponding spatial coordinate sequence information in the navigation path based on the path segment number and time slice index, marking the path segment as an execution interruption segment, and recording the spatial coordinates, time node, and state lock identifier in a lock list mapping table to establish a navigation path instability lock list;

[0119] The path segment number and time index extracted from the path segment continuous offset index group are called, and the spatial coordinate positions of the starting and ending points of the path segment are retrieved in combination with the navigation path structure information table. The segments are classified and recorded as continuous unstable segments according to the numbers, and the task scheduling records corresponding to their time slices in the control log are read. The task execution number, path segment execution identifier, operation node and other contents of the segment are extracted and synchronously integrated, and the segment path is marked as an execution interruption segment. At the same time, a locking status identifier is established and bound to the path scheduling module. Finally, the four types of data, including the path segment number, coordinate information, time segment index and status locking identifier, are summarized and written into the locking information structure table to establish a navigation path instability locking list.

[0120] See also Figure 6 , the steps for obtaining navigation control freeze status information are as follows:

[0121] S501: Calling the navigation path instability lock list, extracting the adjacent path segment indexes of the path segment in the navigation path planning record based on the marked path segment coordinate sequence, calculating the Euclidean distance between the start and end nodes of the path segment and establishing a length sequence, forming a distance change sequence of continuous path segments, and generating a path segment length change trend set;

[0122] Call the marked path segment coordinate sequence in the navigation path instability lock list, first extract the starting and ending coordinate points of each path segment, and construct a three-dimensional coordinate representation (x s ,y s ,z s )、(x e ,y e ,z e ), call the upper and lower path segment numbers adjacent to the path segment in the path planning record file, obtain the corresponding spatial position point pair through the path segment index, combine the starting and ending points of each path segment into two pairs as the distance calculation input, and use the Euclidean formula to calculate the length value. That is, if the starting point of path segment P1 is (2.0, 1.0, 0.0) and the end point is (3.0, 2.0, 0.0), then its length is In this way, all path segments are traversed and a length sequence set is constructed. For example, if the lengths of consecutive segments P1, P2, and P3 are 1.414, 1.322, and 1.220, they are recorded as [1.414, 1.322, 1.220], and finally a set of path segment length change trends is generated.

[0123] S502: Based on the path segment length change trend set, the length differences of adjacent path segments are called to determine whether the lengths of multiple consecutive path segments show a decreasing trend in a single direction. Path segments that satisfy the decreasing relationship are sequence-marked. Obstacle density change records are called, and spatial overlap calculations are performed on the corresponding positions of the marked sequence with the obstacle distribution segments in the scene coordinate system to obtain a set of path segment numbers that meet the contraction feature, thereby obtaining a set of abnormal structure aggregated path numbers.

[0124] Based on the length values ​​of every two adjacent path segments in the path segment length change trend set, calculate their difference sequence. For example, the difference between P1 and P2 is 1.4141.322=0.092, and the difference between P2 and P3 is 1.3221.220=0.102. Determine whether the sequence is in a continuously decreasing relationship, that is, determine whether all differences are positive and whether they continuously exceed the set judgment quantity threshold. Set the continuous decreasing threshold to 2. If two consecutive differences are greater than 0, the decreasing trend is considered valid. Further call the obstacle density change record file to extract the obstacle number density data corresponding to the spatial range of the path segment in the scene coordinates. For example, the average number of obstacles in the area covered by path segment P2 is 15 per square meter. If the density threshold is 12, then this area meets the obstacle density judgment standard. Perform overlap judgment, that is, cross-calculate the path segment coordinate range with the obstacle distribution grid. If the overlapping area ratio exceeds 50%, the path segment is determined to be an abnormal shrinkage area, and its path segment number is extracted and added to the abnormal structure aggregation path number set.

[0125] S503: Calling the abnormal structure aggregated path number set, binding the path segment execution status in the navigation control device, and setting the navigation signal freeze instruction flag within the corresponding time period, synchronizing the status to the path execution schedule and generating a control record, and establishing navigation control freeze status information;

[0126] The number list extracted from the abnormal structure aggregation path number set is called, and the execution control dictionary in the navigation control module is accessed one by one. The execution status field of each path segment is set to "frozen", and a freezing time range field is added. The range is taken from the timestamp index in the instability lock list. At the same time, a freezing task record is added to the path scheduling table, which contains metadata such as the path segment number, freezing time, freezing reason identifier, instruction lock number, etc., and is associated with the scheduling flow chart node. Once the path segment is reached during task execution, the task enters the freezing waiting state. The freezing signal is recorded as FZ state in the control record, and the binding relationship between this state and the path segment and the timestamp information are recorded. Finally, the navigation control freezing state information is established.

[0127] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0128] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0129] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0130] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0131] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0132] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0133] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0134] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0135] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0136] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. Adaptive visual perception and autonomous navigation method for dynamic industrial environments, characterized by: The following steps are involved: S1: Obtain machine vision images of industrial scenes and extract the spatial position sequence of the human skeleton joints for three consecutive frames. Classify the joints with sudden changes in direction and fluctuations in movement amplitude as unstable nodes to obtain information on the stability of the person's movements. S2: Based on the stability information of the personnel's movements, corresponding areas in the image screen used for navigation and recognition in the industrial workshop are called, the occlusion probability density value is calculated, and areas with occlusion trends are screened and layer-labeled to obtain image occlusion prediction labeling information; S3: Obtaining the image occlusion prediction annotation information, judging the response stability of the path segment in a dynamic scene based on the change in the unit time action trigger frequency between adjacent path segments, and classifying the path segments into stable segments and fluctuating segments based on the change trend, thereby obtaining the path segment response consistency distribution information; S4: Call the path segment response consistency distribution information, and based on the identified fluctuation segment, when the state of the path segment continuously deviates from the stable segment and no longer recovers to the original level, it is determined to be an execution interruption segment, and the path execution state is locked to obtain a navigation path instability lock list.

2. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 1 is characterized in that: The personnel motion stability information includes node fluctuation distribution, skeleton structure mapping relationship, regional motion discreteness index, position sequence overlap identification and spatial stability trend type; the image occlusion prediction annotation information specifically includes occlusion position tile index, node-region interaction relationship group, occlusion occurrence time period label, visual conflict direction type and texture fill call mark; the path segment response consistency distribution information includes path segment number identification, event triggering concentration trend, action feedback offset mode, path segment dynamic level partition and action response connection relationship set; the navigation path instability lock list specifically includes path interruption segment position set, execution status deviation sequence, continuous abnormality duration interval, control instruction freeze label and status inconsistency evidence group within time period.

3. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 2 is characterized in that: The steps for obtaining the personnel motion stability information are specifically as follows: S101: Acquire a machine vision image of an industrial scene, extract the three-dimensional coordinate information of the skeleton joint points from each frame, perform structural alignment on the three frames of data according to the node numbers and construct a frame sequence set, identify the three consecutive positions of the same frame under the same number, and establish a skeleton node time sequence position group; S102: Based on the skeleton node temporal position group, the displacement vector between two adjacent frames is called to calculate the direction angle, and whether the angle meets the direction mutation threshold condition is determined. The displacement vector modulus between three frames is then differed to determine whether it meets the amplitude fluctuation range. Joint points that meet the two conditions are selected as unstable nodes to obtain a joint point dynamic fluctuation index table; S103: Call the joint point dynamic fluctuation index table, map the physical area position of the node in the industrial scene image according to the trunk, limbs, and head partition categories to which the node belongs in the standard skeleton structure, and combine the actual positioning coordinates of the node in the navigation mapping area in the picture to establish the motion stability mapping information of the corresponding area and generate the personnel motion stability information.

4. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 3 is characterized in that: The steps for obtaining the image occlusion prediction annotation information are specifically as follows: S201: Based on the stability information of the human motion, a coordinate set in consecutive image frames is sequentially superimposed in a time series order, and a trajectory curve set in consecutive frames is established for each unstable node. The trajectory curve set is mapped to the corresponding navigation area index structure in the workshop image coordinate plane to obtain a node trajectory offset mapping set; S202: Based on the node trajectory offset mapping set, obtain the intersection segment with the edge of the image structure in the two-dimensional plane, filter the trajectory set whose influence value is within the preset occlusion sensitive interval, calculate and obtain the occlusion probability density value of each trajectory, retain the trajectory in the area where the occlusion probability density value is greater than the occlusion risk judgment threshold, and generate an occlusion trend probability distribution map; S203: Based on the occlusion trend probability distribution map, locate the actual area boundary in the image pixel coordinate block, call the layer index information of the area to assign the occlusion label, and register the current timestamp and the index number corresponding to the node trajectory in the layer cache, establish the visual occlusion identification block relationship of the area, and generate image occlusion prediction annotation information.

5. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 4 is characterized in that: The formula for calculating the occlusion probability density value of each trajectory is as follows: Among them, ρ represents the occlusion probability density value corresponding to the trajectory sequence, n represents the number of image frames involved in the calculation, represents the final trajectory direction vector of the unstable node in the i-th frame, Represents the tangent vector of the edge of the image structure corresponding to the node in the i-th frame, represents the modulus of the node trajectory vector in the i-th frame, represents the modulus of the image edge vector in the i-th frame, represents the absolute value of the dot product of the node trajectory vector and the edge vector in the i-th frame, Indicates the distance value from the node to the edge in the i-th frame, Indicates the maximum distance between the node and the edge within three frames in the i-th frame.

6. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 5 is characterized in that: The steps for obtaining the path segment response consistency distribution information are specifically as follows: S301: Obtain the image occlusion prediction annotation information, collect the action trigger log of each path segment during the navigation process of the mobile platform in the current industrial scene, classify and count the number of actions in each path segment in time series based on the action execution records corresponding to each path segment per unit time in the log, establish a time mapping index table, and generate a path segment action frequency set; S302: Based on the path segment action frequency set, the number of events occurring per unit time is counted and compared with the path segment action frequency. A response change intensity value of each path segment is calculated and a state classification of the path segment in the current dynamic scene is determined based on whether the change intensity exceeds a response stability threshold, thereby obtaining a response state label for the path segment. S303: Based on the path segment response status label, a continuous numbering sequence is established for the path segments determined to be in a fluctuating response state. The time index and spatial location information are integrated and mapped, and spatial identifiers of stable segments and fluctuating segments are constructed on the path structure layer to obtain the path segment response consistency distribution information.

7. According to the adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 6, the formula for calculating the response change intensity value of each path segment is specifically: in, f j represents the unit time action frequency of the j-th path, f j-1 represents the unit time action frequency of the j-1th path, represents the normalized mean of the path segment action frequency, s j Indicates the j-th segment personnel movement offset value, s j-1 Indicates the personnel movement offset value of the j-1 segment, represents the normalized mean of personnel deviation in the entire path segment, e j Indicates the number of device status updates in segment j, e j-1 represents the number of device status updates in the j-1th segment, m represents the total number of path segments, and δ is the response change intensity value of the path segment.

8. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 7 is characterized in that: The steps for obtaining the navigation path instability lock list are as follows: S401: Calling the path segment response consistency distribution information, combining the response state values ​​of adjacent path segments in chronological order, constructing the time series of the state values ​​into a path segment state evolution trajectory sequence, and marking the consistency of continuous offset directions in the sequence to generate a path state offset trend sequence; S402: Based on the path state deviation trend sequence, determine whether the state value of each path segment remains consistent with the average state deviation direction of the stable segment within a continuous sliding time window, calculate the deviation interval length, select path segments with deviation lengths greater than the stability persistence threshold, extract the numbers and corresponding time slice indexes, and establish a path segment continuous deviation index group; S403: Call the path segment continuous offset index group, extract the corresponding spatial coordinate sequence information in the navigation path according to the path segment number and time slice index, mark the path segment as an execution interruption segment, and uniformly record the spatial coordinates, time node and status lock identifier in the lock list mapping table to establish a navigation path instability lock list.

9. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 8, characterized in that: The method further comprises the following steps: S5: calling the navigation path instability lock list to identify whether the length change between consecutive segments shows a single direction of reduction, and comparing the change trend with the obstacle density change record in the scene. If the abnormal structure contraction feature is met, adding a pause instruction mark to the path segment to obtain navigation control freeze status information; The navigation control freeze state information includes a path segment shrinkage trend type, an environment congestion section number, a pause trigger timestamp set, a path scheduling pause signal identifier, and a structure constraint strengthening label.

10. The adaptive visual perception and autonomous navigation method for dynamic industrial environments according to claim 9, characterized in that: The steps for obtaining the navigation control freezing state information are specifically as follows: S501: calling the navigation path instability lock list, extracting the adjacent path segment indexes of the path segment in the navigation path planning record based on the marked path segment coordinate sequence, calculating the Euclidean distance between the start and end nodes of the path segment and establishing a length sequence, forming a distance change sequence of continuous path segments, and generating a path segment length change trend set; S502: Based on the path segment length change trend set, determine whether the lengths of multiple consecutive path segments show a decreasing trend in a single direction, and sequence-mark the path segments that satisfy the decreasing relationship. Perform spatial overlap calculation on the corresponding positions of the marked sequence and the obstacle distribution segment in the scene coordinate system to obtain a set of path segment numbers that meet the contraction feature, thereby obtaining a set of abnormal structure aggregated path numbers. S503: Call the abnormal structure aggregated path number set, bind the path segment execution status in the navigation control device, set the navigation signal freeze instruction identifier within the corresponding time period, synchronize the status to the path execution scheduling table and generate a control record, and establish navigation control freeze status information.

Images