Multi-robot cooperative operation route planning generation method and system

Abstract

The application discloses a multi-robot cooperative operation route planning generation method and system, relates to the technical field of agricultural intelligent equipment, and comprises the following steps: collecting fusion perception output, time record and running rhythm information of a multi-robot operation site, generating a fusion observation record, and marking time synchronization information at the end of the fusion observation record; according to the time synchronization information, the fusion observation record is subjected to time segmentation and comparison, time segments in which radar echo intensity suddenly increases and image brightness changes simultaneously are detected, an echo superposition attention list is generated, and a trigger boundary position is marked at the end of the echo superposition attention list. The application realizes unified management of multi-robot perception information through a fusion observation record and a time synchronization mechanism, accurately identifies abnormalities caused by environmental interference, and avoids false triggering of emergency stop; and through a rhythm disturbed position table and dynamic running adjustment, operation rhythm is adaptively recovered, queue balance is maintained, and cooperative operation efficiency and weeding accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of agricultural intelligent equipment technology, specifically to a method and system for generating routes for multi-robot collaborative operations. Background Technology

[0002] Multi-robot collaborative route planning and generation refers to a method used in agricultural weeding operations where multiple weeding robots work simultaneously in the same field. This involves a data acquisition and control system that synchronously collects and models multi-source information, including terrain features, crop distribution, weed density, obstacle locations, and work boundaries. The system utilizes path allocation, task partitioning, and dynamic coordination strategies to automatically generate independent yet collaborative work routes for each robot. This ensures that each robot efficiently covers the entire weeding area while avoiding duplication and mutual interference. This method combines global planning and local obstacle avoidance algorithms, relying on real-time perception data provided by the data acquisition and control system to dynamically adjust the work paths. This achieves a balance of workload among robots and optimal path allocation, thereby ensuring the overall efficiency and accuracy of farmland weeding operations.

[0003] The existing technology has the following shortcomings:

[0004] In existing technologies, agricultural weeding robots typically employ a fusion of radar and vision perception to dynamically identify the working environment. However, in multi-robot collaborative operations, this fusion process is susceptible to transient environmental interference. When wind causes plants to sway rapidly, wetland reflections cause light spots to flicker, or adjacent robots' radar waves overlap briefly, the perception system may generate abnormal superimposed signals during the radar and vision data fusion process, mistakenly identifying obstacles ahead. Such misjudgments can easily trigger emergency braking, causing the robot to stop abruptly, disrupting the work rhythm, decoupling the collaborative path, and even causing confusion in subsequent robots' obstacle avoidance. In severe cases, it can lead to disorder in the multi-robot work sequence and a decrease in coverage, making it difficult to guarantee the continuity and stability of the weeding process.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for generating routes for multi-robot collaborative operations, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for generating routes for multi-robot collaborative operations, comprising the following steps:

[0008] Collect fusion perception output, time records and operation rhythm information of multi-robot operation site, generate fusion observation record, and mark time synchronization information at the end of the fusion observation record;

[0009] Based on the time synchronization information, the fused observation records are compared in time segments to detect time segments in which a sudden increase in radar echo intensity and a change in image brightness occur simultaneously. An echo overlay attention list is generated, and the trigger boundary position is marked at the end of the echo overlay attention list.

[0010] Based on the trigger boundary, the initial segment of the accidental emergency stop is located and fused with the attention list superimposed along the echo. The continuous changes in travel speed and avoidance actions before and after the initial segment are extracted to generate the emergency stop trajectory. The impact time window is marked at the end of the emergency stop trajectory.

[0011] Based on the impact time window, the coordinated queue operation status corresponding to the sudden stop change trajectory is traced back, the position of queue spacing contraction and the position of change in the order of avoidance actions are identified, a rhythm disturbance position table is generated, and adjustment anchor points are marked at the end of the rhythm disturbance position table.

[0012] Based on the adjustment anchor point, dynamic operation adjustment is performed on the rhythm disturbance location table. Segmented deceleration control is performed in the trigger boundary area, short pause transition control is performed in the impact time window, and turn-by-turn traffic control is performed on the avoidance action sequence. The adjusted operation rhythm information is then updated to the fusion observation record.

[0013] Preferably, the time synchronization information annotation steps are as follows:

[0014] Within the multi-robot collaborative operation area, the fusion perception output of each robot is collected, and the source, sampling time and spatial location information of the perception data are integrated, and radar echo information, visual brightness information and operation rhythm information are recorded.

[0015] After the data collection is completed, the perception data of each robot is integrated and arranged in chronological order, aligned with a unified time stamp, forming an aggregated sequence of perception information from multiple robots, and spatiotemporal annotations are inserted according to the intervals between time stamps.

[0016] After obtaining the aggregated sequence, the radar echo intensity sequence, visual brightness sequence and operation rhythm information are superimposed in a unified time order to form a continuous fused observation record.

[0017] Time synchronization information is annotated at the end of the fused observation record. A time mapping is established by matching the time label of the sensing output with the global operation time, so that the sensing data of multiple robots can be aligned under the same time reference system.

[0018] Preferably, the steps for triggering boundary location annotation are as follows:

[0019] Based on the time synchronization information in the fusion observation records, the fusion observation records are segmented and organized in chronological order, dividing the fusion observation records into continuous and interconnected time segments, and merging the multi-robot perception records under the same time reference.

[0020] The sensory feature comparison detection is performed on the segmented and sorted fused observation records. Radar echo intensity data and image brightness change data are extracted in each time slice to match time segments that show both sudden increases and jumps.

[0021] The detected time segments are sequentially integrated to form an echo overlay attention list containing interference intervals, and the start and end times of each time segment are marked.

[0022] Based on the start and end times in the echo overlay attention list, the radar echo change rate and image brightness change rate at both ends of the segment are compared, the trigger boundary position is marked, and the time interval of the interference is recorded.

[0023] Preferably, when marking the trigger boundary position, a two-way boundary analysis is performed on each time segment in the echo overlay attention list. When the radar echo intensity continues to rise at the starting boundary and the image brightness change rate increases synchronously, it is determined as the starting end of the trigger boundary. When the radar echo intensity recovers to the reference level at the ending boundary and the image brightness change rate tends to stabilize, it is determined as the ending end of the trigger boundary. The trigger boundary position is then uniformly recorded at the end of the echo overlay attention list.

[0024] Preferably, the impact time window annotation steps are as follows:

[0025] Using the trigger boundary as a time reference, the recording interval containing the time period of the trigger boundary is selected from the echo overlay attention list, and expanded in the forward and backward directions along the time sequence to construct an analysis time band covering the entire emergency stop process. Radar echo information, image brightness change information and operation rhythm information that match the analysis time band are extracted.

[0026] Based on the trigger boundary position in the echo superposition attention list, the change trend of the sensing data within the analysis time band is compared longitudinally to determine the time point when the radar echo intensity increases and the operating speed suddenly drops, thus locating the starting segment of the fusion sensing accidental emergency stop.

[0027] Extract continuous speed change and avoidance action change data in the forward and backward directions around the initial segment of the emergency stop to generate the emergency stop trajectory and keep it synchronized with the time axis;

[0028] The impact time window is determined based on the speed recovery trend and the stabilization position of the avoidance maneuver in the sudden stop change trajectory, and the start and end times of the impact time window are marked at the end of the sudden stop change trajectory.

[0029] Preferably, the emergency stop trajectory includes an initial speed reduction segment, an emergency stop response segment, and a recovery segment. Each segment is spliced ​​together continuously in chronological order. The starting point of the impact time window is when the travel speed drops to its lowest point, and the ending point is when the travel speed and avoidance direction return to a stable state. The emergency stop trajectory and the impact time window correspond in the time dimension to reflect the continuous impact of accidental emergency stop on the work rhythm.

[0030] Preferably, the steps for adjusting anchor point annotations are as follows:

[0031] Using the impact time window marked in the sudden stop trajectory as the time reference, the operation status data of multi-robot collaborative operation is analyzed by time backtracking. The travel speed, position coordinates, queue number, avoidance direction and operation rhythm information of each robot in the time interval are extracted and arranged in order according to a unified time label.

[0032] Based on the backtracking analysis results, the queue spacing of each robot is dynamically analyzed to identify the position where the queue spacing shrinks due to the decrease in speed of the robot in front, and the start and end times of the continuous queue shrinkage intervals are recorded.

[0033] After identifying the position where the queue spacing shrinks, the changes in the sequence of avoidance actions within the same time range are analyzed to determine the time position of avoidance actions that are advanced, have a sudden change in direction, or have a change in amplitude, and these are correlated with the position where the queue spacing shrinks.

[0034] By integrating the positions of queue spacing contraction and the positions of avoidance action sequence changes, a rhythm disturbance position table is generated, and adjustment anchor points are marked at the end of the rhythm disturbance position table according to the position of the rhythm recovery turning point.

[0035] Preferably, the adjustment anchor points in the rhythm disturbance location table are determined based on the time position when the queue spacing recovers to a stable value and the avoidance action returns to the predetermined direction. The adjustment anchor points are used to identify the turning point of rhythm recovery and provide timing references for segmented control and rhythm balance in dynamic operation adjustment.

[0036] Preferably, the following steps are taken: dynamically adjusting the rhythm disturbance location table based on the adjustment anchor point, performing segmented deceleration control in the trigger boundary area, short pause transition control in the impact time window, and rotating traffic control for the avoidance action sequence, and writing the updated running rhythm information into the fusion observation record:

[0037] Using the adjustment anchor points in the rhythm disturbance location table as time references, the disturbance recovery preparation segment and the rhythm reconstruction segment are divided, and the running speed curves, position change trajectories and avoidance action directions of each robot are extracted to establish a dynamic adjustment range.

[0038] Segmented deceleration control is implemented for the trigger boundary area, with the first interval set as the deceleration segment, the middle interval set as the speed maintenance segment, and the last interval set as the acceleration segment to balance the running rhythm.

[0039] During the impact time window, a short pause transition control is executed to make each robot pause at a unified time point and resume operation sequentially according to the time interval;

[0040] The sequence of avoidance actions is controlled by taking turns, and the updated operation rhythm information is added to the fusion observation record based on the time synchronization information.

[0041] A multi-robot collaborative operation route planning and generation system includes a fusion observation record generation module, an echo superposition detection module, an emergency stop trajectory extraction module, a rhythm disturbance analysis module, and a dynamic operation adjustment module.

[0042] The fusion observation record generation module collects fusion perception output, time records and operation rhythm information of multiple robots in the operation site, generates fusion observation records, and marks time synchronization information at the end of the fusion observation records;

[0043] The echo overlay detection module compares the fused observation records in time segments based on time synchronization information, detects time segments in which a sudden increase in radar echo intensity and a change in image brightness occur simultaneously, generates an echo overlay attention list, and marks the trigger boundary position at the end of the echo overlay attention list.

[0044] The emergency stop trajectory extraction module locates and fuses the initial segment of accidental emergency stop based on the trigger boundary and the echo superimposed on the attention list. It extracts the continuous changes in travel speed and avoidance actions before and after the initial segment, generates the emergency stop trajectory, and marks the impact time window at the end of the emergency stop trajectory.

[0045] The rhythm disturbance analysis module traces back the coordinated queue operation status corresponding to the sudden stop change trajectory based on the impact time window, identifies the position of queue spacing contraction and the position of change in the order of avoidance actions, generates a rhythm disturbance position table, and marks the adjustment anchor point at the end of the rhythm disturbance position table.

[0046] The dynamic operation adjustment module performs dynamic operation adjustment based on the adjustment anchor point and the rhythm disturbance position table. It performs segmented deceleration control in the trigger boundary area, short pause transition control in the impact time window, and turn-by-turn traffic control for the avoidance action sequence. The adjusted operation rhythm information is then updated to the fusion observation record.

[0047] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0048] This invention introduces a fusion observation recording and time synchronization mechanism during multi-robot collaborative operations to achieve unified management of radar perception, visual perception, and operational rhythm information. This enables all robots to perform data alignment and perception fusion under the same time reference. By detecting synchronized segments of sudden increases in radar echo intensity and changes in image brightness during time segment comparison, it can quickly identify perception anomalies caused by transient environmental disturbances, effectively avoiding emergency stops triggered by perception misjudgments. This maintains operational continuity and path coordination stability, allowing multiple robots to maintain an efficient collaborative operational rhythm even in complex farmland environments.

[0049] This invention establishes a rhythm disturbance location table based on the impact time window and performs dynamic operation adjustments, enabling multiple robots to automatically restore their operational order after being affected by accidental sensor touches. Through adjustments such as segmented deceleration, short pauses, and turn-by-turn movement, the collaborative queue achieves adaptive recovery at the temporal level, restoring balance to queue spacing and avoidance order. This method improves the continuity and stability of the multi-robot operation rhythm, reduces coverage blind spots and repetitive tasks caused by sudden stops, and significantly enhances the overall efficiency of collaborative operations and the accuracy of weeding in farmland. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0051] Figure 1 This is a flowchart of the multi-robot collaborative operation route planning and generation method of the present invention.

[0052] Figure 2 This is a schematic diagram of the modules of the multi-robot collaborative operation route planning and generation system of the present invention. Detailed Implementation

[0053] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0054] This invention provides, for example Figure 1 The multi-robot collaborative operation route planning and generation method shown includes the following steps:

[0055] Collect fusion perception output, time records and operation rhythm information of multi-robot operation site, generate fusion observation record, and mark time synchronization information at the end of the fusion observation record;

[0056] In multi-robot collaborative operations, a method for generating fused observation records is adopted to achieve unified management of environmental perception data and temporal alignment of multi-source information. This method continuously collects, organizes, aligns, and labels multi-source perception data from the work site to establish a unified temporal correlation data structure, laying the foundation for subsequent echo detection and dynamic adjustment. The specific implementation steps are as follows:

[0057] Within the multi-robot collaborative operation area, each robot continuously collects fused perception outputs from the work site, integrating the source, sampling time, and spatial location information of the perception data. Each robot possesses radar perception, visual perception, and operational status perception capabilities. During operation, the radar outputs distance echo information and obstacle reflection intensity, the visual outputs image brightness information and texture distribution data, and the operational status perception outputs the robot's running speed, turning angle, work rhythm, and motion trajectory information. All perception data are continuously recorded in time-stamped order during acquisition, ensuring a temporal correspondence between radar echoes, visual images, and operational status at the same moment. To guarantee the consistency of multi-robot perception data across time, each robot initializes its time base before starting operation, allowing its time stamps to grow synchronously across the entire scene. In this way, the fused perception output not only covers environmental elements but also includes the robot's own motion parameters, providing a complete data foundation for subsequent fused observation and recording.

[0058] After collecting the perception outputs from multiple robots, the data streams generated by each robot are integrated and sequentially arranged along the time dimension to form a preliminary aggregated sequence of multi-robot perception information. The perception data uploaded by each robot is arranged in chronological order of sampling time and aligned using a unified time stamp, ensuring that the perception outputs of all robots can be matched on the same timeline. To ensure data consistency, the system calculates the sampling density of each perception sequence based on the interval between time stamps during the integration process, automatically inserting spatiotemporal annotations into data segments with large time spans, so that the perception performance of each robot can be compared at the same time scale in subsequent steps. After integration, a multi-dimensional time-series record containing radar echo intensity, image brightness changes, running speed sequences, and work rhythm information is obtained. This record provides an initial data framework for the subsequent generation of fused observation records. During this process, both the radar and vision components in the fused perception output are indexed using a unified time unit, enabling subsequent time segmentation and correlation analysis to be performed at the same reference scale.

[0059] After obtaining the initial aggregation sequence, the sensory information from different sources is fused to form a fused observation record. During the fusion process, the radar echo intensity sequence and the visual brightness sequence are superimposed according to the sampling time, following a unified temporal order. The operational rhythm information at the corresponding time is then incorporated as an auxiliary parameter into the fusion content of that time node. Each time node forms a fusion unit containing radar echo information, visual brightness information, and operational rhythm information, and a continuous fused observation record is constructed in chronological order. To ensure the continuity of the fused observation record in the temporal dimension, each fusion unit is connected to the previous and next time points, forming a traceable continuous data chain, allowing the operational status, environmental characteristics, and motion behavior within any time period to be simultaneously traced back. The fused observation record not only stores the sensory data of each robot during operation but also records the differences in operational rhythm for each robot within the same time period. This fused observation record reflects both the operational status of a single robot and the collaborative rhythm between multiple robots, providing a reliable basis for subsequent time segment comparison and abnormal segment identification.

[0060] After generating a complete fusion observation record, time synchronization information is annotated at the end of the record. This time synchronization information indicates the correspondence between the sensory data of each robot on the time axis and the distribution of synchronization moments among multiple robots. Through this information, a precise time mapping can be established between the fusion observation records of different robots, enabling the sensory data of multiple robots to be aligned under the same time reference frame. The process of annotating time synchronization information involves matching the time reference identifier of each robot's sensory output with the global operation time and attaching a time synchronization label to the end of the fusion observation record, ensuring that each segment of the fusion observation record corresponds to a unified global time scale. This annotation allows subsequent steps to directly compare time segments based on the time synchronization information, thereby quickly identifying abnormal change segments in the multi-robot sensory signals. After annotation, the fusion observation record achieves a unified expression from single-robot sensory data to multi-robot collaborative data, enabling changes in radar echo intensity, visual brightness, and operational rhythm to be correlated under the same time reference, providing a highly consistent input foundation for subsequent echo superposition detection and rhythm analysis.

[0061] Based on the time synchronization information, the fused observation records are compared in time segments to detect time segments in which a sudden increase in radar echo intensity and a change in image brightness occur simultaneously. An echo overlay attention list is generated, and the trigger boundary position is marked at the end of the echo overlay attention list.

[0062] After obtaining the fused observation records containing time synchronization information, in order to further identify abnormal sensing segments in the continuous sensing data that may be caused by transient environmental interference, the fused observation records undergo time-segmented comparison processing. This processing involves segmented analysis of time synchronization information, feature comparison of sensing signals, cross-analysis of changing trends, and annotation and archiving of results to form an echo superposition interest list containing target interference segments. The specific implementation steps are as follows:

[0063] Based on the time synchronization information marked in the fused observation records, all recorded data segments were segmented and organized in chronological order. During segmentation, the fused observation records were divided into continuous and interconnected time segments based on the time synchronization information, ensuring that each segment fully encompassed radar echo intensity, image brightness information, and operational rhythm information. Continuous division of the time scale ensured that each time segment represented a stable time window during the operation. Considering the time synchronization differences between multiple robots, multiple robot perception records under the same time reference were merged, allowing radar echoes and image brightness at the same moment to be compared within a unified time slice. After this segmentation process, the fused observation records were transformed into a series of chronologically arranged segments, each corresponding to a specific time interval, providing clear time boundaries for subsequent feature detection.

[0064] The sensory features of the fused observation records, segmented over time, are compared and contrasted. In this process, radar echo intensity data and image brightness change data are extracted from each time slice and compared within the same time slice. When a sudden increase in radar echo intensity occurs between adjacent time points, that time point is marked as an echo abrupt change node; similarly, when an abrupt change in image brightness occurs between adjacent time points, that time point is marked as a brightness abrupt change node. Guided by time synchronization information, echo abrupt change nodes and brightness abrupt change nodes are matched on the same time axis. If both occur simultaneously within the same or adjacent time slices, it indicates that radar perception and visual perception are changing synchronously within that time interval, which is highly likely caused by transient environmental interference. To ensure the continuity of the comparison, bidirectional comparisons are performed at the start and end times of each time slice during the detection process, ensuring that abrupt change features can be detected not only in the current time slice but also establish a correspondence with the change trends of preceding and following time slices. Through this slice-by-slice comparison method, time segments in a continuous time series where sudden increases in radar echo intensity and abrupt changes in image brightness occur simultaneously can be accurately identified.

[0065] The time segments where detected radar echo spikes and image brightness jumps occur simultaneously are integrated to form an echo overlay attention list. During generation, all identified time segments are arranged sequentially, and the start and end points of each segment are marked to clarify the duration of the interference. The generated echo overlay attention list not only includes the time range of the interference segment but also records information such as the operational status of each robot within that segment, the peak value of the radar echo, and the magnitude of image brightness changes, thus providing sufficient contextual data for subsequent interference localization. During integration, unmarked data from the preceding and following time slices that are adjacent to the interference segment are compared. If these data show similar trends to the interference segment, they are included in the echo overlay attention list, ensuring the scope of attention covers the entire interference occurrence process. The final echo overlay attention list is ordered chronologically and continuously records all segments potentially affected by environmental interference, providing a clear data foundation for the next step of marking trigger boundaries.

[0066] After creating the echo overlay attention list, each time segment in the list undergoes boundary analysis to determine the trigger boundary location. During this process, based on the start and end times of each segment in the list, the radar echo change rate and image brightness change rate at both ends of the segment are compared. When the echo intensity rapidly increases at the starting boundary and the brightness change rate increases synchronously, that location is designated as the starting end of the trigger boundary; when the echo intensity returns to the baseline level at the end of the segment and the brightness change rate stabilizes, that location is designated as the ending end of the trigger boundary. By bidirectionally labeling the starting and ending ends, the time interval of the interference can be accurately determined. Using this time interval as a benchmark, the trigger boundary location is uniformly labeled at the end of the echo overlay attention list. The labeling of the trigger boundary location not only indicates the specific range of the interference impact but also provides a temporal reference for locating the sudden stop trajectory in subsequent steps. After labeling, the echo overlay attention list achieves a complete closed loop from sensing data comparison to interference interval location. Each attention record has a clear time range, sensing characteristics, and boundary definition, enabling subsequent steps to accurately track the starting segment of the accidental sudden stop.

[0067] Based on the trigger boundary, the initial segment of the accidental emergency stop is located and fused with the attention list superimposed along the echo. The continuous changes in travel speed and avoidance actions before and after the initial segment are extracted to generate the emergency stop trajectory. The impact time window is marked at the end of the emergency stop trajectory.

[0068] After acquiring the echo overlay attention list and completing the trigger boundary marking, in order to further identify the specific starting position of the fusion-sensor accidental emergency stop, the attention list is located and analyzed along the trigger boundary. By extracting the continuous changes in travel speed and avoidance actions before and after the initial segment, an emergency stop trajectory reflecting the dynamic characteristics of the emergency stop process is generated. The impact time window is marked at the end of this trajectory to accurately present the impact range of the accidental emergency stop event on the work rhythm. The specific implementation steps are as follows:

[0069] Using the trigger boundary as a time reference, a recording interval containing the trigger boundary time period is selected from the echo overlay attention list and expanded in both forward and backward directions along the time sequence to construct an analysis time band covering the entire process before and after the emergency stop. The starting point of the analysis time band corresponds to the start time of the time slice before the trigger boundary, and the ending point corresponds to the end time of the time slice after the trigger boundary, ensuring that the time band can completely cover the entire process from disturbance triggering to action response. Within the analysis time band, the fused perception records corresponding to each robot are extracted, focusing on reading radar echo information, image brightness change information, and running rhythm information that match the time band to establish a complete multi-dimensional perception sequence. During the extraction process, all data are arranged in chronological order, so that radar echo, visual brightness, and running rhythm information correspond at the same time scale. In this way, the analysis time band not only includes the perception anomaly corresponding to the trigger boundary but also extends to the changes in the running state before and after the perception anomaly occurs, providing a sufficient temporal context for locating the starting segment of the emergency stop.

[0070] Within the constructed analysis timeframe, based on the trigger boundary positions in the echo overlay attention list, the changing trends of the sensing data are longitudinally compared to track the correspondence between changes in radar echo intensity, image brightness, and operating speed over time. When the radar echo intensity shows a continuous increase at the trigger boundary without accompanying changes in the actual obstacle position, it indicates that the sensing system is subjected to transient environmental interference at that time point. When the operating speed suddenly drops after that time point, it can be preliminarily determined that the robot has triggered an emergency stop within this time range. To accurately determine the starting segment of the emergency stop, multiple consecutive time points before and after the trigger boundary are sequentially compared to identify the transition moment when the operating speed changes from a continuous decrease to zero speed, and cross-validation is performed in conjunction with the time position of the avoidance action change. When the avoidance angle begins to increase rapidly or the avoidance direction changes abruptly, it indicates that the robot control layer has received a misjudged obstacle signal and initiated an avoidance response. In this way, the starting segment of the emergency stop action can be determined on the timeline, which is both the starting point of the response to the perception misjudgment and the initial position of subsequent rhythm disturbance.

[0071] After locating the starting segment of the fusion perception mis-touch emergency stop, extract the continuous data of the traveling speed change and the avoidance action change in the forward and backward time directions around this segment to generate a complete emergency stop change trajectory. During the extraction process, with the starting segment of the emergency stop as the center, incorporate the running speed information and the avoidance action information of several time slices before and after into the analysis scope to ensure that the trajectory can cover the whole process from normal traveling to emergency stop response and then to resuming traveling. For each time slice, record the corresponding traveling speed change amplitude, the change trend of the avoidance direction, and the change of the radar echo intensity, and keep them arranged synchronously with the time axis to form a continuous dynamic change sequence. When generating the emergency stop change trajectory, splice the previous speed reduction segment, the emergency stop response segment, and the later recovery segment in sequence so that the whole trajectory can intuitively reflect the complete behavior change process of the robot after misjudging an obstacle and triggering an emergency stop. This trajectory not only shows the emergency stop response characteristics of a single robot but also reflects the linkage effect among multiple robots within the same time interval through the time synchronization information, providing a data basis for analyzing the disturbance of the collaborative rhythm.

[0072] After generating the emergency stop change trajectory, conduct an impact effect analysis on the time period at the end of the trajectory to determine the impact time window and complete the annotation. The impact time window is used to reflect the duration range of the influence of the mis-touch emergency stop event on the collaborative operation rhythm. During the analysis process, determine the time interval when the robot transitions from the emergency stop state to the normal operation state based on the recovery trend of the running speed and the stable position of the avoidance action in the emergency stop change trajectory. When the traveling speed recovers to the stable level before the emergency stop and the avoidance angle recovers to the original operation direction, it indicates that the impact effect ends. Take the time when the speed drops to the lowest point in the emergency stop change trajectory as the starting point of the impact time window, and take the time when the speed and the avoidance state recover to the stable moment as the end point, and mark the start and end times of this time window at the end of the trajectory. After the annotation is completed, the emergency stop change trajectory not only records the whole process of the emergency stop response but also clarifies the interference duration range of the perception mis-touch on the operation rhythm through the impact time window. By generating the emergency stop change trajectory and annotating the impact time window, it is possible to record the whole process from perception misjudgment to behavior response and then to rhythm recovery, providing a complete time reference and dynamic behavior basis for analyzing the position where the collaborative queue rhythm is disturbed in the subsequent steps.

[0073] Trace back the running state of the collaborative queue corresponding to the emergency stop change trajectory according to the impact time window, identify the positions where the queue spacing shrinks and the positions where the order of the avoidance actions changes, generate a rhythm disturbance position table, and mark the adjustment anchor points at the end of the rhythm disturbance position table;

[0074] After obtaining the trajectory of the sudden stop and marking the impact time window at its end, to further determine the impact range of accidental sudden stop caused by fused perception on the rhythm of multi-robot collaborative operation, time backtracking and state analysis are performed on the collaborative queue operation state corresponding to the sudden stop trajectory. By identifying the positions of queue spacing contraction and changes in the sequence of avoidance actions, the rhythm disturbance caused by the sudden stop event in group operation can be accurately revealed. Adjustment anchor points are marked at the end of the generated rhythm disturbance position table, providing accurate spatiotemporal reference for subsequent dynamic operation adjustment. The specific implementation steps are as follows:

[0075] Using the impact time window marked in the sudden stop trajectory as the time reference, a retrospective analysis of the operational status data of multi-robot collaborative operations is performed. The time range of the retrospective analysis covers the entire process before the sudden stop event, during the event, and after the event. By using the start and end times of the impact time window, the travel speed, position coordinates, queue number, avoidance direction, and operation rhythm information of each robot within that time interval are extracted one by one and arranged sequentially according to a unified time label, so that all robots form a corresponding operational status sequence on the same time axis. During the retrospective process, the time synchronization with the sudden stop trajectory is maintained to ensure that the operational status of each robot at the same point in time can match the time node of the sudden stop event. In this way, a one-to-one correspondence between the impact time window and the multi-robot collaborative operation status is established, enabling subsequent changes in queue spacing and avoidance actions to be compared and analyzed in the time dimension.

[0076] After completing the time alignment of the collaborative operation state, dynamic analysis of queue spacing is performed on the backtracked state data. Based on the trajectory of each robot before and after the impact time window, the spatial distance between adjacent robots is calculated, and the distance change trend is mapped onto the time axis. If, within the time interval corresponding to the initial segment of the sudden stop trajectory change, the speed of the preceding robot rapidly decreases due to accidental sudden stop, the subsequent robots will gradually shrink their travel distance due to inertial delay and avoidance reaction. When the spacing between two adjacent robots decreases below the stable queue distance, this time position is marked as the queue spacing shrinkage position. To ensure complete identification of queue shrinkage, the analysis not only focuses on the directly affected robot pairs but also considers the subsequent robot queues affected by the chain effect. When three or more consecutive robots show similar spacing shrinkage trends in the time series, this phenomenon is recorded as a continuous queue shrinkage interval, and the start and end time positions are marked. In this way, the queue spacing shrinkage position caused by sudden stop can be completely identified on the time axis, providing a basis for spatial changes in subsequent rhythm disturbance analysis.

[0077] After identifying the location of the queue spacing contraction, the sequence of avoidance actions within the same time range is further analyzed. The sequence of avoidance actions refers to the order in which robots in a collaborative work queue respond to perceived anomalies and execute avoidance maneuvers. When a robot in front accidentally triggers an emergency stop and initiates an avoidance maneuver, subsequent robots often adjust their direction of travel and avoidance strategy based on the movement of the robot in front. If, during the impact time window, a subsequent robot exhibits an earlier avoidance action, a sudden change in direction, or an increased avoidance magnitude, it indicates a change in the sequence of avoidance actions within the queue. To accurately identify this change, the analysis uses time synchronization information as a benchmark, longitudinally comparing the timing of avoidance action changes for each robot, recording the start time of avoidance actions and the magnitude of direction adjustments for different robots at the same time. When a robot behind is detected to have an avoidance action earlier than the robot in front or its direction adjustment magnitude exceeds a preset range, this moment is marked as the location of the change in the avoidance action sequence. Based on this, a time series containing multiple avoidance change moments is formed and correlated with the queue spacing contraction location identified in the previous step to determine the correspondence between queue spacing changes and avoidance sequence changes. This dual comparison clearly reveals the timing disturbance characteristics caused by sudden stop events due to accidental sensor touch in the coordinated queue.

[0078] After identifying the positions where queue spacing contractions and the positions where the avoidance action sequence changes, the results are integrated to generate a rhythm disturbance position table, with adjustment anchor points marked at the end of the table. The rhythm disturbance position table is time-based, pairing each queue spacing contraction position with its corresponding avoidance action sequence change position to form a series of interconnected disturbance data items. Each data item includes the occurrence time, affected robot number, magnitude of the change in queue spacing, avoidance action initiation time, and direction adjustment information. All data items are arranged in chronological order, forming a complete rhythm disturbance position table. At the end of the table, based on the temporal distribution of each disturbance item, the position representing the rhythm recovery turning point is extracted as the adjustment anchor point. The time position corresponding to the adjustment anchor point is usually located at the moment when the queue spacing recovers to a stable value and the avoidance action returns to the predetermined direction. By marking adjustment anchor points at the end of the rhythm disturbance position table, a precise timing reference can be provided for subsequent dynamic operation adjustments, enabling the adjustment process to be centered on this anchor point, performing segmented control and rhythm balancing before and after the disturbance interval. After the adjustment anchor points are marked, the rhythm disturbance location table not only records the entire process of rhythm disturbance of the coordinated queue within the impact time window, but also provides quantifiable reference locations for the dynamic adjustment phase.

[0079] Based on the adjustment anchor point, dynamic operation adjustment is performed on the rhythm disturbance location table. Segmented deceleration control is performed in the trigger boundary area, short pause transition control is performed in the impact time window, and turn-by-turn traffic control is performed on the avoidance action sequence. The adjusted operation rhythm information is updated to the fusion observation record.

[0080] After generating a rhythm disturbance location table and marking adjustment anchor points at the end of the table, to enable multi-robot collaborative operation to regain balanced operation after being disturbed by accidental emergency stops caused by fusion perception, dynamic operation adjustment is performed on the rhythm disturbance location table based on the adjustment anchor points. This process comprehensively considers the changing characteristics of the trigger boundary region, impact time window, and avoidance action sequence, continuously repairing and restoring the operation state of multiple robots in time and space, allowing each robot to re-enter a coordinated operation rhythm, and updating the adjusted operation rhythm information to the fusion observation record to achieve a dynamic closed loop of the operation state. The specific implementation steps are as follows:

[0081] Using the adjustment anchor points marked in the rhythm disturbance location table as time references, the disturbance intervals within the preceding and following time periods are divided and reorganized. During the division, the preceding stage is defined as the disturbance recovery preparation segment, and the following stage as the rhythm reconstruction segment, centered on the time position corresponding to the adjustment anchor point. The preceding stage mainly covers the overlapping portion of the trigger boundary region and the impact time window, while the following stage includes the regression process of the change in the sequence of avoidance actions. After division, the running speed curve, position change trajectory, and avoidance action direction of each robot in these two stages are extracted and kept time-synchronized, ensuring that the running states of all robots at the same moment can be compared on a unified timeline. Based on this, a dynamic adjustment interval centered on the adjustment anchor point is established, providing time positioning for subsequent segmented control and rhythm recovery. This method ensures that the dynamic adjustment actions are continuous in time and connected in space, providing a stable transition basis for the running rhythm of the entire collaborative queue.

[0082] After determining the dynamic adjustment range, segmented deceleration control is implemented for the trigger boundary region. The trigger boundary region is the temporal and spatial range in which the initial accidental stop due to perception fusion occurs. This region is typically accompanied by a simultaneous increase in radar echo and a jump in visual brightness. Within this region, to prevent further emergency stops caused by transient interference, the movement speed of each robot is gradually reduced in segments according to the time sequence from the trigger boundary. Specifically, the short time interval before the trigger boundary is designated as the deceleration initiation segment, the interval within the trigger boundary is designated as the speed maintenance segment, and the interval after the trigger boundary is designated as the acceleration transition segment. In the deceleration initiation segment, the rate of change of speed is adjusted to ensure that the robot completes a smooth deceleration before entering the trigger boundary; in the speed maintenance segment, a uniform low speed is maintained to ensure that no further misjudgment is triggered during the duration of the interference signal; in the acceleration transition segment, the original operating speed is gradually restored, allowing the robot to re-enter the normal operating rhythm after leaving the trigger boundary. This segmented deceleration control method effectively mitigates operational fluctuations within the trigger boundary area, smooths out the relative speed difference between robots, avoids a new round of rhythm disturbances caused by rapid changes, and provides a stable input state for subsequent short pause transition control.

[0083] After completing the segmented deceleration control in the trigger boundary area, a short pause transition control is executed for the impact time window. The impact time window reflects the duration of the impact of a perceived accidental stop event on the rhythm of the coordinated queue. Within this window, there are time differences in the speed recovery and avoidance responses of different robots. If not coordinated, this can easily lead to uneven rhythms among the queues. Therefore, a short pause point is set at the beginning of the impact time window, causing all robots to pause synchronously for a very short time at that point to reset the rhythm within the queue. After the pause, each robot starts sequentially according to the time intervals set before and after the adjustment anchor point, forming a gradual recovery rhythm pattern. The pause time and start interval are determined based on the speed recovery trend of each robot within the impact time window to ensure a continuous and smooth recovery process. During the entire short pause transition control process, the robot in front resumes movement first, and subsequent robots enter the running state sequentially according to the time intervals, thus forming a backward-propagating rhythm recovery wave. This sequential recovery method can effectively resolve the rhythm inconsistencies caused by the sudden stop event, allowing the queue to re-establish a uniform running rhythm while reducing cross-interference between avoidance actions. Through this transition control, the coordinated queue transitions from an impact state to a balanced state, and the overall operation tends to be synchronized, laying the foundation for subsequent turn-by-turn traffic control.

[0084] After completing the short pause transition control within the impact time window, a turn-by-turn traffic control is implemented for the avoidance action sequence, and the adjusted running rhythm information is updated to the fusion observation record. During the turn-by-turn traffic control process, the robots with avoidance conflicts are prioritized based on the change position of the avoidance action sequence recorded in the rhythm disturbance position table. When two or more robots have overlapping avoidance paths in the same time period, the right of way is allocated sequentially according to the time order of the adjustment anchor points, so that each robot has a clear time interval when performing avoidance actions. After the previous robot completes the avoidance, the next robot enters the avoidance path, thereby realizing the turn-by-turn traffic among multiple robots. During the passage, the start time, end time, and path offset of each robot's avoidance action are continuously monitored, and the updated running rhythm data is supplemented to the fusion observation record in real time based on the time synchronization information. During the update process, the original time structure remains unchanged, and the newly generated rhythm information is matched one-to-one with the original perception record, speed record, and avoidance record, so that the fusion observation record can fully reflect the real running state after dynamic adjustment. After the update, the fused observation records not only contain the original perception data and operation rhythm information, but also newly added dynamically adjusted rhythm balance data, providing a reference for subsequent collaborative operations. Through this process, the operating rhythm of multi-robot collaborative operations is recalibrated, and the spacing between queues, avoidance order, and operating rhythm return to consistency, realizing a closed loop from accidental perception-induced emergency stop to rhythm recovery.

[0085] This invention introduces a fusion observation recording and time synchronization mechanism during multi-robot collaborative operations to achieve unified management of radar perception, visual perception, and operational rhythm information. This enables all robots to perform data alignment and perception fusion under the same time reference. By detecting synchronized segments of sudden increases in radar echo intensity and changes in image brightness during time segment comparison, it can quickly identify perception anomalies caused by transient environmental disturbances, effectively avoiding emergency stops triggered by perception misjudgments. This maintains operational continuity and path coordination stability, allowing multiple robots to maintain an efficient collaborative operational rhythm even in complex farmland environments.

[0086] This invention establishes a rhythm disturbance location table based on the impact time window and performs dynamic operation adjustments, enabling multiple robots to automatically restore their operational order after being affected by accidental sensor touches. Through adjustments such as segmented deceleration, short pauses, and turn-by-turn movement, the collaborative queue achieves adaptive recovery at the temporal level, restoring balance to queue spacing and avoidance order. This method improves the continuity and stability of the multi-robot operation rhythm, reduces coverage blind spots and repetitive tasks caused by sudden stops, and significantly enhances the overall efficiency of collaborative operations and the accuracy of weeding in farmland.

[0087] This invention provides, for example Figure 2The multi-robot collaborative operation route planning and generation system shown includes a fusion observation record generation module, an echo superposition detection module, an emergency stop trajectory extraction module, a rhythm disturbance analysis module, and a dynamic operation adjustment module.

[0088] The fusion observation record generation module collects fusion perception output, time records and operation rhythm information of multiple robots in the operation site, generates fusion observation records, and marks time synchronization information at the end of the fusion observation records;

[0089] The echo overlay detection module compares the fused observation records in time segments based on time synchronization information, detects time segments in which a sudden increase in radar echo intensity and a change in image brightness occur simultaneously, generates an echo overlay attention list, and marks the trigger boundary position at the end of the echo overlay attention list.

[0090] The emergency stop trajectory extraction module locates and fuses the initial segment of accidental emergency stop based on the trigger boundary and the echo superimposed on the attention list. It extracts the continuous changes in travel speed and avoidance actions before and after the initial segment, generates the emergency stop trajectory, and marks the impact time window at the end of the emergency stop trajectory.

[0091] The rhythm disturbance analysis module traces back the coordinated queue operation status corresponding to the sudden stop change trajectory based on the impact time window, identifies the position of queue spacing contraction and the position of change in the order of avoidance actions, generates a rhythm disturbance position table, and marks the adjustment anchor point at the end of the rhythm disturbance position table.

[0092] The dynamic operation adjustment module performs dynamic operation adjustment based on the adjustment anchor point and the rhythm disturbance position table. It performs segmented deceleration control in the trigger boundary area, short pause transition control in the impact time window, and turn-by-turn traffic control for the avoidance action sequence. The adjusted operation rhythm information is then updated to the fusion observation record.

[0093] The multi-robot collaborative operation route planning and generation method provided in this embodiment of the invention is implemented through the aforementioned multi-robot collaborative operation route planning and generation system. For details of the specific methods and processes of the multi-robot collaborative operation route planning and generation system, please refer to the embodiments of the multi-robot collaborative operation route planning and generation method described above, which will not be repeated here.

[0094] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for generating routes for multi-robot collaborative operations, characterized in that, Includes the following steps: Collect fusion perception output, time records and operation rhythm information of multi-robot operation site, generate fusion observation record, and mark time synchronization information at the end of the fusion observation record; Based on the time synchronization information, the fused observation records are compared in time segments to detect time segments in which a sudden increase in radar echo intensity and a change in image brightness occur simultaneously. An echo overlay attention list is generated, and the trigger boundary position is marked at the end of the echo overlay attention list. Based on the trigger boundary, the initial segment of the accidental emergency stop is located and fused with the attention list superimposed along the echo. The continuous changes in travel speed and avoidance actions before and after the initial segment are extracted to generate the emergency stop trajectory. The impact time window is marked at the end of the emergency stop trajectory. Based on the impact time window, the coordinated queue operation status corresponding to the sudden stop change trajectory is traced back, the position of queue spacing contraction and the position of change in the order of avoidance actions are identified, a rhythm disturbance position table is generated, and adjustment anchor points are marked at the end of the rhythm disturbance position table. Based on the adjustment anchor point, dynamic operation adjustment is performed on the rhythm disturbance location table. Segmented deceleration control is performed in the trigger boundary area, short pause transition control is performed in the impact time window, and turn-by-turn traffic control is performed on the avoidance action sequence. The adjusted operation rhythm information is updated to the fusion observation record. The steps for marking the impact time window are as follows: Using the trigger boundary as a time reference, the recording interval containing the time period of the trigger boundary is selected from the echo overlay attention list, and expanded in the forward and backward directions along the time sequence to construct an analysis time band covering the entire emergency stop process. Radar echo information, image brightness change information and operation rhythm information that match the analysis time band are extracted. Based on the trigger boundary position in the echo superposition attention list, the change trend of the sensing data within the analysis time band is compared longitudinally to determine the time point when the radar echo intensity increases and the operating speed suddenly drops, thus locating the starting segment of the fusion sensing accidental emergency stop. Extract continuous speed change and avoidance action change data in the forward and backward directions around the initial segment of the emergency stop to generate the emergency stop trajectory and keep it synchronized with the time axis; The impact time window is determined based on the speed recovery trend and the stabilization position of the avoidance maneuver in the sudden stop change trajectory, and the start and end times of the impact time window are marked at the end of the sudden stop change trajectory. The steps for adjusting anchor point annotations are as follows: Using the impact time window marked in the sudden stop trajectory as the time reference, the operation status data of multi-robot collaborative operation is analyzed by time backtracking. The travel speed, position coordinates, queue number, avoidance direction and operation rhythm information of each robot in the time range corresponding to the impact time window are extracted and arranged in order according to a unified time label. Based on the backtracking analysis results, the queue spacing of each robot is dynamically analyzed to identify the position where the queue spacing shrinks due to the decrease in speed of the robot in front, and the start and end times of the continuous queue shrinkage intervals are recorded. After identifying the position where the queue spacing shrinks, the changes in the sequence of avoidance actions within the same time range are analyzed to determine the time position of avoidance actions that are advanced, have a sudden change in direction, or have a change in amplitude, and these are correlated with the position where the queue spacing shrinks. By integrating the positions of queue spacing contraction and the positions of changes in the sequence of avoidance actions, a table of rhythm disturbance positions is generated, and adjustment anchor points are marked at the end of the table of rhythm disturbance positions based on the position of the rhythm recovery turning point. The adjusted operational rhythm information was updated to the fused observation records as follows: Using the adjustment anchor points in the rhythm disturbance location table as time references, the disturbance recovery preparation segment and the rhythm reconstruction segment are divided, and the running speed curves, position change trajectories and avoidance action directions of each robot are extracted to establish a dynamic adjustment range. Segmented deceleration control is implemented for the trigger boundary area, with the first interval set as the deceleration segment, the middle interval set as the speed maintenance segment, and the last interval set as the acceleration segment to balance the running rhythm. During the impact time window, a short pause transition control is executed to make each robot pause at a unified time point and resume operation sequentially according to the time interval; The sequence of avoidance actions is controlled by taking turns, and the updated operation rhythm information is added to the fusion observation record based on the time synchronization information.

2. The multi-robot collaborative operation route planning and generation method according to claim 1, characterized in that, The steps for annotating time synchronization information are as follows: Within the multi-robot collaborative operation area, the fusion perception output of each robot is collected, and the source, sampling time and spatial location information of the perception data are integrated, and radar echo information, visual brightness information and operation rhythm information are recorded. After the data collection is completed, the perception data of each robot is integrated and arranged in chronological order, aligned with a unified time stamp, forming an aggregated sequence of multi-robot perception information, and spatiotemporal annotations are inserted according to the interval between time stamps. After obtaining the aggregated sequence, the radar echo intensity sequence, visual brightness sequence and operation rhythm information are superimposed in a unified time order to form a continuous fused observation record. Time synchronization information is annotated at the end of the fused observation record. A time mapping is established by matching the time label of the perception output with the global operation time, so that the perception data of multiple robots can be aligned under the same time reference system.

3. The multi-robot collaborative operation route planning and generation method according to claim 2, characterized in that, The steps to trigger boundary location annotation are as follows: Based on the time synchronization information in the fusion observation records, the fusion observation records are segmented and organized in chronological order, dividing the fusion observation records into continuous and interconnected time segments, and merging the multi-robot perception records under the same time reference. The sensory feature comparison detection was performed on the segmented and sorted fused observation records. Radar echo intensity data and image brightness change data were extracted in each time slice, and time segments with simultaneous sudden increases and jumps were matched. The detected time segments are sequentially integrated to form an echo overlay attention list containing interference intervals, and the start and end times of each time segment are marked. Based on the start and end times in the echo overlay attention list, the radar echo change rate and image brightness change rate at both ends of the segment are compared, the trigger boundary position is marked, and the time interval of the interference is recorded.

4. The multi-robot collaborative operation route planning and generation method according to claim 3, characterized in that, When marking the trigger boundary position, a two-way boundary analysis is performed on each time segment in the echo overlay attention list. When the radar echo intensity continues to rise at the starting boundary and the image brightness change rate increases synchronously, it is determined as the starting end of the trigger boundary. When the radar echo intensity returns to the reference level at the ending boundary and the image brightness change rate tends to stabilize, it is determined as the ending end of the trigger boundary. The trigger boundary position is uniformly recorded at the end of the echo overlay attention list.

5. The method for generating multi-robot collaborative operation routes according to claim 1, characterized in that, The emergency stop trajectory includes an initial speed reduction segment, an emergency stop response segment, and a recovery segment. These segments are spliced ​​together in chronological order. The starting point of the impact time window is when the speed drops to its lowest point, and the ending point is when the speed and avoidance direction return to a stable state. The emergency stop trajectory and the impact time window correspond in the time dimension to reflect the continuous impact of accidental emergency stop on the work rhythm.

6. The multi-robot collaborative operation route planning and generation method according to claim 1, characterized in that, The adjustment anchor points in the rhythm disturbance location table are determined based on the time and position when the queue spacing recovers to a stable value and the avoidance action returns to the predetermined direction. The adjustment anchor points are used to identify the turning point of rhythm recovery and provide timing references for segmented control and rhythm balance in dynamic operation adjustment.

7. A multi-robot collaborative operation route planning and generation system, used to implement the multi-robot collaborative operation route planning and generation method according to any one of claims 1-6, characterized in that, It includes a fusion observation record generation module, an echo superposition detection module, an emergency stop trajectory extraction module, a rhythm disturbance analysis module, and a dynamic operation adjustment module: The fusion observation record generation module collects fusion perception output, time records and operation rhythm information of multiple robots in the operation site, generates fusion observation records, and marks time synchronization information at the end of the fusion observation records; The echo overlay detection module compares the fused observation records in time segments based on time synchronization information, detects time segments in which a sudden increase in radar echo intensity and a change in image brightness occur simultaneously, generates an echo overlay attention list, and marks the trigger boundary position at the end of the echo overlay attention list. The emergency stop trajectory extraction module locates and fuses the initial segment of accidental emergency stop based on the trigger boundary and the echo superimposed on the attention list. It extracts the continuous changes in travel speed and avoidance actions before and after the initial segment, generates the emergency stop trajectory, and marks the impact time window at the end of the emergency stop trajectory. The rhythm disturbance analysis module traces back the coordinated queue operation status corresponding to the sudden stop change trajectory based on the impact time window, identifies the position of queue spacing contraction and the position of change in the order of avoidance actions, generates a rhythm disturbance position table, and marks the adjustment anchor point at the end of the rhythm disturbance position table. The dynamic operation adjustment module performs dynamic operation adjustment based on the adjustment anchor point and the rhythm disturbance position table. It performs segmented deceleration control in the trigger boundary area, short pause transition control in the impact time window, and turn-by-turn traffic control for the avoidance action sequence. The adjusted operation rhythm information is then updated to the fusion observation record.

Images