Intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion

Abstract

This invention relates to the field of vehicle control technology, specifically disclosing an intelligent reversing decision and control system for semi-trailers based on multi-source sensor data fusion. This system uses a data acquisition module to collect environmental and status information of the semi-trailer from multiple sensors, forming multi-source sensor data. The data is first transmitted to a buffer and processed uniformly according to a set fusion interval. The data fusion module projects various sensor data onto the semi-trailer's coordinate system, improving spatial matching accuracy through coordinate transformation. A fusion synchronization index set is introduced to evaluate the consistency and timeliness of the data from each sensor, ensuring that fusion is performed only when the data is valid, avoiding error propagation and information conflicts, and significantly improving the spatial consistency and temporal coordination of the data. The high-quality fused data is input to the decision optimization module, which intelligently plans a reference reversing trajectory and generates control commands. Simultaneously, the system dynamically updates decisions based on the vehicle's status, achieving high-precision reversing control.

Description

Technical Field

[0001] This invention relates to the field of vehicle control technology, and in particular to an intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion. Background Technology

[0002] During the reversing process of a semi-trailer truck, the reversing radar accurately detects the distance to obstacles behind the vehicle by emitting and receiving ultrasonic waves. An alarm is immediately triggered if the distance is less than a set threshold. The reversing camera displays the real-time rear view on the in-vehicle screen for easy driver observation. Furthermore, the vehicle's intelligent decision-making and control system integrates the radar distance data and the scene information from the image, using intelligent algorithms to quickly analyze and automatically generate a reasonable reversing strategy. This, in turn, controls the vehicle's steering, power, and other systems, reducing the difficulty and risk of reversing.

[0003] For example, Chinese invention patent CN111422181B discloses a method for safe control of vehicle reversing. This method aims to utilize existing vehicle speed, acceleration, throttle position sensors, and electronic throttle on the vehicle to achieve safety control functions and solve the safety problem of reversing. The method for updating the intelligent decision-making of semi-trailer trucks is applied to a vehicle reversing safety control system. This system includes a signal acquisition module, a control module, and an execution module. The signal acquisition module includes a gear shift switch collector, a vehicle speed sensor, a vehicle acceleration sensor, and a throttle position sensor to collect relevant reversing signals. The control module includes a signal processing submodule and a decision submodule to receive, process, and output control signals to the execution module after decision-making. The execution module includes an electronic throttle, a braking device, and warning lights to achieve vehicle reversing speed control and early warning.

[0004] For example, Chinese invention patent CN116534010A discloses a method and apparatus for determining the reversing behavior of a vehicle. The method includes: obtaining at least one candidate valid target with at least one corner point in a preset region of interest around the vehicle; determining whether the valid target and the front of the vehicle meet preset passable conditions; if both meet the preset passable conditions, determining whether the vehicle meets preset reversing conditions, and controlling the vehicle to perform a reversing action when the preset reversing conditions are met, until the vehicle or the valid target is detected to meet the preset reversing exit conditions.

[0005] However, in the process of implementing the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems: When existing reversing technology uses data fusion for decision-making, it faces challenges in data integrity and decision robustness. Due to the limitations of sensor installation location and performance, different sensors have different coverage ranges of the vehicle's surrounding environment, resulting in incomplete data collection. Directly making decisions based on this incomplete data will cause the system's perception of the environment to be biased, the decision-making system is easily interfered with, has poor robustness, and cannot guarantee the safety and stability of the reversing process. Summary of the Invention

[0006] To address the technical problem of existing decision-making systems being susceptible to interference, this invention provides an intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion. The technical solution is as follows:

[0007] The intelligent reversing decision and control system for semi-trailers based on multi-source sensor data fusion includes a data acquisition module, which acquires environmental information and status information of the semi-trailer through multi-source sensors, comprehensively labels it as multi-source sensor data, sets up a data buffer, transmits the multi-source sensor data to the data buffer, and performs data fusion operation on the multi-source sensor data in the data buffer according to a preset fusion interval; a data fusion module, which projects the multi-source sensor data onto the semi-trailer's coordinate system through coordinate transformation during the data fusion process, collects and evaluates the fusion synchronization index set of the multi-source sensor data to determine whether the multi-source sensor data is labeled as valid data. If the multi-source sensor data is labeled as valid data, the data fusion operation is completed; if the multi-source sensor data is labeled as invalid data, the data fusion operation is canceled; and a decision optimization module, which, after the data fusion operation is completed, intelligently decides the reference reversing trajectory of the semi-trailer, generates control commands for the semi-trailer, analyzes the reversing anomaly information of the semi-trailer, and determines whether to update the intelligent decision of the semi-trailer.

[0008] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0009] (1) This invention provides a semi-trailer intelligent reversing decision and control system based on multi-source sensor data fusion. Through the data acquisition module, the system uses multi-source sensors to collect environmental and status information of the semi-trailer, forming multi-source sensor data. The data is first transmitted to a buffer and then processed uniformly according to a set fusion interval. The data fusion module projects various sensor data onto the semi-trailer's body coordinate system, improves spatial matching accuracy through coordinate transformation, and introduces a fusion synchronization index set to evaluate the consistency and timeliness of each sensor's data, ensuring that fusion is only performed under the premise that the data is valid, avoiding error propagation and information conflicts, and significantly improving the spatial consistency and temporal coordination of the data. The high-quality fused data is input to the decision optimization module, which intelligently plans the reference reversing trajectory and generates control commands. At the same time, the system dynamically updates the decision based on the vehicle status to achieve high-precision reversing control. The system's fusion mechanism enhances the robustness and accuracy of environmental perception, and is particularly suitable for reversing scenarios of semi-trailers with complex structures and many dynamic interferences.

[0010] (2) This invention dynamically filters and optimizes multi-source sensor data by comparing and judging the synchronization index and growth rate in multiple rounds, avoiding low-quality data from participating in the fusion, significantly improving the temporal consistency and spatial matching accuracy of the data. Frame interpolation optimization can improve the temporal resolution when the synchronization deviation is large, enhancing the reliability of fusion. The hierarchical judgment mechanism constructs a fusion closed loop of "screening - re-evaluation - optimization - re-evaluation", improving the fault tolerance and environmental adaptability of the system in complex scenarios. The fusion parameters are dynamically matched through the database, supporting adaptive adjustment, effectively reducing the impact of erroneous fusion on reversing path planning, and ensuring the stability and accuracy of control output.

[0011] (3) By comparing the abnormal coefficient of the reversing decision execution with the threshold, this invention can accurately identify the deviation of the reversing path, update the reference trajectory in a timely manner and verify its feasibility, and avoid misleading control commands. By recognizing the number, state and movement trend of obstacles, the system realizes a graded early warning mechanism from static waiting to dynamic obstacle avoidance, and can actively intervene in braking or speed limiting to reduce the risk of collision. At the same time, the early warning level is linked to visual prompts, which improves the driver's perception of reversing risks and reaction speed, especially in complex scenarios, effectively ensuring the continuity and safety of semi-trailer truck reversing operations. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of system module connections provided in an embodiment of the present invention.

[0014] Figure 2 This is a schematic diagram of the information fusion process provided in an embodiment of the present invention.

[0015] Figure 3 This is a schematic diagram of the decision control process provided in an embodiment of the present invention. Detailed Implementation

[0016] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0017] 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.

[0018] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0019] 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.

[0020] 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.

[0021] See Figure 1 As shown, this embodiment of the invention provides a technical solution: a semi-trailer intelligent reversing decision and control system based on multi-source sensor data fusion, including a data acquisition module, a data fusion module, a decision optimization module, and a database.

[0022] The database is used to store the parameters involved in the intelligent reversing decision and control system for semi-trailer trucks, which integrates multi-source sensor data.

[0023] The data acquisition module is connected to the data fusion module and the decision optimization module, respectively. The data fusion module is connected to the decision optimization module, and the data acquisition module, data fusion module, and decision optimization module are all connected to the database.

[0024] The data acquisition module is used to acquire environmental information and status information of the semi-trailer through multi-source sensors, and comprehensively label them as multi-source sensor data. A data buffer is set up, and the multi-source sensor data is first transmitted to the data buffer. According to the preset fusion interval, the multi-source sensor data in the data buffer is fused.

[0025] The data fusion module is used to project multi-source sensor data onto the semi-trailer's body coordinate system through coordinate transformation during the data fusion process. It collects and evaluates the fusion synchronization index set of multi-source sensor data to determine whether the multi-source sensor data is marked as valid data. If the multi-source sensor data is marked as valid data, the data fusion operation is completed. If the multi-source sensor data is marked as invalid data, the data fusion operation is canceled.

[0026] The decision optimization module is used to intelligently determine the reference reversing trajectory of the semi-trailer after the data fusion operation is completed, generate control commands for the semi-trailer, analyze the abnormal reversing information of the semi-trailer, and thus determine whether to update the intelligent decision of the semi-trailer.

[0027] The data acquisition module uses various sensors installed on the semi-trailer (such as cameras, millimeter-wave radar, ultrasonic sensors, GPS, and IMU) to collect real-time environmental information (e.g., obstacle positions, road boundaries) and the vehicle's own status information (e.g., speed, steering angle, and position coordinates). This data from different sources is collectively referred to as "multi-source sensor data." It is first transmitted to a temporary storage area called the data buffer. The system periodically retrieves this data from the buffer according to time intervals set by technicians (e.g., every 0.5 seconds) and performs data fusion. Using the ROS system combined with the Extended Kalman Filter (EKF) algorithm, the data from GPS, IMU, radar, and cameras is time-aligned and coordinate-transformed, uniformly converting it to the vehicle's own coordinate system. This process automatically runs on the vehicle's edge computing unit (such as NVIDIA Jetson) without manual intervention, providing high-quality data input for subsequent reversing path planning.

[0028] In a specific implementation, as one example, the vehicle body coordinate system is a three-dimensional spatial coordinate system established with the semi-trailer itself as a reference. It is used to uniformly describe the positional relationships of various information in the semi-trailer and its surrounding environment. The origin of this coordinate system is usually set at the geometric center of the semi-trailer or the midpoint of the front axle. The X-axis points in the direction of travel of the semi-trailer, and the Y-axis is perpendicular to the X-axis and points to the left side of the vehicle. By transforming the data collected by various sensors (such as LiDAR, cameras, GPS, etc.) from their local coordinate system or geographic coordinate system to the vehicle body coordinate system, spatial alignment of data from different sensors can be achieved, enabling the targets sensed by each sensor to be accurately located and identified under the same spatial reference.

[0029] In one specific embodiment, the present invention dynamically filters and optimizes multi-source sensor data through multiple rounds of synchronization index comparison and growth rate judgment, avoiding low-quality data from participating in fusion, significantly improving the temporal consistency and spatial matching accuracy of the data. Frame interpolation optimization can improve temporal resolution when the synchronization deviation is large, enhancing the reliability of fusion. The hierarchical judgment mechanism constructs a fusion closed loop of "screening-re-evaluation-optimization-re-assessment", improving the fault tolerance and environmental adaptability of the system in complex scenarios. The fusion parameters are dynamically matched through the database, supporting adaptive adjustment, effectively reducing the impact of erroneous fusion on reversing path planning, and ensuring the stability and accuracy of control output.

[0030] Figure 2 This is a schematic diagram of the information fusion process provided in this embodiment of the invention. The system determines whether the information fusion synchronization index is greater than or equal to the information fusion synchronization threshold. If the condition is met, the data is deemed valid, and the current round of data fusion is completed directly. Otherwise, the system extends the fusion interval and performs a fusion process again. After the fusion is repeated, the system again determines whether the newly calculated synchronization index reaches the information fusion synchronization threshold. If the condition is met, the data fusion continues. If the condition is still not met, the synchronization index growth rate is calculated, and it is determined whether it is greater than or equal to the growth threshold. If the growth rate does not exceed the growth threshold, the system determines that the data is invalid and cancels the current round of fusion operation. If the growth rate exceeds the growth threshold, the system performs frame interpolation optimization on the original data, performs fusion again, and determines its validity. After the valid data fusion is completed, the system enters the reversing decision stage.

[0031] In a specific implementation, as one example, multi-source sensor data is uniformly projected onto the semi-trailer's body coordinate system through coordinate transformation. First, based on the installation position and attitude parameters of each sensor on the semi-trailer, a geometric mapping relationship is established between the sensor coordinate system and the semi-trailer's body coordinate system. This typically includes two parts: translation (position offset) and rotation (attitude difference). By constructing a coordinate transformation matrix containing rotation and translation vectors, the original sensor data points can be converted from their local coordinate system to their corresponding coordinates in the vehicle's body coordinate system, achieving projection mapping of data under a unified coordinate reference. This operation ensures spatial consistency of data from different sensors, thus providing a reliable fusion foundation for subsequent environmental modeling, target recognition, and path planning.

[0032] Specifically, the process for determining whether multi-source sensor data is marked as valid data is as follows: evaluate the fusion synchronization index set of multi-source sensor data to obtain the information fusion synchronization index of multi-source sensor data within the data fusion cycle, and compare it with the preset information fusion synchronization threshold in the database; the information fusion synchronization threshold represents the lower limit of the reasonable range of the information fusion synchronization index.

[0033] If the information fusion synchronization index of the multi-source sensor data within the data fusion period is greater than or equal to the information fusion synchronization threshold, the multi-source sensor data is marked as valid data. If the information fusion synchronization index of the multi-source sensor data within the data fusion period is less than the information fusion synchronization threshold, the fusion interval is extended based on the information fusion synchronization index of the multi-source sensor data within the data fusion period, and the multi-source sensor data is reacquired and the data fusion operation is performed again to obtain a secondary information fusion synchronization index of the multi-source sensor data. This index is then compared with the information fusion synchronization threshold to determine whether the multi-source sensor data is marked as valid data. If the multi-source sensor data is marked as valid data, the data fusion operation is completed. If the multi-source sensor data is not marked as valid data, it is determined whether to perform data optimization on the multi-source sensor data.

[0034] Extending the fusion interval based on the information fusion synchronization index of multi-source sensor data within the data fusion cycle is achieved by consulting a pre-stored information fusion synchronization index and its corresponding duration extension coefficient in a database. Specifically, the process involves: first, retrieving the corresponding duration extension coefficient from the database based on the information fusion synchronization index of the current multi-source sensor data within the data fusion cycle; then, multiplying this duration extension coefficient by the originally set fusion interval to calculate the extended fusion interval. The duration extension coefficient represents the proportional increase in the original fusion interval duration.

[0035] Extending the fusion interval allows for the acquisition of more temporally continuous multi-source sensor data, improving the completeness and richness of data acquisition and helping to alleviate the problem of low information fusion synchronization index caused by insufficient data synchronization in a short period. By extending the time window, the system can capture more sensor state changes and environmental dynamics, thereby improving the accuracy and stability of data fusion. In addition, more data samples help filter out occasional anomalies, enhance the robustness of the fusion results, ensure that subsequent decisions are based on a more sufficient and effective information foundation, and improve the overall performance of the system.

[0036] The secondary information fusion synchronization index of multi-source sensor data refers to the information fusion synchronization index obtained when the system extends the fusion interval and re-collects more data after failing the first fusion operation due to insufficient synchronization indicators, and then performs a second fusion evaluation on the newly added data. It reflects the consistency and coordination of the data collected after the delay in time and space, measures the matching and synchronization effect of multi-source sensor information in the second fusion cycle, and is used to judge whether the supplementary data has improved the fusion quality, thereby determining the validity of the data.

[0037] Specifically, the information fusion synchronization index of multi-source sensor data within the data fusion cycle is as follows: The fusion synchronization index set of multi-source sensor data includes the maximum spatial registration error value, the maximum temporal synchronization error value, and the data integrity ratio within the data fusion cycle. Spatial registration refers to aligning and matching data from different sensors in a spatial coordinate system so that they can accurately describe targets in the same physical space. The maximum spatial registration error value is the maximum deviation generated by all sensor data during spatial registration within the data fusion cycle. Within the data fusion cycle, each sensor collects data at its own frequency, using the coordinate transformation formula Pcar=R×P. sensor+T, where Pcar is the coordinate of the data point transformed to the vehicle's body coordinate system, Psensor is the coordinate of the data point in the sensor's coordinate system, R is the rotation matrix, and T is the translation vector, transforms each sensor data point from its own coordinate system to the vehicle's body coordinate system. A feature matching algorithm (such as nearest neighbor matching) is used to match data points from different sensors describing the same target. For each pair of matched data points, the Euclidean distance formula is used to calculate their spatial position deviation in the vehicle's body coordinate system. Within the data fusion period, the position deviations of all matched data points are statistically analyzed, and the maximum value is identified as the maximum spatial registration error. Time synchronization ensures that the data collected by different sensors are accurately aligned in time so as to correctly reflect the target's state changes at different times.The maximum time synchronization error is the maximum time deviation that occurs during the time synchronization process of all sensor data within the data fusion period. A precise timestamp is added to the data collected by each sensor to record the specific moment of data acquisition. The accuracy of the timestamp is determined by technicians based on actual application requirements, typically needing to reach the microsecond or even nanosecond level. Time synchronization algorithms are used to align the data from different sensors. Common time synchronization algorithms include Network Time Protocol (NTP) and Precise Time Protocol (PTP). These algorithms can achieve time synchronization between sensors through network communication. For each set of data from different sensors, their timestamps are compared, and the time difference is calculated. The sign of the time difference indicates the order of data acquisition. Within the data fusion period, the time deviation of all data sets is considered. Statistical analysis is performed to identify the maximum value, which is the maximum time synchronization error. The data integrity ratio refers to the ratio of the effective data successfully acquired and transmitted to the fusion center within the data fusion cycle to the expected data acquisition amount. Based on the design requirements of the data fusion system and the sampling frequency of the sensors, the expected data acquisition amount of each sensor within the data fusion cycle is calculated. After the data fusion cycle ends, the actual effective data acquired and transmitted to the fusion center by each sensor is statistically analyzed. The validity of the data can be determined by checking data integrity identifiers, checksums, etc. For each sensor, the actual data amount is divided by the expected data amount to obtain the data integrity ratio of that sensor. The average data integrity ratio of all sensors is then used to obtain the data integrity ratio of multi-source sensor data within the data fusion cycle.

[0038] The maximum spatial registration error should typically be less than 0.2 meters (20 centimeters) to ensure that the spatial alignment error of multi-source data is within the acceptable range for vehicle control; the maximum time synchronization error should be controlled within 50 microseconds, preferably within 20 microseconds, which is especially critical in high-speed reversing or fine obstacle avoidance scenarios; the data integrity ratio should be no less than 95%, meaning that most sensors should upload data completely within the fusion cycle.

[0039] The data fusion cycle refers to the fixed time interval for the fusion processing of multi-source sensor data in the system. This time interval is usually short to ensure that the fusion operation can reflect the latest changes in the surrounding environment and the semi-trailer's own status in real time and efficiently. Due to the reasonable setting of the cycle time, the data fusion process will not cause significant delays to subsequent intelligent decision-making, ensuring that the system can respond quickly in dynamic environments and continuously output accurate fusion results, thereby supporting the safe and stable operation of the semi-trailer.

[0040] Extract the spatial registration error value, the time synchronization error value, and the data integrity ratio from the database. The spatial registration error value is used to characterize the maximum acceptable value of the spatial registration error. The time synchronization error value is used to characterize the maximum acceptable value of the time synchronization error. The data integrity ratio is used to characterize the minimum value required for the data integrity ratio.

[0041] The measurement ratios are extracted from the database, and quantitative analysis is performed on the proportional relationships between the maximum spatial registration error and the defined spatial registration error, the maximum time synchronization error and the defined time synchronization error, and the data integrity ratio and the defined data integrity ratio within the data fusion cycle. This clarifies the degree of influence of the above proportional relationships on the information fusion synchronization index. Finally, the degree of influence of each proportional relationship is summarized and integrated to obtain the information fusion synchronization index.

[0042] The information fusion synchronization index of multi-source sensor data within the data fusion cycle is used to digitally represent the degree of information fusion synchronization of multi-source sensor data in time and space within the data fusion cycle. The specific expression is as follows:

[0043] ;

[0044] In the formula, SM is the information fusion synchronization index of multi-source sensor data within the data fusion period, A is the measurement ratio corresponding to the preset maximum spatial registration error value in the database, B is the measurement ratio corresponding to the preset maximum time synchronization error value in the database, C is the measurement ratio corresponding to the preset data integrity ratio in the database, MSZ is the maximum spatial registration error value of multi-source sensor data within the data fusion period, JDMSZ is the defined spatial registration error value, TSZ is the maximum time synchronization error value of multi-source sensor data within the data fusion period, JDTSZ is the defined time synchronization error value, DR is the data integrity ratio of multi-source sensor data within the data fusion period, and JDDR is the defined data integrity ratio.

[0045] The maximum spatial registration error of multi-source sensor data within the data fusion cycle reflects the alignment accuracy of different sensor data in spatial location. Increased spatial registration error leads to inconsistencies in the position of perceived information between sensors, thus affecting the accuracy of the fusion result. The maximum time synchronization error indicates the acquisition delay or asynchrony of multi-source sensor data in time. Increased time synchronization error will cause data temporal misalignment, reducing the timeliness and relevance of the fused data. The data integrity ratio measures the proportion of data effectively acquired within the fusion cycle. A lower data integrity ratio means missing data, reducing the amount of usable information. Increased spatial registration error and time synchronization error usually lead to a decrease in the data integrity ratio. The three factors influence each other and jointly determine the information fusion synchronization index of multi-source sensor data—the greater the spatial and temporal errors, the more missing data and the lower the synchronization index, and vice versa, thus affecting the effectiveness and stability of fusion.

[0046] The metric ratio corresponding to the maximum spatial registration error value is used to digitally characterize the impact of the ratio between the maximum spatial registration error value and the defined spatial registration error value within the fusion cycle on the information fusion synchronization index. The metric ratio corresponding to the maximum time synchronization error value is used to digitally characterize the impact of the ratio between the maximum time synchronization error value and the defined time synchronization error value on the information fusion synchronization index. Several metric ratio mapping tables are stored in the database, so the metric ratio corresponding to the maximum spatial registration error value, the metric ratio corresponding to the maximum time synchronization error value, and the metric ratio corresponding to the data integrity ratio can be directly queried from the database. The values ​​of all three are between 0 and 1.

[0047] Furthermore, the determination of whether to optimize multi-source sensor data involves the following process: Based on the information fusion synchronization index of multi-source sensor data within the data fusion period and the secondary information fusion synchronization index of multi-source sensor data, the growth rate of the information fusion synchronization index is obtained and compared with the preset defined index growth rate in the database; the defined index growth rate represents the lower limit of the information fusion synchronization index growth rate; the information fusion synchronization index growth rate is calculated by subtracting the information fusion synchronization index of multi-source sensor data within the data fusion period from the secondary information fusion synchronization index of multi-source sensor data, and then dividing the result by the information fusion synchronization index of multi-source sensor data within the data fusion period to obtain the final information fusion synchronization index growth rate.

[0048] If the information fusion synchronization index growth rate is less than the defined index growth rate, it is determined that no data optimization will be performed on the multi-source sensor data, and the multi-source sensor data will be marked as invalid data, and the data fusion operation will be cancelled; if the information fusion synchronization index growth rate is greater than or equal to the defined index growth rate, it is determined that data optimization will be performed on the multi-source sensor data.

[0049] If the information fusion synchronization index growth rate is lower than the defined index growth rate, it indicates that even with an increase in data volume, the fusion result has not achieved significant improvement. This may be due to abnormal data from the semi-trailer truck sensors, excessive environmental interference, or structural inconsistencies between data sources. In this case, continuing optimization will not only fail to effectively improve the fusion quality but may also increase the computational burden and delay decision-making. Therefore, when the growth rate is lower than the set threshold, the system will determine that the current fusion attempt is unacceptable, directly mark the multi-source sensor data of that round as invalid data, and cancel the fusion operation. This avoids transmitting low-quality data to subsequent decision-making modules and ensures the stability and reliability of the overall system.

[0050] Furthermore, the multi-source sensor data is optimized. The specific optimization process is as follows: Based on the secondary information fusion synchronization index of the multi-source sensor data, the frame interpolation interval reduction duration is matched from the database, thereby reducing and optimizing the frame interpolation interval duration of the multi-source sensor data. Then, the frame interpolation operation is re-performed on the multi-source sensor data. The core reason is to improve the alignment accuracy of the data in the time dimension, thereby improving the overall synchronization performance. The database stores a mapping table of information fusion synchronization index and frame interpolation interval reduction duration. By directly querying the secondary information fusion synchronization index of the multi-source sensor data in the database, the corresponding frame interpolation interval reduction duration can be obtained. Subtracting the frame interpolation interval reduction duration from the current frame interpolation interval duration yields the reduced and optimized frame interpolation interval duration.

[0051] After the frame interpolation operation is completed, the data fusion operation is re-executed to obtain the three information fusion synchronization indices of the multi-source sensor data and compare them three times with the information fusion synchronization threshold. This is to determine again whether the multi-source sensor data is marked as valid data. If the multi-source sensor data is marked as valid data, the data fusion operation is completed. If the multi-source sensor data is not marked as valid data, the multi-source sensor data is marked as invalid data, and the current data fusion operation is canceled.

[0052] The synchronization index of the three-stage information fusion of multi-source sensor data refers to the synchronization performance evaluation index obtained by performing data fusion again after the system performs frame interpolation optimization processing on the data after the first two rounds of data fusion attempts fail to meet the standard (i.e., the original fusion index and the secondary fusion index both fail to reach the set threshold).

[0053] Figure 3This is a schematic diagram of the decision control process provided in this embodiment of the invention. The system first generates a reference reversing trajectory and control commands, and further determines whether the existing reversing decision needs to be updated. If no update is needed, the original control commands are executed directly. If an update is needed, a new reference reversing trajectory is generated, and the validity of the trajectory is determined. The validity is determined based on whether the number of obstacles equals the defined number of obstacles. If the condition is met, the new trajectory is considered valid, and the system regenerates the control commands. If the number of obstacles is greater than the defined number of obstacles, the new trajectory is deemed invalid, and the system initiates a graded warning process. Next, it is determined whether there are dynamic obstacles in the new trajectory that are moving towards the vehicle. If so, the system will issue a level four warning and execute a braking operation. If there are no approaching dynamic obstacles, the system will assess the current vehicle speed and determine other warning levels accordingly, such as level one, level two, or level three warnings. At the same time, the system determines whether a visual command needs to be generated: if the condition is met, the system will generate a visual prompt of direction and speed to guide the vehicle back to the trajectory; otherwise, it will attempt to re-determine the trajectory. If the system cannot generate a valid reference reversing trajectory, a "cannot reverse" warning is executed.

[0054] In one specific embodiment, this invention accurately identifies reversing path deviations by comparing the abnormal coefficient of reversing decision execution with a threshold, promptly updates the reference trajectory, and verifies its feasibility, avoiding misleading control commands. By recognizing the number, state, and movement trends of obstacles, the system implements a tiered early warning mechanism from static waiting to dynamic obstacle avoidance, proactively intervening with braking or speed limiting to reduce collision risks. Simultaneously, the warning level is linked to visual prompts, improving the driver's perception and reaction speed to reversing risks, effectively ensuring the continuity and safety of semi-trailer truck reversing operations, especially in complex scenarios.

[0055] It needs to be explained that the aforementioned intelligent decision-making for the semi-trailer's reference reversing trajectory involves using a trajectory generation algorithm to intelligently determine this trajectory. This process first constructs a kinematic model of the semi-trailer (commonly a two- or three-wheeled constrained model) based on fused environmental perception results (such as obstacle distribution and road boundaries) and vehicle state information (such as current position, speed, steering angle, and vehicle orientation). Then, combining this model with the target reversing endpoint pose, a feasible reversing trajectory that satisfies constraints, ensures obstacle avoidance safety, and allows for continuous steering is generated using spline curve fitting methods (such as cubic Bezier curves or B-splines) or search-based path planning methods (such as the Hybrid A* algorithm).

[0056] It needs to be explained that the generation of control commands for the semi-trailer is achieved after the reference reversing trajectory is determined. Based on the vehicle's current state and kinematic model, the system calculates the specific control quantities that the vehicle should execute at each moment using a trajectory tracking algorithm. These control commands mainly include: steering angle commands, used to adjust the steering wheel or wheel angles to ensure the vehicle's posture continuously conforms to the reference trajectory; target speed commands, setting the reversing speed according to the trajectory curvature and path changes to ensure smooth reversing; acceleration / deceleration control signals, used to finely control the throttle or braking system to adjust speed; trailer posture coordination commands, dynamically adjusting the traction relationship based on the angle between the tractor and trailer to prevent trailer instability; and safety control commands triggered in emergency situations, such as emergency stop or automatic braking. Through the coordinated execution of these control commands, the semi-trailer can accurately and safely complete the reversing process along the reference trajectory.

[0057] Specifically, the process for determining whether to update the intelligent decision-making of the semi-trailer is as follows: analyze the reversing anomaly information of the semi-trailer, obtain the reversing decision execution anomaly coefficient of the semi-trailer within the monitoring period, and compare it with the reversing decision execution anomaly threshold stored in the database; the reversing decision execution anomaly threshold represents the maximum value of the reasonable range of the reversing decision execution anomaly coefficient.

[0058] If the abnormal coefficient of the semi-trailer's reversing decision execution during the monitoring period is less than the abnormal threshold for reversing decision execution, it is determined that the intelligent decision of the semi-trailer will not be updated; if the abnormal coefficient of the semi-trailer's reversing decision execution during the monitoring period is greater than or equal to the abnormal threshold for reversing decision execution, it is determined that the intelligent decision of the semi-trailer will be updated, a new intelligent decision will be made for the reference reversing trajectory of the semi-trailer, it will be marked as the new reference reversing trajectory, and it will be determined whether the new reference reversing trajectory is valid.

[0059] Specifically, the analysis process for the abnormal reversing decision execution coefficient of the semi-trailer truck within the monitoring period is as follows: The abnormal reversing information of the semi-trailer truck includes the total deviation value of its trajectory position, the speed fluctuation amplitude, and the total deviation value of its steering angle within the monitoring period. The total deviation value of the trajectory position is used to measure the degree of deviation between the actual driving trajectory of the semi-trailer truck and the preset reference trajectory within the monitoring period. The actual position coordinates of the semi-trailer truck at each moment are obtained in real time by positioning sensors (such as GPS sensors, LiDAR positioning modules, etc.) installed on the semi-trailer truck. Simultaneously, the system stores the coordinate positions on the reference reversing trajectory at the corresponding moments. At each monitoring moment, the distance deviation between the actual position coordinates and the corresponding position coordinates on the reference trajectory is calculated. This process is repeated for all moments within the entire monitoring period. The distance deviations at different times are accumulated to obtain the total trajectory position deviation value; the speed fluctuation amplitude is used to measure the degree of fluctuation of the semi-trailer's reversing speed within the monitoring period. It can be obtained by recording the speed value of the semi-trailer in real time within the monitoring period through speed sensors (such as wheel encoders, inertial navigation IMU, or GPS speed calculation), and averaging the values. The result is marked as the speed fluctuation amplitude; the total steering angle deviation value is used to measure the degree of deviation between the actual steering angle of the semi-trailer and the preset reference steering angle within the monitoring period. The steering angle sensor collects the steering angle in real time. At the same time, the system stores the expected steering angle at the corresponding time on the reference reversing trajectory. For each sampling time, the absolute value of the difference between the current steering angle and the expected steering angle is calculated. The calculation results for the entire monitoring period are accumulated to obtain the total steering angle deviation value.

[0060] Extract the total deviation value of the trajectory position, the amplitude of the speed fluctuation, and the total deviation value of the steering angle from the database; the total deviation value of the trajectory position represents the upper limit of the total deviation value of the trajectory position; the amplitude of the speed fluctuation represents the upper limit of the speed fluctuation amplitude; and the total deviation value of the steering angle represents the upper limit of the total deviation value of the steering angle.

[0061] The monitoring cycle refers to the fixed time interval set by the system for continuously monitoring and evaluating the reversing status of a semi-trailer truck. This cycle is usually short to achieve high-frequency sampling and rapid response of the vehicle's operating status. By collecting and analyzing multi-source sensor data in real time within each monitoring cycle, the system can promptly detect abnormal changes during the reversing process, providing a rapid and effective basis for intelligent decision-making updates and safety control.

[0062] The system acquires the information fusion synchronization index of multi-source sensor data within the monitoring period and matches the decision anomaly increments from the database. The information fusion synchronization index of multi-source sensor data within the monitoring period refers to the information fusion synchronization index of all multi-source sensor data used by the system within a complete monitoring period.

[0063] It should be explained that the decision anomaly increment refers to an incremental parameter used to increase the numerical value of the reversing decision execution anomaly coefficient. The specific matching process is as follows: A mapping relationship between the information fusion synchronization index and the corresponding decision anomaly increment is pre-stored in the database. By querying the information fusion synchronization index of multi-source sensor data within the monitoring period in the database, the corresponding decision anomaly increment can be retrieved, thereby incrementally supplementing the reversing decision execution anomaly coefficient.

[0064] The measurement ratios are extracted from the database, and the proportional relationships between the total deviation of the trajectory position and the defined total deviation of the trajectory position, the proportional relationship between the speed fluctuation amplitude and the defined speed fluctuation amplitude, and the proportional relationship between the total deviation of the steering angle and the defined total deviation of the steering angle are quantified. This clarifies the degree of influence of the above proportional relationships on the abnormal coefficient of reversing decision execution. Finally, the degree of influence of each proportional relationship and the decision abnormality increment are summarized and integrated to obtain the abnormal coefficient of reversing decision execution.

[0065] The anomaly coefficient of semi-trailer truck reversing decision execution during the monitoring period is used to digitally characterize the degree of fluctuation anomaly when semi-trailer trucks execute reversing decisions during the monitoring period. The specific expression is as follows:

[0066] ;

[0067] In the formula, DEP is the reversing decision execution anomaly coefficient of the semi-trailer truck within the monitoring period, W is the measurement ratio corresponding to the preset total deviation value of trajectory position in the database, R is the measurement ratio corresponding to the preset speed fluctuation amplitude in the database, T is the measurement ratio corresponding to the preset total deviation value of steering angle in the database, SUSM is the decision anomaly increment, TDP is the total deviation value of trajectory position of the semi-trailer truck within the monitoring period, JDTDP is the defined total deviation value of trajectory position, SMP is the speed fluctuation amplitude of the semi-trailer truck within the monitoring period, JDSMP is the defined speed fluctuation amplitude, SAV is the total deviation value of steering angle of the semi-trailer truck within the monitoring period, and JDSAV is the defined total deviation value of steering angle.

[0068] During the monitoring period, the information fusion synchronization index of multi-source sensor data directly affects the selection of decision anomaly increments. The lower the information fusion synchronization index, the worse the consistency and reliability of the fused data, and the larger the decision anomaly increment value matched by the system from the database, which is used to enhance the sensitivity of anomaly assessment. The total trajectory position deviation reflects the degree of deviation between the actual driving trajectory of the vehicle and the reference reversing trajectory. Affected by the fusion quality and the path tracking execution effect, the trajectory deviation value usually increases when the information fusion synchronization index decreases or the control execution is unstable. The speed fluctuation amplitude represents the speed stability of the vehicle during reversing. Affected by the path complexity and vehicle response delay, increased speed fluctuation will lead to a decrease in trajectory control accuracy, which will further aggravate the position deviation. The total steering angle deviation measures the deviation between the vehicle steering command and the actual steering action. Steering error is often caused by control system lag or increased environmental disturbances, which will also have a cumulative effect on trajectory accuracy. The above four parameters are interrelated, thereby triggering the system to dynamically update intelligent decisions and control risks.

[0069] The measurement ratio corresponding to the total deviation value of the trajectory position is used to digitally represent the influence of the ratio between the total deviation value of the trajectory position and the defined total deviation value of the trajectory position on the abnormal coefficient of the reversing decision execution; the measurement ratio corresponding to the speed fluctuation amplitude is used to digitally represent the influence of the ratio between the speed fluctuation amplitude and the defined speed fluctuation amplitude on the abnormal coefficient of the reversing decision execution; the measurement ratio corresponding to the total deviation value of the steering angle is used to digitally represent the influence of the ratio between the total deviation value of the steering angle and the defined total deviation value of the steering angle on the abnormal coefficient of the reversing decision execution. The database stores several measurement ratio value mapping tables, so the measurement ratio values ​​corresponding to the total deviation value of the trajectory position, the speed fluctuation amplitude, and the steering angle can be directly queried from the database. The values ​​of all three are between 0 and 1.

[0070] Furthermore, the validity of the new reference reversing trajectory is determined through the following process: The number of obstacles in the new reference reversing trajectory is obtained. If the number of obstacles in the reference reversing trajectory equals the defined number of obstacles, the new reference reversing trajectory is deemed valid, and the semi-trailer control commands are regenerated. The defined number of obstacles is usually set to zero, meaning that when determining the validity of the new reference reversing trajectory, the system expects that there are no obstacles on the trajectory path that could affect the safe reversing of the semi-trailer. Here, "obstacles" specifically refer to objects within the reference reversing trajectory range of the semi-trailer that pose a potential collision risk, including stationary or moving vehicles, pedestrians, roadside facilities, etc.—any entity that may interfere with or hinder the safe reversing of the vehicle. The system acquires environmental information through multi-source sensors (such as LiDAR, cameras, millimeter-wave radar, etc.), and combines sensor data fusion and target detection algorithms to identify and locate obstacles in real time. The identification process determines whether an obstacle is within a danger zone based on the spatial overlap relationship between the obstacle and the reference reversing trajectory and a distance threshold, thereby determining whether it should be included in the obstacle count to assess the safety and validity of the trajectory.

[0071] If the number of obstacles in the reference reversing trajectory is greater than the defined number of obstacles, the new reference reversing trajectory is determined to be invalid. At the same time, the status of each obstacle in the new reference reversing trajectory is obtained, thereby providing graded warnings for the reversing process of the semi-trailer truck.

[0072] Furthermore, a tiered warning system is implemented for the reversing process of the semi-trailer truck. The specific warning process is as follows: if all obstacles in the new reference reversing trajectory are stationary, the shortest distance between the semi-trailer truck and the obstacle, as well as the reversing speed of the semi-trailer truck, are obtained. This allows for the analysis of the intervention waiting time of the semi-trailer truck, which is then compared with a preset intervention waiting time reference range in the database. The intervention waiting time refers to the shortest distance between the semi-trailer truck and the obstacle divided by the reversing speed of the semi-trailer truck, i.e., the shortest time required for the vehicle to maintain a safe distance from the obstacle without taking additional actions. The intervention waiting time reference range is a numerical range used to distinguish the warning levels. The shortest distance between the semi-trailer truck and the obstacle can be obtained through lidar.

[0073] If the intervention waiting time of the semi-trailer exceeds the maximum value of the intervention waiting time reference range, a Level 1 warning will be issued for the semi-trailer's reversing process. At this time, the distance between the vehicle and the obstacle is relatively far, the reaction time is sufficient, and it is a low-risk state. The system will remind the driver to pay attention through the Level 1 warning.

[0074] If the intervention waiting time of the semi-trailer truck falls within the reference range for intervention waiting time, a level-two warning will be issued for the semi-trailer truck's reversing process. If the intervention waiting time is within the normal safe range but the risk has increased, the system will issue a level-two warning to remind the driver to pay more attention and prepare to take measures.

[0075] If the intervention waiting time of the semi-trailer is less than the minimum value of the intervention waiting time reference range, a level three warning will be issued for the semi-trailer's reversing process. If the reaction time is significantly insufficient and the vehicle is in a high-risk state, the system will issue a level three warning, prompting the driver to take evasive measures immediately, or even prepare for emergency braking.

[0076] Level 1 Warning (Low Risk Alert): A green warning message is displayed on the instrument panel or central control screen, accompanied by a slight audible alert, reminding the driver to remain alert as the environment is relatively safe. Level 2 Warning (Medium Risk Alert): A yellow warning sign is displayed on the screen, accompanied by a moderate-volume warning sound, and a slight vibration of the steering wheel or seat, enhancing driver alertness. Level 3 Warning (High Risk Alert): A flashing red warning message is displayed, accompanied by a high-decibel warning sound, and a strong vibration of the steering wheel or seat, prompting the driver to take immediate evasive action.

[0077] If there are moving obstacles in the new reference reversing trajectory, the movement direction vector of each moving obstacle is obtained. If the movement direction of a moving obstacle is towards the semi-trailer, it means that in the reversing environment, some moving obstacles are detected, and their movement trajectory and speed vector point to the position of the semi-trailer. This indicates that these obstacles are approaching the vehicle. In this case, the semi-trailer is braked and a four-level warning is issued. Level four warning (extremely high risk, automatic braking intervention): The system automatically starts braking, accompanied by a continuous red warning light and a sharp alarm sound, to ensure that the driver and surrounding personnel are highly alert and to prevent a collision.

[0078] If there are no moving obstacles moving towards the semi-trailer, the minimum speed among all moving obstacles is obtained and marked as the maximum reversing speed of the semi-trailer. If the reversing speed of the semi-trailer is less than its maximum reversing speed, a level 5 warning is issued. The system automatically decelerates and displays a red overspeed warning, accompanied by continuous warning sounds, to urge the driver to control the reversing speed and ensure safety.

[0079] If the semi-trailer's reversing speed is greater than or equal to its maximum reversing speed, the reversing speed will be reduced to the minimum reversing speed preset by the technicians, and an overspeed warning will be issued, displaying a red overspeed warning and accompanied by continuous warning sounds to urge the driver to control the reversing speed and ensure safety.

[0080] When the new reference reversing trajectory is invalid, a visual instruction is generated based on the reversing decision execution anomaly coefficient and reversing decision execution anomaly threshold of the semi-trailer truck within the monitoring period.

[0081] Specifically, the process for determining whether to generate a visualization instruction is as follows: the abnormal coefficient of the semi-trailer's reversing decision execution within the monitoring period is differentiated from the abnormal threshold of the reversing decision execution. The result is marked as the abnormal difference value of the reversing decision execution and compared with the preset defined difference value in the database. The abnormal difference value of the reversing decision execution refers to the abnormal threshold of the reversing decision execution minus the abnormal coefficient of the semi-trailer's reversing decision execution within the monitoring period. The defined difference value represents the maximum value within the reasonable range of the abnormal difference value of the reversing decision execution.

[0082] If the abnormal difference value of the reversing decision execution is less than the defined difference value, a visual instruction is generated to provide visual prompts on the semi-trailer's driving direction and speed, thereby guiding the semi-trailer back to the reference reversing trajectory. If the abnormal difference value of the reversing decision execution is greater than or equal to the defined difference value, a new intelligent decision is made for the semi-trailer's reference reversing trajectory. If a valid reference reversing trajectory cannot be determined for the semi-trailer, a reversing failure warning is issued.

[0083] When the abnormal difference value in the reversing decision execution is less than the preset threshold, it indicates that the vehicle's deviation from the reference reversing trajectory is minor. The system can generate visual instructions to provide clear directional and speed cues to the driver or automatic control system, helping the vehicle adjust its posture in time and smoothly return to the original reference trajectory, ensuring the continuity and safety of the reversing process. Conversely, when the abnormal difference value is greater than or equal to the threshold, it indicates that the vehicle's deviation is significant and exceeds the range of simple adjustments. The system needs to re-perform intelligent path planning and generate a new reference reversing trajectory to adapt to the current environment and vehicle status. If an effective new trajectory cannot be generated, a reversing failure warning is triggered, reminding relevant personnel to take further measures to avoid reversing accidents.

[0084] The process of generating visual instructions is based on real-time collected vehicle position, attitude, and speed data. It calculates the deviation between the vehicle's current state and a reference reversing trajectory, including lateral and heading angle deviations. The system utilizes a vehicle kinematics model and trajectory tracking algorithm to calculate the steering angle and speed adjustments needed to correct these deviations. These adjustments are then transformed into intuitive visual elements, such as directional arrows, trajectory deviation indicators, and speedometers, accompanied by warning colors and prompts. These are presented to the driver in real-time via the in-vehicle display or head-up display, helping them quickly understand and adjust their actions. The entire process relies on fused multi-source sensor data and an intelligent decision-making module to ensure that the instructions accurately reflect the vehicle's deviation from the ideal trajectory, thereby improving the safety and accuracy of reversing.

[0085] In one specific embodiment, the present invention provides an intelligent reversing decision and control system for semi-trailers based on multi-source sensor data fusion. Through a data acquisition module, environmental and status information of the semi-trailer is collected using multi-source sensors, forming multi-source sensor data. This data is first transmitted to a buffer and processed uniformly according to a set fusion interval. The data fusion module projects various sensor data onto the semi-trailer's coordinate system, improving spatial matching accuracy through coordinate transformation. A fusion synchronization index set is introduced to evaluate the consistency and timeliness of the data from each sensor, ensuring that fusion is performed only when the data is valid, avoiding error propagation and information conflicts, and significantly improving the spatial consistency and temporal coordination of the data. The high-quality fused data is input to the decision optimization module, which intelligently plans a reference reversing trajectory and generates control commands. Simultaneously, the decision is dynamically updated based on the vehicle's status, achieving high-precision reversing control. This system's fusion mechanism enhances the robustness and accuracy of environmental perception, making it particularly suitable for reversing scenarios involving complex structures and numerous dynamic disturbances in semi-trailer vehicles.

[0086] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0087] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A semi-trailer intelligent reversing decision and control system based on multi-source sensor data fusion, characterized in that, include: The data acquisition module is used to acquire environmental information and status information of the semi-trailer through multi-source sensors, and comprehensively label them as multi-source sensor data. A data buffer is set up, and the multi-source sensor data is first transmitted to the data buffer. According to the preset fusion interval, the multi-source sensor data in the data buffer is fused. The data fusion module is used to project multi-source sensor data onto the semi-trailer body coordinate system through coordinate transformation during the data fusion process. It collects and evaluates the fusion synchronization index set of multi-source sensor data to determine whether the multi-source sensor data is marked as valid data. If the multi-source sensor data is marked as valid data, the data fusion operation is completed. If the multi-source sensor data is marked as invalid data, the data fusion operation is canceled. The decision optimization module is used to intelligently determine the reference reversing trajectory of the semi-trailer after the data fusion operation is completed, generate control commands for the semi-trailer, analyze the abnormal reversing information of the semi-trailer, and thus determine whether to update the intelligent decision of the semi-trailer. The specific determination process for whether to update the intelligent decision-making of the semi-trailer truck is as follows: Analyze the reversing anomaly information of semi-trailer trucks to obtain the reversing decision execution anomaly coefficient of semi-trailer trucks within the monitoring period, and compare it with the reversing decision execution anomaly threshold stored in the database; If the reversing decision execution anomaly coefficient of the semi-trailer is less than the reversing decision execution anomaly threshold during the monitoring period, it is determined that the intelligent decision of the semi-trailer will not be updated. If the abnormal coefficient of the semi-trailer's reversing decision execution within the monitoring period is greater than or equal to the abnormal threshold of the reversing decision execution, it is determined that the intelligent decision of the semi-trailer should be updated, a new intelligent decision should be made for the semi-trailer's reference reversing trajectory, which is marked as the new reference reversing trajectory, and it is determined whether the new reference reversing trajectory is valid. The specific analysis process for the abnormal coefficient of the semi-trailer's reversing decision execution during the monitoring period is as follows: The abnormal reversing information of the semi-trailer includes the total deviation of the semi-trailer's trajectory position during the monitoring period, the speed fluctuation amplitude of the semi-trailer during the monitoring period, and the total deviation of the semi-trailer's steering angle during the monitoring period. Extract the total deviation value of the defined trajectory position, the defined speed fluctuation amplitude, and the defined total deviation value of the steering angle from the database; Obtain the information fusion synchronization index of multi-source sensor data within the monitoring period, and match the decision anomaly increment from the database; The measurement ratio values ​​are extracted from the database, and the proportional relationship between the total deviation value of the trajectory position and the defined total deviation value of the trajectory position, the proportional relationship between the speed fluctuation amplitude and the defined speed fluctuation amplitude, and the proportional relationship between the total deviation value of the steering angle and the defined total deviation value of the steering angle are quantified. In this way, the influence of the above proportional relationships on the abnormal coefficient of reversing decision execution is clarified. Finally, the influence of each proportional relationship and the decision abnormality increment are summarized and integrated to obtain the abnormal coefficient of reversing decision execution. The abnormal coefficient of the semi-trailer's reversing decision execution during the monitoring period is used to digitally characterize the degree of fluctuation abnormality when the semi-trailer executes the reversing decision during the monitoring period. The set of fusion synchronization indicators for multi-source sensor data includes the maximum spatial registration error value, the maximum temporal synchronization error value, and the data integrity ratio of multi-source sensor data within the data fusion period; the information fusion synchronization index of multi-source sensor data within the data fusion period is used to digitally characterize the degree of information fusion synchronization of multi-source sensor data in time and space within the data fusion period.

2. The intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion according to claim 1, characterized in that: The specific process for determining whether multi-source sensor data is marked as valid data is as follows: The fusion synchronization index set of multi-source sensor data is evaluated to obtain the information fusion synchronization index of multi-source sensor data within the data fusion cycle, and compared with the preset information fusion synchronization threshold in the database. If the information fusion synchronization index of multi-source sensor data within the data fusion cycle is greater than or equal to the information fusion synchronization threshold, then the multi-source sensor data will be marked as valid data. If the information fusion synchronization index of the multi-source sensor data within the data fusion period is less than the information fusion synchronization threshold, the fusion interval is extended based on the information fusion synchronization index of the multi-source sensor data within the data fusion period. The multi-source sensor data is then reacquired, and the data fusion operation is performed again to obtain a secondary information fusion synchronization index of the multi-source sensor data. This index is then compared with the information fusion synchronization threshold to determine whether the multi-source sensor data is marked as valid data. If the multi-source sensor data is marked as valid data, the data fusion operation is completed. If the multi-source sensor data is not marked as valid data, it is determined whether to optimize the multi-source sensor data.

3. The intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion according to claim 2, characterized in that: The specific process for determining whether to perform data optimization on multi-source sensor data is as follows: Based on the information fusion synchronization index of multi-source sensor data within the data fusion cycle and the secondary information fusion synchronization index of multi-source sensor data, the growth rate of the information fusion synchronization index is obtained and compared with the preset definition index growth rate in the database. If the information fusion synchronization index growth rate is less than the defined index growth rate, it is determined that no data optimization will be performed on the multi-source sensor data, and the multi-source sensor data will be marked as invalid data, and the data fusion operation will be cancelled. If the information fusion synchronization index growth rate is greater than or equal to the defined index growth rate, then it is determined that data optimization should be performed on the multi-source sensor data.

4. The intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion according to claim 3, characterized in that: The data optimization process for multi-source sensor data is as follows: Based on the secondary information fusion synchronization index of multi-source sensor data, the frame interpolation interval reduction time is matched from the database, thereby reducing and optimizing the frame interpolation interval time of multi-source sensor data, and then re-performing the frame interpolation operation on the multi-source sensor data. After the frame interpolation operation is completed, the data fusion operation is re-executed to obtain the three information fusion synchronization indices of the multi-source sensor data and compare them three times with the information fusion synchronization threshold. This is to determine again whether the multi-source sensor data is marked as valid data. If the multi-source sensor data is marked as valid data, the data fusion operation is completed. If the multi-source sensor data is not marked as valid data, the multi-source sensor data is marked as invalid data, and the current data fusion operation is canceled.

5. The intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion according to claim 1, characterized in that: The information fusion synchronization index of the multi-source sensor data during the data fusion cycle is specifically as follows: Extract and define spatial registration error values, temporal synchronization error values, and data integrity ratios from the database; The measurement ratios are extracted from the database, and quantitative analysis is performed on the proportional relationships between the maximum spatial registration error and the defined spatial registration error, the maximum time synchronization error and the defined time synchronization error, and the data integrity ratio and the defined data integrity ratio within the data fusion cycle. This clarifies the degree of influence of the above proportional relationships on the information fusion synchronization index. Finally, the degree of influence of each proportional relationship is summarized and integrated to obtain the information fusion synchronization index.

6. The intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion according to claim 1, characterized in that: The specific process for determining whether the new reference reversing trajectory is valid is as follows: Obtain the number of obstacles in the new reference reversing trajectory. If the number of obstacles in the reference reversing trajectory is equal to the number of defined obstacles, the new reference reversing trajectory is deemed valid, and the semi-trailer control command is regenerated. If the number of obstacles in the reference reversing trajectory is greater than the defined number of obstacles, the new reference reversing trajectory is determined to be invalid. At the same time, the status of each obstacle in the new reference reversing trajectory is obtained, thereby providing graded warnings for the reversing process of the semi-trailer truck.

7. The intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion according to claim 6, characterized in that: The aforementioned tiered warning system for the reversing process of semi-trailer trucks is as follows: If all obstacles in the new reference reversing trajectory are stationary, the shortest distance between the semi-trailer and the obstacle and the reversing speed of the semi-trailer are obtained, thereby analyzing the intervention waiting time of the semi-trailer and comparing it with the preset intervention waiting time reference range in the database. If the intervention waiting time of the semi-trailer exceeds the maximum value of the intervention waiting time reference range, a level one warning will be issued for the reversing process of the semi-trailer. If the intervention waiting time of the semi-trailer truck falls within the intervention waiting time reference range, a level-two warning will be issued for the semi-trailer truck's reversing process. If the intervention waiting time of the semi-trailer is less than the minimum value of the intervention waiting time reference range, a level three warning will be issued for the reversing process of the semi-trailer. If there are moving obstacles in the new reference reversing trajectory, the motion direction vector of each moving obstacle is obtained. If the moving direction of the moving obstacle is towards the semi-trailer, the semi-trailer is braked and a level four warning is issued. If there are no moving obstacles and their direction of movement is towards the semi-trailer, then the minimum value of the moving speed of each moving obstacle is obtained and marked as the maximum reversing speed of the semi-trailer. If the reversing speed of the semi-trailer is less than the maximum reversing speed of the semi-trailer, then a level 5 warning is issued. If the semi-trailer's reversing speed is greater than or equal to its maximum reversing speed, the reversing speed of the semi-trailer will be reduced and an overspeed warning will be issued. When the new reference reversing trajectory is invalid, a visual instruction is generated based on the reversing decision execution anomaly coefficient and reversing decision execution anomaly threshold of the semi-trailer truck within the monitoring period.

8. The intelligent reversing decision and control system for semi-trailer trucks based on multi-source sensor data fusion according to claim 7, characterized in that: The specific process for determining whether to generate a visualization command is as follows: The abnormal coefficient of reversing decision execution and the abnormal threshold of reversing decision execution of semi-trailer trucks within the monitoring period are differentiated, and the processing result is marked as the abnormal difference value of reversing decision execution, and compared with the preset definition difference value in the database. If the abnormal difference value of the reversing decision execution is less than the defined difference value, a visual instruction is generated to provide visual prompts on the driving direction and speed of the semi-trailer, so as to bring the semi-trailer back to the reference reversing trajectory. If the abnormal difference value of the reversing decision execution is greater than or equal to the defined difference value, the intelligent system will re-determine the reference reversing trajectory of the semi-trailer. If a valid reference reversing trajectory for the semi-trailer cannot be determined, a reversing failure warning will be issued.

Images