Unmanned express delivery vehicle control system and method based on operation state perception

Abstract

The invention discloses an unmanned express delivery vehicle control system and method based on operation state perception, and relates to the field of unmanned express delivery vehicle control systems. The method comprises the following steps: acquiring vehicle-mounted data in a driving period through a vehicle-mounted sensor array, performing time alignment and validity marking on the vehicle-mounted data to obtain a data quality label, constructing an evidence item set based on the data quality label, generating a sub-portrait label set according to the evidence item set, and displaying the sub-portrait label set in the driving period. And generating a portrait parameter set based on the sub-portrait label set through a preset priority rule and a persistence judgment strategy. Although the scheme has excellent performance in the aspects of improving system adaptability, safety, dynamic decision-making capability, reviewable performance and resource utilization efficiency, the problems of sensor faults, data loss, real-time data processing bottleneck, label interpretation errors and the like may influence the real-time response and decision-making precision of the system in a complex or extreme environment. Therefore, the overall performance and stability are reduced, especially under the condition of high load or low signal.

Description

Technical Field

[0001] This invention relates to the field of unmanned delivery vehicle control systems, specifically to an unmanned delivery vehicle control system and method based on operational status perception. Background Technology

[0002] With the continuous development of intelligent technology, unmanned delivery vehicles, as an important component of automated logistics, have been widely used in modern urban delivery. Unmanned delivery vehicles possess advantages such as high efficiency and environmental friendliness, significantly improving the speed and quality of logistics delivery. However, due to the complex and ever-changing environment and the stability issues of system hardware and software, unmanned delivery vehicles still face many challenges during operation, especially in ensuring their safety and stability. Currently, existing unmanned delivery vehicle control systems mainly rely on single sensors or health indicators to assess the vehicle's status and make control decisions accordingly. Traditional methods typically focus only on the health status of a single subsystem, such as the accuracy of the positioning system or the battery level, while ignoring the synergistic effects between various subsystems. As unmanned delivery vehicle technology continues to develop, single-indicator control schemes are insufficient to cope with the uncertainties in complex environments, the system's ability to respond to abnormal events is poor, and control strategies often lack dynamic adjustment capabilities. While existing technologies offer significant advantages in enhancing the safety, dynamic decision-making capabilities, auditability, and resource utilization efficiency of unmanned delivery vehicle control systems, some potential technical shortcomings remain in practical applications. First, the system heavily relies on the health of multiple sensors and subsystems. If some sensors malfunction or lose data, especially in low-signal environments, the system may fail to obtain critical data in a timely manner, affecting the stability of the entire system. Particularly in cases of sensor failure or data transmission obstruction, the system's inability to quickly adjust its decision-making strategy may delay response time, preventing the vehicle from avoiding obstacles or adjusting its path in time, thus increasing safety risks. Second, although the review process can effectively address fluctuations in abnormal data, in extremely... In extreme weather conditions or electromagnetic interference, the reliability of sensors and the accuracy of data may be greatly reduced. This can affect the system's timely handling of abnormal situations and decision optimization, increasing the risk of system errors in complex environments. In addition, although multi-level data processing and resource filtering improve data processing efficiency, the system may encounter computational bottlenecks in scenarios with high-density obstacles, complex path planning, or real-time high-frequency data processing, leading to response delays or even the inability to update control strategies in a timely manner. This affects the vehicle's real-time response capability in rapidly changing environments. Therefore, although this invention has made optimizations in many aspects, in actual deployment, it is still necessary to pay attention to the system's adaptability to abnormal environments, the processing power of computing resources, and the stability of data transmission. Summary of the Invention

[0003] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide a control system and method for unmanned delivery vehicles based on operational status perception, so as to solve the above-mentioned technical problems.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a control method for unmanned delivery vehicles based on operational status perception, comprising: acquiring vehicle data within a driving cycle through an onboard sensor array; performing time alignment and validity marking on the onboard data to obtain data quality labels; constructing an evidence item set based on the data quality labels; generating a sub-portrait label set based on the evidence item set; generating a portrait parameter set based on the sub-portrait label set through preset priority rules and persistence judgment strategies; generating an operational status portrait code through structured encoding based on the portrait parameter set and the sub-portrait label set; generating a control constraint package and determining a control mode based on the operational status portrait code; constraining path planning, trajectory generation, and execution control based on the control mode; outputting vehicle control commands; executing vehicle control commands and synchronously obtaining execution feedback; performing closed-loop verification on the operational status portrait code; triggering a verification process when preset verification conditions are met; and updating the operational status portrait code and control constraint package based on the verification results.

[0005] The present invention is further configured such that the vehicle data is time-aligned using either cache alignment or maintain alignment, and data exceeding a preset time threshold is marked as expired. The time-aligned vehicle data is then marked for validity using at least one of range verification, jump verification, and continuity verification. Data quality labels are generated based on the expired and valid labels.

[0006] The present invention is further configured such that, based on the data quality label, a preset subsystem mapping table is used to group the vehicle data into subsystems, and evidence rules are constructed and evidence judgments are performed for each group of data based on a preset evidence rule configuration table. The evidence judgment results are discretized and output to generate evidence conclusions. Based on the data quality label and combined with the evidence conclusions, stability judgments are performed within a preset sliding time window to generate evidence strength labels. Finally, the evidence conclusions and evidence strength labels are combined to form a set of evidence items.

[0007] The present invention is further configured such that the evidence items in the evidence item set are merged by subsystem grouping, an initial level of the corresponding sub-portrait tag is generated by a preset evidence merging strategy, the initial level is mapped to the sub-portrait level through a preset level mapping table, and a continuous determination is performed on the sub-portrait level to generate a sub-portrait tag set.

[0008] The present invention is further configured such that the profile parameter set includes an overall profile level, a dominant risk source, and a credibility level; the dominant risk source is determined in the sub-profile tag set according to a preset priority rule; the sub-profile tag set is gating processed according to a preset persistence judgment strategy to generate an overall profile level; the credibility level is generated based on data quality tags and the consistency judgment of the evidence item set; and the overall profile level, dominant risk source, credibility level, and sub-profile tag set are structured and encoded to generate an operational status profile code.

[0009] The present invention is further configured such that, based on the total profile level and dominant risk source in the profile parameter set, a control constraint template is selected from the preset control constraint template library using at least one of the following methods: table lookup matching, rule matching, and state machine switching. A control constraint package is generated from the control constraint template based on the running status profile code and using parameter instantiation. A control mode is selected according to the total profile level. The control constraint package and the control mode are associated through a preset constraint association table.

[0010] The present invention is further configured such that, according to the control constraint package, the candidate routes and candidate maneuvers are restricted in the path planning process, and a planning result is generated using a preset constrained path planning method. The planning result is used as a reference path or constraint input for trajectory generation. Boundary constraints are applied to the trajectory generation process according to the control constraint package, and a target trajectory is generated using a preset trajectory generation method. The target trajectory is used as the tracking target for execution control. Control quantity constraint processing is performed on the execution control process according to the control constraint package, and vehicle control commands are output based on the target trajectory.

[0011] The present invention is further configured such that, when executing vehicle control commands, collecting execution feedback, aligning the vehicle control commands and execution feedback in a timely manner, performing consistency verification of commands and responses based on the alignment results and obtaining consistency results, when the consistency results meet preset deviation conditions, updating the evidence item set and the evidence items or sub-portrait tags associated with the deviation, and regenerating the running status portrait code and control constraint package accordingly.

[0012] The present invention is further configured such that, in the step of parsing the running status profile code to obtain the credibility level, at least one of the credibility level, data quality label and running status profile code historical sequence is used to determine whether the preset review conditions are met, and when the preset review conditions are met, the review process is triggered to obtain review data, and the evidence item set, sub-profile label set and running status profile code are reconstructed based on the review data, and the control constraint package is regenerated based on the updated running status profile code.

[0013] This invention also provides a control system for unmanned delivery vehicles based on operational status perception. The system includes: a data acquisition and labeling module: acquiring vehicle data during the driving cycle through an onboard sensor array, performing time alignment and validity labeling on the onboard data to obtain data quality labels; a sub-profile generation module: constructing an evidence item set based on the data quality labels, generating a sub-profile label set based on the evidence item set; a profile code synthesis module: generating a profile parameter set based on the sub-profile label set through preset priority rules and persistence judgment strategies, generating an operational status profile code based on the profile parameter set and the sub-profile label set through structured encoding, wherein the profile parameter set includes the overall profile level, dominant risk source, and credibility level; a profile-driven control module: generating a control constraint package and determining the control mode based on the operational status profile code, constraining path planning, trajectory generation, and execution control based on the control mode, and outputting vehicle control commands; and a closed-loop verification module: executing vehicle control commands and synchronously acquiring execution feedback, performing closed-loop verification on the operational status profile code, triggering a verification process when preset verification conditions are met, and updating the operational status profile code and control constraint package based on the verification results.

[0014] This invention provides a control system and method for unmanned delivery vehicles based on operational status perception. The method acquires vehicle-mounted data during the driving cycle through an onboard sensor array, performs time alignment and validity marking on the data to obtain data quality labels, constructs an evidence item set based on the data quality labels, generates a sub-profile label set based on the evidence item set, generates a profile parameter set based on the sub-profile label set using preset priority rules and persistence judgment strategies, generates an operational status profile code based on the profile parameter set and the sub-profile label set using structured encoding, generates a control constraint package based on the operational status profile code and determines the control mode, constrains path planning, trajectory generation, and execution control based on the control mode, outputs vehicle control commands, executes the vehicle control commands and simultaneously obtains execution feedback, performs closed-loop verification of the operational status profile code, triggers a verification process when preset verification conditions are met, and updates the operational status profile code and control constraint package based on the verification results. The beneficial effects include: Based on the operational status perception and multi-level processing generated by multi-source data, the operational status profile code can comprehensively and accurately assess the current operational status of unmanned delivery vehicles. The system adjusts the control mode according to the health status of each subsystem and generates corresponding control constraint packages based on the profile code, optimizing the vehicle's safety and adaptability in complex environments. Through continuous monitoring and flexible response, the robustness of the vehicle in dynamic environments is improved.

[0015] By analyzing and verifying the operational status profile code in real time, the system can dynamically adjust the control strategy based on the vehicle's execution feedback and environmental changes. When an anomaly or deviation is detected, a review process is triggered to update the system's evidence items and sub-profile labels, ensuring that the system's decisions are based on the latest and most accurate operational status. This dynamic decision-making mechanism can effectively respond to emergencies and improve the autonomous decision-making capabilities of unmanned delivery vehicles.

[0016] The operational status profile code clearly identifies the health status and risk sources of each subsystem, generating an interpretable set of sub-profile tags. This allows each control decision to be traced back to the specific source of the status. Combined with the dominant risk source and credibility level, it provides auditors and operators with clear decision-making basis, enhancing the auditability and transparency of the system. This helps to strengthen the compliance of unmanned delivery vehicles and facilitates supervision and auditing.

[0017] By generating data quality labels and designing a reasonable data grouping mechanism, the system can perform validity and freshness screening during the data acquisition stage, avoiding the transmission and processing of redundant data. Through hierarchical processing of evidence items and sub-profile labels, the system can reduce the consumption of computing resources by irrelevant data without affecting decision accuracy. By combining data quality labels with evidence strength labels and performing stability judgment within a sliding time window, the system solves the problem of misjudgment when a single sensor fails, and improves the robustness of the system in abnormal environments.

[0018] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0019] 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 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. In the drawings: Figure 1 A flowchart illustrating an exemplary embodiment of the present invention is shown below, illustrating a method for controlling an unmanned delivery vehicle based on operational status perception. Figure 2 This is a schematic diagram illustrating the structure of an unmanned delivery vehicle control system based on operational status perception, as an exemplary embodiment of the present invention. Detailed Implementation

[0020] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0021] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0022] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention. Example 1

[0023] A control method for unmanned delivery vehicles based on operational status awareness, such as Figure 1 As shown, it includes: S1: Acquiring vehicle data during the driving cycle through the vehicle sensor array, performing time alignment and validity marking on the vehicle data to obtain data quality labels; S2: Construct a set of evidence items based on data quality labels, and generate a set of sub-profile labels based on the set of evidence items; S3: Generate a portrait parameter set based on the sub-portrait tag set through preset priority rules and persistence determination strategies, and generate a running status portrait code based on the structured encoding of the portrait parameter set and the sub-portrait tag set.

[0024] S4: Generate a control constraint package based on the running status profile code and determine the control mode. Based on the control mode, constrain path planning, trajectory generation and execution control, and output vehicle control commands. S5: Execute vehicle control commands and synchronously obtain execution feedback, perform closed-loop verification of the operating status profile code, trigger the verification process when the preset verification conditions are met, and update the operating status profile code and control constraint package based on the verification results.

[0025] The present invention is further configured such that S1 specifically includes: performing time alignment on the vehicle data using either cache alignment or maintain alignment, marking data exceeding a preset timeliness threshold as expired, marking the time-aligned vehicle data for validity using at least one of range verification, jump verification, and continuity verification, and generating data quality labels based on the expired and valid labels. Specifically, in the unmanned delivery vehicle control method based on operational status perception, data acquisition and quality labeling are the first steps, mainly acquiring vehicle data within the driving cycle through an onboard sensor array. This data comes from multiple sensors and covers vehicle position, speed, etc. Information such as acceleration, battery level, communication status, and power system operating status is collected. Since different sensors have different sampling frequencies, this data is first time-aligned to ensure that it can be analyzed and processed under a unified time reference. To this end, two alignment methods are used: buffer alignment and hold alignment. Buffer alignment buffers the data with lower sampling frequency to match it with the high-frequency sampling data and fill in the missing time points. Hold alignment selects the high-frequency data closest to the target time point and aligns it with the low-frequency data without interpolation. These two alignment methods ensure that all sensor data can be compared and analyzed within the same time period. After completing time alignment, the next step is to mark the vehicle data for validity. This is to ensure the validity of the data. The data must be verified through range verification, jump verification, and continuity verification. Range verification ensures that the data falls within a reasonable range. For example, the location data should not exceed the map range, and the speed should not exceed the maximum vehicle speed. Jump verification checks whether there are abrupt changes or abnormal changes in the data. For example, whether the abrupt change in the location or speed data exceeds a reasonable threshold, usually defined as a change of more than 10 km / h. Continuity verification checks whether the data changes smoothly to prevent abnormal fluctuations caused by sensor failure or data transmission errors. It is usually required that the change within a continuous time window does not exceed the preset maximum change. The precise value of the maximum change is within 10%. After these verification steps are completed, the data will be marked as valid or invalid, depending on whether the data meets the above verification criteria. If the data is marked as valid, its freshness, completeness, and consistency will be further assessed. The freshness label is determined based on the difference between the data acquisition time and the current time. Specifically, if the data acquisition time is within 10 seconds, it is marked as normally fresh; if the data acquisition time is between 10 and 30 seconds, it is marked as slightly expired; and if it exceeds 30 seconds, it is marked as expired. The completeness label checks whether the data is missing key fields, such as location or velocity data. If the data is complete, it is marked as complete; otherwise, it is marked as incomplete. Finally, the consistency label determines the reliability of the data by comparing the consistency of data from different sensors with the data within the time window. If there are significant differences or instability between multiple time points, the data is marked as inconsistent for subsequent processing. Then, based on time alignment, expiration marking, validity verification, and consistency judgment, the generated data quality labels will serve as inputs for subsequent steps to ensure that the system uses high-quality vehicle data when making control decisions. These data quality labels include freshness labels, completeness labels, and initial consistency judgment labels. The generation rules for each label are preset. The default threshold for the freshness label is 10 seconds, the completeness label is determined based on the missing fields, and the consistency label is determined based on the comparison of multi-source data. Through these steps, the system can ensure that subsequent decisions are based on fully verified and reliable vehicle data.

[0026] The present invention is further configured such that S2 includes: Based on the data quality labels, a preset subsystem mapping table is used to group the vehicle data into subsystems. Based on a preset evidence rule configuration table, evidence rules are constructed for each group of data, and evidence judgment is performed. The evidence judgment results are discretized and output to generate evidence conclusions. Based on data quality labels and combined with evidence conclusions, stability is determined within a preset sliding time window to generate evidence strength labels. Then, the evidence conclusions and evidence strength labels are combined to form a set of evidence items. Specifically, in the process of evidence construction and stability determination based on data quality labels, firstly, the vehicle data is grouped according to a preset subsystem mapping table. Vehicle data includes, but is not limited to, various data types such as positioning, perception, and power systems. These data are assigned to the corresponding subsystems according to their source or type, such as positioning data being assigned to the positioning subsystem and perception data being assigned to the perception subsystem. The role of the mapping table is to accurately allocate data to the corresponding processing rules according to the characteristics of different data types, thereby ensuring that each type of data can be analyzed and processed according to its characteristics. Next, for each group of data, evidence rules are constructed according to the preset evidence rule configuration table, and evidence judgment is performed according to these rules. The evidence rules are used to evaluate the validity and consistency of each type of data. For example, the evidence rules for location data may include "check whether the location data is consistent with the wheel speed and IMU data". If the data meets the preset standard, it is judged as "valid"; if the data does not meet the standard, it is judged as "invalid"; if the data cannot be judged or is missing, it is judged as "unknown". After each piece of data is judged, an evidence conclusion is generated, reflecting whether the data meets the preset rules. After generating the evidentiary conclusion, the next step is to determine the stability of the data. This stability determination is based on the data quality label and the evidentiary conclusion, and is performed within a preset sliding time window. The sliding time window is a duration, set to 15 seconds by default, used to assess whether the data is stable within this time period. If the evidentiary conclusion of the data does not change significantly within this time period, and the data quality label remains stable, then the evidence is considered stable, and an evidence strength label is generated, marked as "strong" or "stable". If the data quality fluctuates significantly or the evidentiary conclusion changes frequently, an evidence strength label is generated, marked as "weak" or "unstable". This process ensures the stability of the data by comparing and analyzing data from multiple periods, and can generate corresponding strength labels based on the stability. Finally, the evidence conclusions generated by each subsystem and their corresponding evidence strength labels are summarized to form a set of evidence items.

[0027] The present invention is further configured such that S2 further includes: The evidence items in the evidence item set are merged according to subsystems, and the initial level of the corresponding sub-profile label is generated by using a preset evidence merging strategy; The initial level is mapped to a sub-profile level through a preset level mapping table. The sub-profile level is then continuously judged to generate a sub-profile tag set. Specifically, the vehicle data is first grouped according to a preset subsystem mapping table. The vehicle data includes data collected by multiple sensors, such as positioning data, perception data, and power system data. Each data source is assigned to the corresponding subsystem according to its type. All positioning data is classified into the positioning subsystem, all perception data into the perception subsystem, and power system data into the power system subsystem, etc. By using the preset subsystem mapping table, the system ensures that each type of data is assigned to the appropriate subsystem for subsequent processing. After data grouping, evidence rules applicable to each group are constructed according to a pre-defined evidence rule configuration table, and the data is used for evidence judgment. The evidence rules are set according to the characteristics of the data type, with the aim of evaluating the validity and consistency of each data point. For example, for location data, the evidence rules may include judging whether the location data is consistent with wheel speed and IMU data; for perception data, the evidence rules may include judging whether the target identified by the detection sensor is consistent with the expectation. Each evidence judgment result generates an evidence conclusion, which may be "valid" indicating that the data meets the expectation, "invalid" indicating that the data does not meet the expectation, or "unknown" indicating that the data is undeterminable or missing. After the evidence assessment is completed, the process of generating sub-profile labels begins. First, the system merges the evidence items using an evidence merging strategy (which employs a weighted average method). This strategy can either calculate a weighted average based on the credibility of the evidence items or prioritize the evidence item with the highest credibility as the merged result. The purpose of evidence merging is to integrate the assessment results of multiple evidence items in the same subsystem into a single final result. Through this merging process, the system generates an initial level label for each subsystem. The initial level label may be "normal," "slightly restricted," "significantly restricted," or "unavailable," representing different subsystem states. Next, the initial level of each subsystem is converted into a standardized sub-profile level using a preset level mapping table. The preset level mapping table maps the initial level "normal" to sub-profile level 0, indicating that the subsystem is operating well; "slightly restricted" to sub-profile level 1, indicating that the subsystem has slight restrictions; "significantly restricted" to sub-profile level 2, indicating that the subsystem is significantly restricted; and "unavailable" to sub-profile level 3, indicating that the subsystem is in a faulty state. After mapping to sub-profile levels, the system performs a persistence determination based on the evidence conclusions and data quality labels of each subsystem. The purpose of the persistence determination is to assess whether the operating status of the subsystem is stable over multiple time periods. The system compares the evidence conclusions within a preset sliding time window. If the evidence conclusions remain consistent and the data quality labels remain stable within the time window, the data status is considered stable, and a stable sub-profile label is generated. If the evidence conclusions change frequently or the data quality fluctuates greatly, the subsystem status is considered unstable, and an unstable sub-profile label is generated. Finally, after the persistence determination, the sub-profile labels of all subsystems form the final set of sub-profile labels. The present invention is further configured such that S3 includes: The profile parameter set includes the overall profile level, the dominant risk source, and the credibility level; The dominant risk source is determined from the sub-profile tag set based on preset priority rules; The sub-portrait tag set is gating processed according to a preset persistence judgment strategy to generate the overall portrait level; A credibility level is generated based on data quality labels and the consistency of the evidence set. The overall profile level, dominant risk source, credibility level, and sub-profile tag set are structured and encoded to generate an operational status profile code. Specifically, the dominant risk source is determined in the sub-profile tag set through preset priority rules. The dominant risk source refers to the subsystem that is most likely to affect the overall system status among multiple subsystems. Each subsystem is assigned a priority according to its impact on the overall system operation. Subsystems with higher priority are usually considered to be dominant risk sources in the current system status. For example, the power system may be considered to have a higher priority than the positioning system because a failure of the power system may directly cause the unmanned delivery vehicle to malfunction, while a failure of the positioning system may be compensated for through other means. Subsequently, based on the preset persistence judgment strategy, the sub-profile tag set is gating to generate an overall profile level. The persistence judgment strategy judges whether the state of each subsystem is stable by comparing data over multiple periods. If the state of a subsystem does not change much over multiple time periods and the data quality remains consistent, the system considers the subsystem stable, and the gating result maintains the original level. If the state of a subsystem fluctuates frequently, the instability of the subsystem is handled by adjusting the level or temporarily downgrading it. Finally, by combining the gating results of each subsystem, an overall profile level reflecting the overall system state is generated, which usually includes several levels such as "normal", "cautious", "downgraded" and "lowest risk" to indicate the current health status of the system. After generating the overall profile level, a credibility level is generated by combining the consistency judgment of data quality labels and evidence items. Data quality labels reflect the freshness, completeness, and consistency of data. The generation of credibility level depends on the judgment results of these labels and each evidence item. If the evidence items are consistent and the data quality is good, the credibility level is "high credibility". If there are some missing or inconsistent parts, a "medium credibility" or "low credibility" level is generated. The credibility level is used to assess the system's trust in the current data, thereby providing a reference for subsequent control decisions. Finally, the system performs structured encoding of the overall profile level, dominant risk source, credibility level, and sub-profile tag set to generate an operational status profile code.

[0028] The present invention is further configured such that S4 includes: Based on the total profile level and dominant risk source in the profile parameter set, control constraint templates are selected from the preset control constraint template library using at least one of the following methods: table lookup matching, rule matching, and state machine switching. Based on the operational status profile code and using parameter instantiation, a control constraint package is generated from the control constraint template. The control mode is selected according to the overall profile level. A pre-defined constraint association table is used to associate the control constraint package with the control mode. Specifically, firstly, based on the overall profile level and dominant risk source in the generated operational status profile code, a suitable control constraint template is selected from a pre-defined control constraint template library. This library contains multiple predefined control templates, each corresponding to a different system state and risk source, and providing corresponding control strategies. When selecting a suitable control constraint template, the system matches the template most suitable for the current system state based on the combination of the overall profile level and the dominant risk source. For example, when the overall profile level indicates that the system is operating normally and the dominant risk source is the sensing subsystem, the system may choose a more lenient control constraint template; however, if the overall profile level is "degraded" and the dominant risk source is the power system, the system will choose a more stringent control constraint template to limit speed and acceleration, reducing system risk. Once a control constraint template is selected, the system can generate a control constraint package through parameter instantiation. The control constraint package combines the abstract control strategy in the template with the actual system data to specifically generate constraints such as path planning, speed limits, acceleration limits, and trajectory generation. For example, if the control constraint template requires imposing restrictions on path planning, the system will instantiate actionable path planning constraints based on the overall profile level and the specific data involved in the dominant risk source. If the overall profile level is "downgraded", the system will restrict the selection of simpler and safer paths to avoid complex environments or areas with high uncertainty. After generating the control constraint package, the system selects the corresponding control mode based on the overall profile level. The selection of the control mode is based on the overall profile level and usually includes multiple levels, such as "normal mode", "cautious mode", "degraded mode" and "minimum risk mode". Each mode corresponds to a different control strategy. For example, in "normal mode", the vehicle travels at a normal speed and along a normal path, while in "minimum risk mode", the system will forcibly decelerate and adopt the most conservative operating strategy. Finally, the relationship between control constraint packages and control modes is mapped through a pre-defined constraint association table, which defines the control constraint package corresponding to each control mode.

[0029] The present invention is further configured such that S4 further includes: Based on the control constraint package, candidate routes and candidate maneuvers are restricted in the path planning process, and the planning results are generated using a preset constrained path planning method. The planning results are used as reference paths or constraint inputs for trajectory generation. Boundary constraints are applied to the trajectory generation process based on the control constraint package, and the target trajectory is generated using a preset trajectory generation method. The target trajectory is then used as the tracking target for execution control. The system performs control constraint processing on the execution control process based on the control constraint package and outputs vehicle control commands based on the target trajectory. Specifically, the path planning process is first constrained based on the control constraint package, limiting candidate routes and maneuvers. During path planning, the system filters out multiple possible candidate routes and maneuvers based on the vehicle's current state, environmental data, and the requirements in the control constraint package. Candidate routes and maneuvers are limited by constraints such as maximum speed, acceleration, and turning radius set in the control constraint package. When the maximum turning radius is set to 10 meters in the control constraint package, the turning radius of all candidate routes must not exceed this value. If the control constraint package limits the maximum speed to 50 kilometers per hour, the speed of all candidate paths must not exceed 50 kilometers per hour. After these constraint filters, the final generated path planning result is the optimal path that meets the control constraint conditions. Next, based on the path planning results and the boundary constraints in the control constraint package, the trajectory generation process is carried out. The control constraint package applies boundary constraints to the trajectory generation process to ensure that the generated trajectory meets safety requirements. Boundary constraints include restrictions such as maximum turning angle, maximum acceleration, and minimum turning radius. During trajectory generation, common algorithms such as Bézier curve generation are used to ensure that the generated trajectory is smooth and meets the set constraints. If the maximum acceleration is limited to 0.5 m / s², the trajectory generation algorithm will ensure that the acceleration of the entire trajectory will not exceed this value. Through these boundary constraints, the target trajectory generated by the system has high operability and safety. After the trajectory is generated, the system enters the execution control process. Based on the target trajectory and the control constraint package, it performs control quantity constraint processing to generate vehicle control commands. The control quantity constraints in the control constraint package include controls on various aspects of the vehicle, such as steering angle, acceleration, and speed. This ensures that the constraints in the control constraint package can ensure that the vehicle does not exceed the set safety range during execution. If a maximum value for the steering angle is set in the control constraint package, the system will ensure that the vehicle's steering angle does not exceed this limit, thereby preventing the vehicle from making overly sharp turns and ensuring safe driving. Finally, the system calculates the control commands such as steering angle, acceleration, and speed that the vehicle needs to execute, guiding the vehicle to travel according to the generated trajectory. The entire process involves path planning, trajectory generation, and execution control. Each step is strictly constrained by the control constraint package to ensure that the system adopts the most appropriate control strategy under different operating conditions. Specifically, trajectory generation includes boundary constraint application and trajectory optimization algorithm. Boundary constraint application involves the control constraint package defining multiple sets of constraints such as the maximum velocity and maximum acceleration of the trajectory during trajectory generation to further ensure that the trajectory does not exceed a safe predetermined range. The trajectory optimization algorithm specifically uses Bézier curves to generate trajectories, thereby ensuring the smoothness and stability of the trajectories.

[0030] The present invention is further configured such that S5 includes: Execute vehicle control commands and collect execution feedback; The vehicle control commands and execution feedback are time-aligned, and the consistency of the commands and responses is verified based on the alignment results to obtain a consistency result. When the consistency result meets the preset deviation conditions, the evidence item set and the evidence item or sub-profile label associated with the deviation are updated, and the running status profile code and control constraint package are regenerated accordingly. Specifically, after the vehicle control command is executed, the system collects the vehicle execution feedback data and performs time alignment. The vehicle control command includes various control quantities such as steering, acceleration, and speed that should be followed during vehicle driving. The feedback data is the actual performance of the vehicle after executing these commands, including speed, acceleration, and steering response. In order to ensure that the control command and the execution feedback data can correspond in time, the system synchronizes the control command and the execution feedback through time alignment technology for further consistency verification. It is important to note that the consistency verification process compares the differences between the control command and the execution feedback. If there is a deviation between the actual feedback and the expected control command, the system will determine whether it meets the preset deviation conditions. The preset threshold of the deviation conditions can be flexibly adjusted according to the actual application scenario. For example, if the difference between the vehicle's acceleration and the expected value exceeds the preset threshold (the default threshold is 0.2 m / s²), it is considered that a deviation has occurred, and an update process is triggered. When a deviation occurs, the system updates the evidence set, which contains health status information of multiple subsystems, reflecting the operating status of each subsystem. The system updates the evidence items based on the latest data from the execution feedback and may need to adjust the relevant sub-profile labels. For example, if the status of a subsystem changes and the performance of the power system is lower than expected, the system may change the label of the subsystem from "normal" to "slightly restricted" or "significantly restricted". After the evidence set and sub-profile labels are updated, the system will regenerate the running status profile code. Combining the overall profile level, dominant risk source, credibility level and updated sub-profile labels, a new profile code will be generated. In addition, the system will regenerate the control constraint package based on the new running status profile code. The control constraint package will adjust the control strategy according to the current system status, such as resetting the speed limit or adjusting the path planning strategy based on the vehicle acceleration, speed and path planning constraints. Specifically, path planning mainly includes candidate path selection and path optimization. Candidate path selection involves the system filtering out multiple candidate paths based on constraints such as maximum turning radius, maximum speed, and maximum acceleration set in the control constraint package, and prioritizing the shortest path that meets these constraints.

[0031] Path optimization further optimizes the path using pre-defined control constraint packages, ensuring that the path is as short as possible while satisfying the constraints.

[0032] The present invention is further configured such that S5 includes: The credibility level is obtained by parsing the running status profile code; The system determines whether a preset review condition is met based on at least one of the following: credibility level, data quality label, and historical sequence of operational status profile code. If the preset review condition is met, a review process is triggered to obtain review data. Based on the review data, the evidence item set, sub-profile label set, and operational status profile code are reconstructed. The control constraint package is regenerated based on the updated operational status profile code. Specifically, after executing the vehicle control command, the system parses the operational status profile code and extracts the credibility level. The credibility level is determined by the stability and consistency assessment of the evidence item set and data quality label. The credibility level is based on multiple factors, including the completeness, freshness, and consistency of the data. If the credibility level is low, it indicates that the system currently has low trust in the data and there may be unstable factors. In this case, the review process will be triggered. The system further determines whether to trigger a review based on preset review conditions. The review conditions include the following situations: if the confidence level is low, it indicates that the system does not have enough trust in the current data, and the review process will be triggered; if the data quality label shows that the data is missing or inconsistent, the review will also be initiated; if the historical sequence of the running status profile code shows that the system status fluctuates greatly and the current status is inconsistent with the historical status, the system will also trigger a review. When a review is triggered, the system will supplement the missing or inaccurate data points by re-collecting or updating sensor data. After the data collection is completed, the system will update the evidence item set based on the new data and adjust the sub-profile labels according to the new evidence items. If the operating status of a subsystem changes, such as the performance of the power system being lower than expected, the system will update the label of the subsystem, possibly changing its status from "normal" to "slightly restricted" or "unavailable". The updated evidence item set reflects the latest subsystem status, while the sub-profile label set represents the latest assessment of the health status of each subsystem. After the evidence item set and sub-profile tag set are updated, the system will regenerate the operating status profile code. The new operating status profile code integrates the overall profile level, dominant risk source, credibility level and sub-profile tag set, and comprehensively reflects the current operating status of the vehicle. Based on the updated operating status profile code, the system will regenerate the control constraint package and adjust the control strategy to ensure the safety and stability of the vehicle in the current state. The control constraint package will adjust the maximum speed, acceleration limit or select different path planning strategies according to the new profile code to cope with the changes in the current system state. Example 2

[0033] Please see Figure 2 This exemplary unmanned delivery vehicle control system based on operational status perception includes: Acquisition and Labeling Module: Acquires vehicle data during the driving cycle through the vehicle sensor array, performs time alignment and validity labeling on the vehicle data, and obtains data quality labels; Sub-profile generation module: Constructs a set of evidence items based on data quality tags, and generates a set of sub-profile tags based on the set of evidence items; The profile code synthesis module generates a profile parameter set based on the sub-profile tag set through preset priority rules and continuous judgment strategies. It generates the running status profile code based on the profile parameter set and the sub-profile tag set through structured encoding. The profile parameter set includes the total profile level, the dominant risk source, and the credibility level. Image-driven control module: Generates control constraint package and determines control mode based on the image code of running status, constrains path planning, trajectory generation and execution control based on control mode, and outputs vehicle control commands; Closed-loop verification module: Executes vehicle control commands and synchronously obtains execution feedback, performs closed-loop verification on the running status profile code, triggers the verification process when the preset verification conditions are met, and updates the running status profile code and control constraint package based on the verification results; It should be noted that the specific steps for the review include: Data re-acquisition: When the verification conditions are met, the system re-acquisitions the current sensor data, especially key information such as positioning and sensing. Data correction: The system corrects newly collected data and repairs missing or abnormal data; Evidence item update: Based on the review data, update the evidence items of each subsystem. If the evidence item of a certain subsystem changes from "normal" to "restricted" or "unavailable", then update the sub-profile label. Regenerate the running status profile code: Based on the updated evidence items and sub-profile tags, regenerate the complete running status profile code; Control strategy update: Adjust the control constraint package and decision strategy based on the updated profile code.

[0034] It should be noted that the unmanned delivery vehicle control system based on operational status awareness provided in the above embodiments and the unmanned delivery vehicle control method based on operational status awareness provided in the above embodiments belong to the same concept. The specific methods by which each module and unit performs operations have been described in detail in the method embodiments and will not be repeated here. In practical applications, the unmanned delivery vehicle control system based on operational status awareness provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0035] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A control method for unmanned delivery vehicles based on operational status perception, characterized in that, Includes the following steps: S1: Acquire vehicle data during the driving cycle through the vehicle sensor array, perform time alignment and validity marking on the vehicle data, and obtain data quality labels; S2: Construct a set of evidence items based on data quality labels, and generate a set of sub-profile labels based on the set of evidence items; S3: Generate a portrait parameter set based on the sub-portrait tag set through preset priority rules and persistence judgment strategies, and generate a running status portrait code based on the structured encoding of the portrait parameter set and the sub-portrait tag set. S4: Generate a control constraint package based on the running status profile code and determine the control mode. Based on the control mode, constrain path planning, trajectory generation and execution control, and output vehicle control commands. S5: Execute vehicle control commands and synchronously obtain execution feedback, perform closed-loop verification of the operating status profile code, trigger the verification process when the preset verification conditions are met, and update the operating status profile code and control constraint package based on the verification results.

2. The unmanned delivery vehicle control method based on operational status perception according to claim 1, characterized in that, S1 includes: The vehicle data is time-aligned using either cache alignment or maintain alignment, and data exceeding a preset time threshold is marked as expired. The time-aligned vehicle data is then marked for validity using at least one of range verification, jump verification, and continuity verification. Data quality labels are generated based on the expired and valid labels.

3. A control method for unmanned delivery vehicles based on operational status perception according to claim 1, characterized in that, S2 includes: Based on the data quality labels, the vehicle data is grouped into subsystems using a preset subsystem mapping table; Based on a preset evidence rule configuration table, evidence rules are constructed for each group of data, and evidence judgment is performed. The evidence judgment results are discretized and output to generate evidence conclusions. Based on data quality labels and combined with evidence conclusions, stability is determined within a preset sliding time window, and evidence strength labels are generated. Then, evidence conclusions and evidence strength labels are combined to form a set of evidence items.

4. The unmanned delivery vehicle control method based on operational status perception according to claim 3, characterized in that, S2 further includes: The evidence items in the evidence item set are merged according to subsystems, and the initial level of the corresponding sub-profile label is generated by using a preset evidence merging strategy; The initial level is mapped to a sub-portrait level through a preset level mapping table, and a continuous determination is performed on the sub-portrait level to generate a sub-portrait tag set.

5. The unmanned delivery vehicle control method based on operational status perception according to claim 1, characterized in that, S3 includes: The profile parameter set includes the overall profile level, the dominant risk source, and the credibility level; The dominant risk source is determined from the sub-profile tag set based on preset priority rules; Based on a preset persistence judgment strategy, the sub-portrait tag set is gating processed to generate the overall portrait level; A credibility level is generated based on data quality labels and the consistency of the evidence set. The overall profile level, dominant risk source, credibility level, and sub-profile tag set are structured and encoded to generate an operational status profile code.

6. The unmanned delivery vehicle control method based on operational status perception according to claim 1, characterized in that, S4 includes: Based on the total profile level and dominant risk source in the profile parameter set, control constraint templates are selected from the preset control constraint template library using at least one of the following methods: table lookup matching, rule matching, and state machine switching. Based on the running status profile code, a control constraint package is generated from the control constraint template using parameter instantiation. The control mode is selected according to the total profile level, and the control constraint package is associated with the control mode through a preset constraint association table.

7. The unmanned delivery vehicle control method based on operational status perception according to claim 6, characterized in that, S4 further includes: Based on the control constraint package, candidate routes and candidate maneuvers are restricted in the path planning process, and the planning results are generated using a preset constrained path planning method. The planning results are used as reference paths or constraint inputs for trajectory generation. Boundary constraints are applied to the trajectory generation process based on the control constraint package, and the target trajectory is generated using a preset trajectory generation method. The target trajectory is then used as the tracking target for execution control. The control constraint package is used to perform control quantity constraint processing on the execution control process, and vehicle control commands are output based on the target trajectory.

8. The unmanned delivery vehicle control method based on operational status perception according to claim 1, characterized in that, S5 includes: Execute vehicle control commands and collect execution feedback; The vehicle control commands and execution feedback are time-aligned, and the consistency of the commands and responses is verified based on the alignment results to obtain a consistency result. When the consistency result meets the preset deviation conditions, update the evidence item set and the evidence item or sub-profile label associated with the deviation, and regenerate the running status profile code and control constraint package accordingly.

9. A control method for an unmanned delivery vehicle based on operational status perception according to claim 8, characterized in that, The S5 also includes: The credibility level is obtained by parsing the running status profile code; The system determines whether the preset review conditions are met based on at least one of the following: credibility level, data quality label, and historical sequence of operation status profile code. If the preset review conditions are met, the review process is triggered to obtain review data. Based on the review data, the evidence item set, sub-profile label set, and operation status profile code are reconstructed. The control constraint package is then regenerated based on the updated operation status profile code.

10. A control system for an unmanned delivery vehicle based on operational status perception, used to implement the unmanned delivery vehicle control method based on operational status perception as described in any one of claims 1-9, characterized in that, include: Acquisition and Labeling Module: Acquires vehicle data during the driving cycle through the vehicle sensor array, performs time alignment and validity labeling on the vehicle data, and obtains data quality labels; Sub-profile generation module: Constructs a set of evidence items based on data quality tags, and generates a set of sub-profile tags based on the set of evidence items; The profile code synthesis module generates a profile parameter set based on the sub-profile tag set through preset priority rules and continuous judgment strategies. It generates the running status profile code based on the profile parameter set and the sub-profile tag set through structured encoding. The profile parameter set includes the total profile level, the dominant risk source, and the credibility level. Image-driven control module: Generates control constraint package and determines control mode based on the image code of running status, constrains path planning, trajectory generation and execution control based on control mode, and outputs vehicle control commands; Closed-loop verification module: Executes vehicle control commands and synchronously obtains execution feedback, performs closed-loop verification on the operating status profile code, and triggers the verification process when the preset verification conditions are met, and updates the operating status profile code and control constraint package based on the verification results.

Images