Trigger type data acquisition system

Abstract

The invention relates to the technical field of signal acquisition, and provides a trigger type data acquisition system, which comprises a cloud data platform used for configuring a target data acquisition task, the target data acquisition task comprises at least one target event trigger, the target event triggers are in one-to-one correspondence with target data acquisition sub-tasks in the target data acquisition task; the task scheduling subsystem is used for receiving a target data acquisition task issued by the cloud data platform and updating a current data acquisition task of the task execution subsystem based on the target data acquisition task; and the task execution subsystem is used for monitoring the to-be-collected data according to the target data collection task, and collecting the to-be-collected data according to the target data collection subtask matched with the target event trigger to form target data when the target event trigger is triggered by the to-be-collected data. According to the technical scheme provided by one or more embodiments of the invention, the efficiency and convenience of data acquisition can be improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of signal acquisition, and in particular to a triggered data acquisition system. Background Art

[0002] In today's highly information-based and intelligent era, data collection and processing are of vital importance to the development of various fields. Especially in fields such as autonomous driving, which have extremely high requirements for data accuracy and real-time performance, efficient and accurate data collection systems are a key factor in optimizing vehicle system performance.

[0003] With the rapid development of automobile intelligence and networking technology, data acquisition is becoming increasingly important as a key link in automobile system monitoring, fault diagnosis and performance optimization. In current automobile data acquisition systems, a technical solution based on static configuration is generally adopted for the data acquisition process inside the domain controller.

[0004] However, since the trigger rules are usually statically configured, it is difficult to dynamically adjust the trigger conditions for data collection according to changes in the driving area, real-time road conditions, driving behavior, or vehicle system status during actual vehicle operation. In the rapidly changing automotive market, this inflexible configuration method makes it difficult for data collection systems to keep up with the pace of technological development and cannot meet users' needs for efficient, convenient, and real-time data collection systems. Summary of the invention

[0005] In view of this, one or more embodiments of the present disclosure provide a trigger data acquisition system that can meet the diverse needs of data acquisition in complex scenarios and improve the efficiency and convenience of data acquisition.

[0006] On the one hand, the present disclosure provides a triggered data acquisition system, which includes a cloud data platform, a task scheduling subsystem and a task execution subsystem; the cloud data platform is used to configure a target data acquisition task, the target data acquisition task includes at least one target event trigger, and the target event trigger corresponds one-to-one to the target data acquisition subtask in the target data acquisition task; the task scheduling subsystem is used to receive the target data acquisition task issued by the cloud data platform, and update the current data acquisition task of the task execution subsystem based on the target data acquisition task; the task execution subsystem is used to monitor the data to be collected according to the target data acquisition task, and when the target event trigger is triggered by the data to be collected, collect the data to be collected according to the target data acquisition subtask matching the target event trigger to form target data.

[0007] In one embodiment, the cloud data platform includes a device configuration module, a trigger configuration module and a collection task configuration module; the device configuration module is used to manage candidate smart devices with data collection functions; the trigger configuration module is used to manage candidate event triggers, and the attributes of the candidate event triggers include at least one of the trigger name, trigger type, compatible device type, trigger condition and trigger action; the collection task configuration module is used to generate the target data collection task based on the target smart device among the candidate smart devices and the target event trigger among the candidate event triggers.

[0008] In one embodiment, the type of the target event trigger includes at least one of the following: a fault detection trigger, which is used to trigger a first data acquisition task when the data acquisition device cannot operate normally; a rule recognition trigger, which is used to trigger a second data acquisition task when the data to be collected meets predefined data rules; an algorithm abnormality trigger, which is used to trigger a third data acquisition task when the operating algorithm of the data acquisition device is abnormal; a shadow mode trigger, which is used to trigger a fourth data acquisition task when the user's operating data for the data acquisition device does not match the algorithm predicted data.

[0009] In one embodiment, when the user's operation data on the data acquisition device does not match the algorithm predicted data, the fourth data acquisition task is triggered, including: determining whether the data difference between the user's operation data on the data acquisition device and the algorithm predicted data exceeds a preset threshold; if the data difference exceeds the preset threshold, determining that the operation data does not match the algorithm predicted data, and triggering the fourth data acquisition task; wherein the data difference is determined based on the rank between a first data sample of the operation data and a second data sample of the algorithm predicted data.

[0010] In one embodiment, updating the current data acquisition task of the task execution subsystem based on the target data acquisition task includes: determining whether the target data acquisition task is consistent with the current data acquisition task; if they are consistent, instructing the task execution subsystem to continue executing the current data acquisition task; if they are inconsistent, sending the target data acquisition task to the task execution subsystem, and instructing the task execution subsystem to stop the current data acquisition task and run the target data acquisition task.

[0011] In one embodiment, the collecting of the data to be collected to form target data includes: dividing the data to be collected into periodic data packets according to a preset time period; caching the periodic data packets according to preset management rules; and sampling the periodic data packets to form the target data.

[0012] In one embodiment, the task execution subsystem includes a first chip and a second chip, and the collecting of the data to be collected to form target data includes: the first chip confirms that the target event trigger is triggered by the data to be collected; the first chip determines the target data collection subtask that matches the target event trigger, and the target data collection subtask includes at least one of an event label, an event start time, and an event end time; the first chip sends the target data collection subtask to the second chip; according to the target data collection subtask, the first chip collects the first data to be collected to form the first target data; according to the target data collection subtask, the second chip collects the second data to be collected to form the second target data.

[0013] In one embodiment, the task execution subsystem is further used to upload the target data to the cloud data platform so that the cloud data platform processes the target data.

[0014] In one embodiment, the cloud data platform processes the target data, including: classifying and merging the target data corresponding to different target event triggers to form at least one event data packet; storing and managing the event data packet according to a predefined data storage strategy.

[0015] In one embodiment, the cloud data platform processes the target data, including: converting the target data into parsed files of different data formats according to the device source of the target data.

[0016] The technical solution provided by one or more embodiments of the present invention configures the target data collection task through a cloud data platform, and transmits the target data collection task to the task execution subsystem for updating, so that the data collection rules can be configured and updated remotely and in real time. This makes the data collection system more adaptable and scalable, which can effectively reduce the cost of data collection, improve the efficiency of data collection, and meet the diverse needs of data collection.

[0017] The technical solution provided by one or more embodiments of the present disclosure can collect data in a more targeted manner by configuring target event triggers, and can collect data more specifically for some typical scenarios. In this way, full data collection at all times is avoided, and data storage and transmission costs can be saved. At the same time, the workload of later data analysis is greatly reduced, making the collected data more valuable. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The features and advantages of the embodiments of the present disclosure may be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the present disclosure in any way. In the accompanying drawings:

[0019] Figure 1 A schematic diagram of the structure of a trigger data acquisition system in one embodiment of the present disclosure is shown;

[0020] Figure 2 A schematic diagram of the structure of a cloud data platform in one embodiment of the present disclosure is shown;

[0021] Figure 3 A schematic diagram of the workflow of a shadow mode trigger in one embodiment of the present disclosure is shown;

[0022] Figure 4 A schematic diagram of the structure of a task scheduling subsystem in one embodiment of the present disclosure is shown;

[0023] Figure 5 A structural schematic diagram of a triggered data acquisition system in a dual-chip working mode on a device side in one embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present disclosure.

[0025] In the related art, the vehicle data acquisition system generally adopts a technical solution based on static configuration for the data acquisition process inside the domain controller. Specifically, this solution deploys a special trigger rule static configuration module in the domain controller. As one of the core components of the data acquisition system, this module can store and manage all preset data acquisition trigger rules. These rules have been defined in the system design phase and solidified in the software or firmware of the domain controller. With the update and upgrade of the Software Development Kit (SDK) version, the trigger rule static configuration module can also be adjusted and optimized accordingly. However, this adjustment and optimization often depends on the release cycle of the SDK version, and users need to manually update the operation. Whenever there is a new functional requirement or a change in the trigger condition, the user needs to wait for the next version of the SDK to be released before obtaining the corresponding support.

[0026] In view of this, one or more embodiments of the present disclosure provide a trigger data acquisition system that can meet the diverse needs of data acquisition in complex scenarios and improve the efficiency and convenience of data acquisition.

[0027] See also Figure 1 A triggered data acquisition system provided in one embodiment of the present disclosure may include a cloud data platform 100, a task scheduling subsystem 200 and a task execution subsystem 300.

[0028] In this embodiment, the cloud data platform 100 is used to configure a target data collection task, wherein the target data collection task includes at least one target event trigger, and the target event trigger corresponds one-to-one to a target data collection subtask in the target data collection task.

[0029] Specifically, the cloud data platform can configure the target data collection task to obtain the target data according to the data collection method expected by the user. Among them, the target data collection task can include multiple target data collection subtasks, and each target data collection subtask corresponds to a target event trigger. Through the target event trigger, data collection can be triggered, and data can be collected in a targeted manner.

[0030] See also Figure 2 In some implementations, the cloud data platform may include a device configuration module, a trigger configuration module, and a collection task configuration module. The device configuration module is used to manage candidate smart devices with data collection functions. The trigger configuration module is used to manage candidate event triggers, and the attributes of the candidate event triggers may include information such as trigger name, trigger type, compatible device type, trigger condition, and trigger action. The collection task configuration module is used to generate a target data collection task based on the target smart device in the candidate smart devices and the target event trigger in the candidate event triggers.

[0031] In some embodiments, the device configuration module can manage and maintain the information of various types of smart devices in the trigger data acquisition system. Each smart device can have key information such as device number, device name, device brand, device type and device key. Optionally, setting a unique device number can accurately distinguish different smart devices. This information of smart devices is not only the basis for identifying devices, but also the key dependence and support for data acquisition tasks. The data of the device configuration module can be stored in a relational database in the cloud. Based on the device configuration module, device information in the trigger data acquisition system can be dynamically added, updated and deleted. Based on the device configuration module, data acquisition tasks for a variety of smart devices can be configured, such as smart cars, smart robots, smart drones, etc. This flexibility and scalability ensures that the trigger data acquisition system can adapt to the access and management needs of different smart devices.

[0032] In some embodiments, the trigger configuration module allows the user to define various event triggers according to actual data collection needs. Each event trigger may have configuration items such as trigger name, trigger type, applicable device type, trigger condition, and action to be performed when triggered. Optionally, setting a unique trigger name can accurately distinguish different event triggers. Through programming development, the event trigger can output a string containing an event trigger mark according to the data format specified by the smart device end and combined with the user-defined data processing logic. The actual existence form of the event trigger can be a plug-in file compiled from the code, which can be loaded into the process by the main program of the smart device. The flexible configuration of the event trigger enables the triggered data acquisition system to meet complex and changeable data acquisition scenarios, and improves the accuracy and efficiency of data acquisition.

[0033] In an actual application example, the types of target event triggers may include: a fault detection trigger, which is used to trigger a first data acquisition task when the data acquisition device fails to operate normally; a rule recognition trigger, which is used to trigger a second data acquisition task when the data to be collected meets predefined data rules; an algorithm exception trigger, which is used to trigger a third data acquisition task when the operating algorithm of the data acquisition device is abnormal; and a shadow mode trigger, which is used to trigger a fourth data acquisition task when the user's operating data for the data acquisition device does not match the algorithm predicted data.

[0034] Specifically, the data acquisition device (equivalent to the above-mentioned candidate intelligent device with data acquisition function) cannot operate normally, which may be manifested as abnormal functions or performance of the data acquisition device, making it difficult to implement the data acquisition function. The failure of the data acquisition device to operate normally may be caused by a variety of reasons, such as hardware damage, software problems, improper operation, environmental factors, etc. The failure of the data acquisition device to operate normally may affect the working efficiency and quality of the data acquisition device, and requires timely diagnosis and repair. "Fault detection" triggers may include "positioning failure" triggers, "navigation failure" triggers, "control node loss" triggers, "drive failure" triggers, "perception failure" triggers, "map engine failure" triggers, "brake system failure" triggers and other branch fault detection triggers.

[0035] Predefined data rules can refer to data judgment logic predefined by development and test personnel based on typical scene conditions, such as sudden braking of the front vehicle and the front vehicle cutting in, using prior knowledge. Predefined data rules can be implemented through code. Based on rule recognition triggers, trigger-type data collection can focus on specific typical scene conditions to ensure that the collected data is more targeted, thereby greatly saving collection time and equipment resources and improving data collection efficiency. "Rule recognition" triggers can include "sudden acceleration" triggers, "sudden deceleration" triggers, "front vehicle cutting in" triggers, "collision warning" triggers, "unsmooth steering wheel control" triggers, "running out of the lane" triggers, "frustration" triggers, "shaking" triggers, "obstacles cannot be avoided" triggers, "speeding" triggers and other branch rule recognition triggers.

[0036] The abnormal operation algorithm of the data acquisition device can be manifested as the operation status of the autonomous driving algorithm in the autonomous driving system of the smart car not meeting expectations. This may include incorrect calculations, misjudgments, and wrong decisions of the autonomous driving algorithm. These abnormal states of the running algorithm may cause the autonomous driving system to fail to work properly, thereby affecting the safety and reliability of the car. In order to ensure the safety and reliability of the autonomous driving system, it is necessary to monitor and diagnose the state of the running algorithm in real time. The "algorithm abnormality" type triggers may include branch algorithm abnormality triggers such as "different sensor perception results are too different" triggers, "planning trajectory jump" triggers, and "manual takeover trigger" triggers.

[0037] Shadow mode can be used to train and improve the autonomous driving system of smart cars. Shadow mode can effectively ensure the safety of the test in the early stage of open road testing of autonomous driving. Through shadow mode, the autonomous driving system can accumulate a lot of experience in the actual environment without causing risks to passengers and other road users. Shadow mode helps train autonomous driving algorithms to have the level of human drivers, thereby improving the maturity and reliability of autonomous driving technology and laying the foundation for its widespread application in the future.

[0038] When the user's operation data on the data acquisition device does not match the algorithm prediction data, the shadow mode trigger can be activated to trigger the data acquisition task. The process includes: determining whether the data difference between the user's operation data on the data acquisition device and the algorithm prediction data exceeds a preset threshold; if the data difference exceeds the preset threshold, it is determined that the operation data does not match the algorithm prediction data, and the fourth data acquisition task is triggered. The data difference can be determined based on the rank between the first data sample of the operation data and the second data sample of the algorithm prediction data.

[0039] For example, the development of "shadow mode" type triggers can use the "Mann-Whitney U test" algorithm of statistical non-parametric tests. This algorithm is a non-parametric test method used to compare whether there is a significant difference in the distribution of two independent samples. This algorithm does not rely on the specific distribution of the sample data, but judges the difference by comparing the ranks of the two samples. "Shadow mode" type triggers can determine whether to trigger data collection by comparing whether there is a significant difference in the data of the three dimensions of steering wheel angle, accelerator pedal opening, and steering wheel pedal opening between human drivers and autonomous driving planning and control algorithms (referred to as planning and control algorithms).

[0040] See also Figure 3 ,In actual application scenarios, the implementation logic of the shadow mode ,trigger can be as follows.

[0041] The autonomous vehicle is driving on the actual road, and the control algorithm is running in the background. The control instructions generated by the control algorithm will not be sent to the wire-controlled chassis, that is, the human driver still maintains actual control of the vehicle. The triggered data acquisition system will monitor the output data of the control algorithm, the vehicle's sensor data, the driver's operation data, and the surrounding environment information in real time. When it is detected that there is a statistically significant difference between the driver's operation data of the vehicle and the output results of the control algorithm in a certain time window, the shadow mode trigger can be activated to collect the corresponding data and upload it to the cloud data platform.

[0042] In some embodiments, the collection task configuration module can effectively combine the smart device and the event trigger to form a specific data collection task. Each data collection task can be composed of one or more smart devices and one or more event triggers. In each data collection task, task parameters such as data collection frequency, collection volume, storage location, and data processing flow can be defined. In addition, the smart device information and event trigger information bound to each data collection task can be dynamically modified. The modified data collection task can be sent to the data collection device end in real time.

[0043] In this embodiment, the task scheduling subsystem 200 is used to receive the target data collection task issued by the cloud data platform, and update the current data collection task of the task execution subsystem based on the target data collection task.

[0044] In this embodiment, the task scheduling subsystem and the task execution subsystem can be arranged together at the data acquisition device end. As a medium, the task scheduling subsystem can receive the target data acquisition task configured by the cloud data platform, and transmit the target data acquisition task to the task execution subsystem for updating, so that the data acquisition rules at the data acquisition device end can be configured and updated remotely and in real time. In this way, the trigger data acquisition system has stronger adaptability and higher scalability, which can effectively reduce the cost of data acquisition, improve the efficiency of data acquisition, and meet the diverse needs of data acquisition.

[0045] In some embodiments, based on the target data acquisition task, the current data acquisition task of the task execution subsystem is updated, including: determining whether the target data acquisition task is consistent with the current data acquisition task; if they are consistent, instructing the task execution subsystem to continue executing the current data acquisition task; if they are inconsistent, sending the target data acquisition task to the task execution subsystem, and instructing the task execution subsystem to stop the current data acquisition task and run the target data acquisition task.

[0046] See also Figure 4 ,In some practical application examples, the task scheduling subsystem may include a cloud ,communication module and a local communication module.

[0047] The cloud communication module can establish communication with the cloud data platform through the HTTP interface to obtain and synchronize the target data collection tasks configured by the cloud data platform in real time. Once a new target data collection task is received, the cloud communication module can update the data collection tasks currently running in the task execution subsystem to ensure that the data collection tasks on the data collection device side are synchronized and consistent with the data collection tasks on the cloud data platform.

[0048] After receiving the target data collection task configured by the cloud data platform, the local communication module will first perform a repeatability check. If it is determined that the task information received from the cloud data platform is consistent with the task information currently being run by the task execution subsystem, the task execution subsystem will be instructed to continue to execute the current data collection task; if not, a termination signal will be sent to the task execution subsystem, causing the task execution subsystem to stop the currently running data collection task and start a new data collection task based on the new task information.

[0049] This dynamic task management and scheduling mechanism ensures timely response and efficient execution of data collection tasks on the data collection device side, and can effectively cope with complex equipment changes and environmental changes.

[0050] In this embodiment, the task execution subsystem 300 is used to monitor the data to be collected according to the target data collection task, and when the target event trigger is triggered by the data to be collected, collect the data to be collected according to the target data collection subtask matching the target event trigger to form target data.

[0051] In this embodiment, the data to be collected can be a data stream monitored by the data collection device based on its own sensors. For example, a data stream in an automatic driving system of a smart car. When the data characteristics of the data to be collected meet the trigger conditions preset in the target event trigger, the trigger action preset in the target event trigger can be executed to collect data for the data to be collected.

[0052] In a practical application example, the data to be collected can be initially collected by the sensor unit on the vehicle side and then transmitted to the vehicle domain controller via the vehicle's controller area network (CAN) bus. After the vehicle domain controller receives the data to be collected transmitted by the CAN bus, it can monitor the data features of the data to be collected and determine whether the data features of the data to be collected trigger a certain target event trigger. When the data features of the data to be collected trigger a certain target event trigger, the trigger action in the target event trigger can be executed, so that the data to be collected is collected according to the data collection rules in the target event trigger to form target data. Optionally, the vehicle domain controller can store, calculate, and send target data.

[0053] In some implementations, collecting the data to be collected to form target data includes: dividing the data to be collected into periodic data packets according to a preset time period; caching the periodic data packets according to preset management rules; and sampling the periodic data packets to form the target data.

[0054] Specifically, the data to be collected may be a data stream, and data segmentation may be performed according to a certain time period (eg, predetermined milliseconds, predetermined seconds, predetermined minutes, etc.). The periodic data packets obtained after segmentation may be cached, and managed and cleaned according to a preset policy.

[0055] In some embodiments, the task execution subsystem may include a first chip and a second chip. The task execution subsystem collects the data to be collected to form the target data, including: the first chip confirms that the target event trigger is triggered by the data to be collected; the first chip determines the target data collection subtask that matches the target event trigger, and the target data collection subtask includes at least one of the event label, the event start time, and the event end time; the first chip sends the target data collection subtask to the second chip; according to the target data collection subtask, the first chip collects the first data to be collected to form the first target data; according to the target data collection subtask, the second chip collects the second data to be collected to form the second target data.

[0056] In some practical application scenarios, the architecture of vehicle domain controllers is undergoing a transition from single-chip to multi-chip, especially dual-chip design. This transition can improve the processing power, redundancy and integration of domain controllers. Through the collaborative work of two high-performance chips, dual-chip domain controllers can allocate resources more efficiently, handle more complex tasks, and improve the overall stability and reliability of triggered data acquisition systems.

[0057] However, in the related art, the domain controller under the dual-chip architecture usually uses a single chip as the main export and processing center of the data, and the communication efficiency and stability between the two chips (or multiple chips) may be seriously affected. When a single chip is responsible for processing and forwarding data from other chips, it may become a bottleneck in the entire data flow. With the increase in the amount of data collected and the improvement of real-time requirements, this single-export design mode can easily lead to data transmission delays, congestion or even loss, which in turn affects the integrity and accuracy of data collection. In addition, the communication quality between the two chips will also be affected by many factors, such as the choice of communication protocol, the format of the data packet, the error handling mechanism, etc., which may reduce the reliability and stability of communication to varying degrees.

[0058] Therefore, in some embodiments provided by the present disclosure, the task execution subsystem can adopt a dual-chip architecture, and the first chip and the second chip can both receive data, both can perform data collection, and both can upload the collected data to the cloud data platform. The first chip and the second chip can process data of different data types, thereby improving the overall efficiency of the triggered data collection system. For example, the first chip can receive and process camera data, and the second chip can receive and process lidar data. The first chip can cache camera image data, and the second chip can cache point cloud data.

[0059] It should be noted that the target event trigger can only run on the first chip. The first chip can determine the first data feature based on the data it receives. The second chip can pre-process the data it receives, determine the second data feature, and send the second data feature to the first chip. Combining the first data feature and the second data feature, the first chip can determine whether the target event trigger is triggered. If the first chip determines that the target event trigger is triggered, it will execute the corresponding target data acquisition subtask. At the same time, using the chip communication module, the first chip can send the target data acquisition subtask to the second chip for execution. Since the target data acquisition subtask sent by the first chip to the second chip includes the event label, event start time, and event end time, the time alignment of the data collected by the two chips can be guaranteed. In other words, the timestamps in the data packets collected by the two chips correspond, which is conducive to subsequent merging processing.

[0060] See also Figure 5 ,In a practical application example, the data acquisition task execution system corresponding to ,the task execution subsystem is the core module that actually executes the data acquisition ,task on the device side, which includes multiple sub-modules such as the data ,caching module, trigger plug-in loading module, data monitoring module, event processing module and chip ,communication module.

[0061] The data cache module can adopt a second-level segmentation strategy to segment the data generated every second into a data packet, and manage and clean up the data packet according to the preset strategy. This data segmentation and management strategy not only ensures the real-time and integrity of the data, but also effectively reduces the resource consumption of data processing and storage.

[0062] The trigger plug-in loading module can compare the trigger information configured in the cloud with the existing downloaded trigger plug-ins on the vehicle side, and dynamically download and load the trigger plug-in script program incrementally through the network communication protocol. After loading, these plug-in programs can monitor the data flow generated by the device side in real time, and trigger data collection and processing actions when the preset conditions are met.

[0063] The data monitoring module can receive the data stream of the autonomous driving system from the device side, and judge and process the event trigger time in combination with the preset trigger rules. All trigger events can be added to the event processing queue for subsequent processing and analysis.

[0064] The event processing module can process each event in the event processing queue in turn. Each event can be a time window. The event processing module can copy the second-level data packets in the time window from the data cache directory to the file directory corresponding to the event, and send the event information to the chip communication module.

[0065] The chip communication module can complete the data exchange and communication within the device and between the devices. For example, the chip communication module of chip 1 can send the trigger event and the time window of each trigger event (including the start time and the end time) to the chip communication module of chip 2, so that the data collected by chip 1 and chip 2 have aligned timestamps.

[0066] In this application example, the dual-chip collaboration on the device side is the key to the efficient operation of the entire triggered data acquisition system. The division of labor and cooperation between the two chips in data processing and communication not only improves the efficiency of data processing on the device side, but also enhances the reliability and stability of the triggered data acquisition system. After the data communication and data copy operations are completed, the event processing modules of the first chip and the second chip compress the data under each event directory. This data compression can not only effectively reduce the bandwidth consumption when data is transmitted on the network, but also improve the efficiency and speed of data transmission. The compressed data packets are uploaded to the cloud storage through the network protocol. The cloud storage system will separate and store the data according to the different events for subsequent data analysis and use.

[0067] In some implementations, the task execution subsystem is further used to upload the target data to the cloud data platform so that the cloud data platform processes the target data.

[0068] Specifically, the cloud data platform processes and stores target data, which can serve as the final link of the entire triggered data acquisition system. The cloud data platform can receive, integrate and store a large amount of target data from the device side, and provide data analysis and decision support.

[0069] In some implementations, the cloud data platform may classify and merge target data corresponding to different target event triggers to form at least one event data packet. The cloud data platform may store and manage the event data packets according to a predefined data storage strategy.

[0070] In an actual application example, the cloud data platform may include a data packet merging module and a data storage management module.

[0071] The data packet merging module is the core part of the cloud data processing flow. When all the second-level data packets related to a certain event are uploaded, the data packet merging module will call the data merging function to merge all the second-level data packets of the event into a complete event data packet. The merged event data packet can be used in subsequent data analysis, data mining, real-time decision support and other application scenarios.

[0072] The data storage management module can store and manage the merged event data packets according to the pre-defined data storage strategy. The cloud data storage management module can ensure the security, integrity and availability of data through efficient storage engines and data management strategies. Event data packets of different events can be stored separately, and each event data packet can have a unique identifier and metadata information so that users can quickly retrieve and access the required data.

[0073] In some implementations, the cloud data platform may convert the target data into parsed files of different data formats according to the device source of the target data.

[0074] In a practical application example, after the target data collected by multiple smart devices are uploaded to the cloud data platform, they will be uniformly analyzed by the data analysis module of the cloud data platform. The data analysis module will automatically execute the corresponding specific data analysis program according to different types of devices. These data can cover multiple aspects, including positioning data, image data, point cloud data, perception fusion target list data, regulatory control algorithm prediction data, user operation data and wire-controlled chassis data. Among them, positioning data, regulatory control algorithm prediction data, user operation data and wire-controlled chassis data can be converted into csv format files after analysis; image data can be converted into jpg format image files after analysis; point cloud data can generate pcd format point cloud files and jpg format image files after analysis.

[0075] This processing method makes the data more universal and usable in different devices and application scenarios. Through unified parsing and format conversion, the target data collected by the triggered data acquisition system can be better utilized and analyzed, thus providing more accurate and valuable information support for various intelligent applications.

[0076] In a practical application example, based on the files generated by the data analysis module, image data, point cloud data, positioning data, and wire-controlled chassis data can be used for visualization on the cloud data platform. The cloud data platform can provide users with a visual interface. Through this interface, users can intuitively see the vehicle driving trajectory of the intelligent device under the triggering scenario, as well as key driving parameters including vehicle speed, yaw angular velocity, steering wheel angle, brake and accelerator pedal opening, etc. This display method enables users to more conveniently understand the actual working conditions corresponding to the triggering scenario, so as to better analyze and make decisions.

[0077] In addition, based on the positioning data and the perception fusion target list data, the scene restoration module in the cloud data platform can be used to generate simulation scene files that meet industrial standards. These simulation scene files can be directly imported into mainstream simulation software for algorithm access and simulation testing. This provides users with a convenient way to test and optimize various algorithms and strategies in a virtual environment, thereby improving the performance and reliability of the autonomous driving system of intelligent vehicles. By analyzing the simulation results, users can further improve and perfect the algorithm to better cope with various situations in actual applications.

[0078] The technical solution provided by one or more embodiments of the present invention configures the target data collection task through a cloud data platform, and transmits the target data collection task to the task execution subsystem for updating, so that the data collection rules can be configured and updated remotely and in real time. This makes the trigger data collection system more adaptable and scalable, which can effectively reduce the cost of data collection, improve the efficiency of data collection, and meet the diverse needs of data collection.

[0079] The technical solution provided by one or more embodiments of the present disclosure can collect data in a more targeted manner by configuring target event triggers, and can collect data more specifically for some typical scenarios. In this way, full data collection at all times is avoided, and data storage and transmission costs can be saved. At the same time, the workload of later data analysis is greatly reduced, making the collected data more valuable.

[0080] For the convenience of description, when describing the above trigger data acquisition system, the functions are divided into multiple subsystems or even multiple functional modules of the subsystems and described separately. Of course, when implementing this application, the functions of each module or each subsystem can be integrated into the same or multiple software and / or hardware for implementation.

[0081] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0082] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

[0083] Although the embodiments of the present disclosure have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A triggered data acquisition system, characterized in that: The system includes a cloud data platform, a task scheduling subsystem and a task execution subsystem; The cloud data platform is used to configure a target data collection task, wherein the target data collection task includes at least one target event trigger, and the target event trigger corresponds one-to-one to a target data collection subtask in the target data collection task; The task scheduling subsystem is used to receive the target data collection task issued by the cloud data platform, and update the current data collection task of the task execution subsystem based on the target data collection task; The task execution subsystem is used to monitor the data to be collected according to the target data collection task, and when the target event trigger is triggered by the data to be collected, collect the data to be collected according to the target data collection subtask matching the target event trigger to form target data.

2. The system according to claim 1, characterized in that The cloud data platform includes a device configuration module, a trigger configuration module and a collection task configuration module; The device configuration module is used to manage candidate smart devices with data collection functions; The trigger configuration module is used to manage candidate event triggers, and the attributes of the candidate event triggers include at least one of a trigger name, a trigger type, an adaptation device type, a trigger condition, and a trigger action; The acquisition task configuration module is used to generate the target data acquisition task according to the target smart device among the candidate smart devices and the target event trigger among the candidate event triggers.

3. The system according to claim 1 or 2, characterized in that: The type of the target event trigger includes at least one of the following: A fault detection trigger, used to trigger the first data collection task when the data collection device fails to operate normally; A rule identification trigger, used to trigger a second data collection task when the data to be collected meets a predefined data rule; An algorithm abnormality trigger, used to trigger a third data acquisition task when an operation algorithm of the data acquisition device is abnormal; A shadow mode trigger is used to trigger the fourth data collection task when the user's operation data on the data collection device does not match the algorithm prediction data.

4. The system according to claim 3, characterized in that When the user's operation data on the data collection device does not match the algorithm prediction data, triggering the fourth data collection task includes: Determine whether the data difference between the operation data of the user on the data acquisition device and the data predicted by the algorithm exceeds a preset threshold; If the data difference exceeds the preset threshold, it is determined that the operation data does not match the algorithm prediction data, and the fourth data collection task is triggered; The data difference is determined based on the rank between a first data sample of the operation data and a second data sample of the algorithm prediction data.

5. The system according to claim 1, characterized in that The updating of the current data collection task of the task execution subsystem based on the target data collection task includes: Determining whether the target data collection task is consistent with the current data collection task; If they are consistent, instruct the task execution subsystem to continue executing the current data collection task; If they are inconsistent, the target data collection task is sent to the task execution subsystem, and the task execution subsystem is instructed to stop the current data collection task and run the target data collection task.

6. The system according to claim 1, characterized in that The collecting the data to be collected to form target data includes: According to a preset time period, the data to be collected is divided into periodic data packets; caching the periodic data packets according to preset management rules; The periodic data packets are sampled to form the target data.

7. The system according to claim 1, characterized in that The task execution subsystem includes a first chip and a second chip, and the collecting the data to be collected to form target data includes: The first chip confirms that the target event trigger is triggered by the data to be collected; The first chip determines the target data collection subtask that matches the target event trigger, wherein the target data collection subtask includes at least one of an event label, an event start time, and an event end time; The first chip sends the target data acquisition subtask to the second chip; According to the target data acquisition subtask, the first chip acquires first data to be acquired to form first target data; According to the target data acquisition subtask, the second chip acquires second data to be acquired to form second target data.

8. The system according to claim 1, characterized in that The task execution subsystem is also used to upload the target data to the cloud data platform so that the cloud data platform processes the target data.

9. The system according to claim 8, characterized in that The cloud data platform performs data processing on the target data, including: Classify and merge target data corresponding to different target event triggers to form at least one event data packet; The event data packets are stored and managed according to a predefined data storage strategy.

10. The system according to claim 8 or 9, characterized in that The cloud data platform performs data processing on the target data, including: According to the device source of the target data, the target data is converted into parsing files of different data formats.

Images