CN117194414A

CN117194414A - Automatic driving data processing system and automatic driving data processing method

Info

Publication number: CN117194414A
Application number: CN202310967301.0A
Authority: CN
Inventors: 陈贺; 单佳炜; 沈罗丰
Original assignee: Jiangsu Youtan Intelligent Technology Co ltd
Current assignee: Jiangsu Youtan Intelligent Technology Co ltd
Priority date: 2023-08-02
Filing date: 2023-08-02
Publication date: 2023-12-08
Anticipated expiration: 2043-08-02

Abstract

The present application relates to an automatic driving data processing system and an automatic driving data processing method. The method comprises the following steps: the data interface acquires a data packet acquired by executing a data acquisition task; the verification module is used for generating verification data according to the validity of the data packet verification data acquisition task and dividing the data packet into a valid data packet and an invalid data packet; the marking module visualizes the effective data packet, displays a label panel for editing labels of the effective data packet, and generates label data of the effective data packet based on the operation of a user on the label panel, wherein the label data comprises object labels; the labeling module is used for dividing and labeling the laser radar point cloud data in the effective data packet with the object label, generating labeling data, and performing quality inspection on the labeling data to generate a quality inspection result; the database stores data packets, verification data, label data, labeling data and quality inspection results. According to the application, the processing efficiency of the automatic driving data is improved.

Description

Automatic driving data processing system and automatic driving data processing method

Technical Field

The application relates to the technical field of automatic driving, in particular to an automatic driving data processing system and an automatic driving data processing method.

Background

In the conventional data production process, collected data is stored in a database, and then operations such as data cleaning and the like are performed. However, since the amount of data of the type of cloud data, video data, etc. in the automatic driving data is generally very large, the data processing efficiency is low because the original data is directly converted and stored in the database and then the data is cleaned. At present, there is no efficient processing flow for the collected data processing of automatic driving data.

Disclosure of Invention

The automatic driving data processing system and the automatic driving data processing method at least solve the problem of low automatic driving data processing efficiency in the related technology.

An autopilot data processing system comprising:

the data interface is used for acquiring a data packet acquired by executing a data acquisition task, wherein the data packet comprises data acquired by various sensors within a preset duration;

the verification module is used for verifying the validity of the data acquisition task according to the data packet, generating verification data and dividing the data packet into an effective data packet and an invalid data packet;

the marking module is used for visualizing the effective data packet, displaying a label panel for editing labels of the effective data packet, and generating label data of the effective data packet based on the operation of a user on the label panel, wherein the label data comprises object labels;

the labeling module is used for dividing and labeling the laser radar point cloud data in the effective data packet with the object label, generating labeling data, and performing quality inspection on the labeling data to generate a quality inspection result;

and the database is used for storing the data packet, the check data, the tag data, the labeling data and the quality inspection result.

In some of these embodiments, the data packet includes: the system comprises laser radar point cloud data and laser radar calibration parameters collected by a laser radar, video data and camera calibration parameters collected by a camera, positioning data collected by a positioning system, inertial navigation data collected by an inertial navigation system and vehicle data, wherein the vehicle data at least comprises: wheel speed data and steering wheel angle data.

In some embodiments, the data packet includes multiple message sets corresponding to the multiple sensors one by one, where each message set includes multiple message sets arranged according to an acquisition sequence, and the message sets carry a timestamp; the verification module verifies the validity of the data acquisition task by verifying the total duration of the data packet and the frame rate stability of the message packet in the data packet; the total duration is determined based on the time stamps of the first message and the last message of the data Bao Nadi, and the frame rate of the message group is determined based on the time stamps of the first message and the last message in the statistical interval and the number of message in the statistical interval.

In some of these embodiments, the frame rate stability of the message packet includes a total frame rate stability of the data packet and a segment frame rate stability of each segment after averaging segments of the data packet.

In some of these embodiments, the label panel includes a label control for editing at least the following labels: road type, special targets, driving behavior, special scenes, buildings, abnormal conditions, time periods, weather, and scenes.

In some embodiments, the data packet is a data packet in a ROSbag format, and the marking module comprises a ROS node and creates a visual graphical interface based on PyQt, tkilter, or Django; the ROS node is used for subscribing to the data of the data packet and mapping the data of the data packet to the visual graphical interface.

In some embodiments, the database comprises a web application framework with a visual graphical interface, a conversion unit and a database management system, wherein the conversion unit is used for combining query parameters input by the visual graphical interface into query sentences of the database management system and providing query results returned by the database management system to the web application framework so as to visualize the query results through the visual graphical interface.

In some of these embodiments, the web application framework is created based on Django, the conversion unit is written based on Python, and the database management system comprises a mysql database management system.

An autopilot data processing method applied to the system of any one of claims 1 to 8, in some embodiments comprising:

acquiring a data packet acquired by executing a data acquisition task, wherein the data packet comprises data acquired by various sensors within a preset duration;

checking the validity of the data acquisition task according to the data packet, generating check data, and dividing the data packet into a valid data packet and an invalid data packet;

visualizing the valid data packet, displaying a label panel for editing labels of the valid data packet, and generating label data of the valid data packet based on the operation of a user on the label panel, wherein the label data comprises object labels;

dividing and marking laser radar point cloud data in an effective data packet with an object tag, generating marking data, and performing quality inspection on the marking data to generate a quality inspection result;

and storing the data packet, the verification data, the tag data, the labeling data and the quality inspection result.

In some embodiments, the data packet includes multiple message sets corresponding to the multiple sensors one by one, where each message set includes multiple message sets arranged according to an acquisition sequence, and the message sets carry a timestamp; checking the validity of the data acquisition task according to the data packet comprises the following steps:

verifying the validity of the data acquisition task by verifying the total duration of the data packet and the frame rate stability of the message group in the data packet; the total duration is determined based on the time stamps of the first message and the last message of the data Bao Nadi, and the frame rate of the message group is determined based on the time stamps of the first message and the last message in the statistical interval and the number of message in the statistical interval.

In some embodiments, the data packet is a data packet in a ROSbag format, and visualizing the valid data packet includes:

creating a visual graphical interface based on PyQt, tkiner or Django;

subscribing the data of the data package through the ROS node, and mapping the data of the data package to the visual graphical interface.

In some of these embodiments, the database includes a web application framework with a visual graphical interface, a conversion unit, and a database management system, the method further comprising:

and combining the query parameters input by the visual graphical interface into a query statement of the database management system through the conversion unit, and providing a query result returned by the database management system to the web application framework so as to visualize the query result through the visual graphical interface.

According to the automatic driving data processing system and the automatic driving data processing method provided by the embodiment of the application, a data interface acquires a data packet acquired by executing a data acquisition task, and the data packet comprises data acquired by various sensors within a preset duration; the verification module is used for generating verification data according to the validity of the data packet verification data acquisition task and dividing the data packet into a valid data packet and an invalid data packet; the marking module visualizes the effective data packet, displays a label panel for editing labels of the effective data packet, and generates label data of the effective data packet based on the operation of a user on the label panel, wherein the label data comprises object labels; the labeling module is used for dividing and labeling the laser radar point cloud data in the effective data packet with the object label, generating labeling data, and performing quality inspection on the labeling data to generate a quality inspection result; the database stores data packets, verification data, label data, labeling data and quality inspection results, at least solves the problem of low automatic driving data processing efficiency in the related technology, and improves the automatic driving data processing efficiency.

The details of one or more embodiments of the application are set forth in the accompanying drawings and the description below to provide a more thorough understanding of the other features, objects, and advantages of the application.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It is evident that the drawings in the following description are only some embodiments of the application, from which other embodiments can be obtained for a person skilled in the art without inventive effort.

Fig. 1 is a schematic diagram of the structure of an automated driving data processing system of the present embodiment.

Fig. 2 is a schematic diagram of a graphical user interface provided by the marking module of the present embodiment.

Fig. 3 is a flowchart of the automatic driving data processing method of the present embodiment.

Detailed Description

Embodiments of the present embodiment will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present embodiments are illustrated in the accompanying drawings, it is to be understood that the present embodiments may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided to provide a more thorough and complete understanding of the present embodiments. It should be understood that the drawings and the embodiments of the present embodiments are presented for purposes of illustration only and are not intended to limit the scope of the embodiments.

The embodiment aims at providing a systematically and mechanically applicable automatic driving data production flow and system, combining a visualized data marking and labeling module and a visualized database system to realize low-cost and high-efficiency automatic driving data processing.

The automatic driving data of the present embodiment includes, but is not limited to, the following data: laser radar point cloud data and laser radar calibration parameters acquired by a laser radar, video data and camera calibration parameters acquired by a camera, positioning data acquired by a positioning system, inertial navigation data acquired by an inertial navigation system and vehicle data, wherein the vehicle data comprises: wheel speed data and steering wheel angle data.

The laser radar point cloud data is a data set of points obtained by data acquisition through a three-dimensional laser radar, and comprises three-dimensional coordinates, colors, classification values, intensity values and time information. In actual business, the data volume of the point cloud data acquired by each data acquisition task is large, the cost is high, the time is long, if a systematic data production flow is not available, the problems of low data efficiency, poor data quality and low data production efficiency are easily caused, the time and the cost are consumed, and the subsequent work is also more difficult.

The automatic driving data of the embodiment is collected by the collecting vehicle, and the collecting vehicle is provided with sensors such as a radar, a camera, a satellite positioning system, an inertial navigation system and the like, and devices such as a PTP time service module, a vehicle-mounted computing unit and the like, so that synchronous collection of the automatic driving data is realized. When automatic driving data are collected, one collection package is generated every time when preset time (for example, 5 minutes) is collected, each collection package comprises data collected by various sensors, and meanwhile, in order to facilitate subsequent processing, calibration parameters of a laser radar and a camera can be added into each collection package. The sensors are synchronized through a PTP time service module, namely, the laser radar point cloud time stamp, the inertial navigation data time stamp, the positioning data time stamp, the bicycle information time stamp and the like are unified, and for example, the UNIX time stamp can be adopted uniformly.

The automatic driving data acquisition scene covers places such as expressways, urban loops, crossroads, overpasses, tunnels, bridges, culverts and the like, covers different periods such as early morning, daytime, evening, nighttime and the like, and comprises various common weather conditions such as sunny days, cloudy days, rain and fog and the like.

The data packet format in this embodiment adopts a topic data format that can be subscribed to by the ros node, i.e. a rosbag format. The data packet is named by time, and fields of acquisition scenes and weather can be added in the naming of the data packet for subsequent processing.

When automatic driving data acquisition is carried out, in order to ensure the effectiveness of data acquisition, each sensor is confirmed to be started and work normally before each data acquisition task is executed, and each acquisition data stream (namely each topic of rosbag, each topic corresponds to one sensor) has continuous, stable and correct data; when automatic driving data is started to be collected, the first rosbag is recorded and is firstly stationary in place for a plurality of minutes, and after equipment such as an inertial navigation system and the like is initialized, the vehicle is started to be collected.

In order to avoid invalid data acquisition as much as possible, each sensor data can be checked periodically during the acquisition process to see if the data is time stamped continuously, frame rate stable, complete and correct.

Stopping the data acquisition task when the acquisition personnel rest and the acquisition vehicle is stationary for a long time; after recording the rosbag interrupt, when the next data acquisition task needs to be restarted, the steps should be executed again to avoid invalidation of the acquired data.

The acquired rosbag data packet is not directly stored in the database, but is stored in the database after the data is cleaned, screened and marked by the automatic driving data processing system after being acquired.

The embodiment provides an automatic driving data processing system. Fig. 1 is a schematic structural diagram of an automatic driving data processing system of the present embodiment, as shown in fig. 1, including:

the data interface 11 is used for acquiring a data packet acquired by executing a data acquisition task, wherein the data packet comprises data acquired by various sensors within a preset duration;

the verification module 12 is connected with the data interface 11 and is used for generating verification data according to the validity of the data acquisition task of the data packet verification data and dividing the data packet into a valid data packet and an invalid data packet;

the marking module 13 is connected with the checking module 12 and is used for visualizing the effective data packet, displaying a label panel for editing labels of the effective data packet, and generating label data of the effective data packet based on the operation of a user on the label panel, wherein the label data comprises an object label;

the marking module 14 is connected with the marking module 13 and is used for dividing and marking the laser radar point cloud data in the effective data packet with the object tag to generate marking data and performing quality inspection on the marking data to generate a quality inspection result;

the database 15 is connected with the labeling module 14 and is used for storing data packets, verification data, label data, labeling data and quality inspection results.

Through the automatic driving data processing system, after the data packet is subjected to validity verification, marking and labeling, the labor waste caused by invalid data packet processing is avoided; the marking efficiency is improved by adopting a visual mode and a label panel; the automatic driving data is stored in the database after validity verification, marking and labeling, so that frequent conversion of automatic driving data formats is avoided, and the automatic driving data processing efficiency is improved.

In some embodiments, the data packet includes multiple message sets corresponding to the multiple sensors one to one, where each message set includes multiple message sets arranged in an acquisition order, and the message sets carry a timestamp. For example, the data packets are data packets in a rosbag format, each data packet includes a plurality of topics, each topic is a message packet, and each topic corresponds to a sensor one by one. Each topic comprises a plurality of message messages arranged according to the acquisition sequence, and each message carries a uniform acquisition time stamp. The verification module verifies the validity of the data acquisition task by verifying the total duration of the data packet and the frame rate stability of the message group in the data packet; the total duration is determined based on the time stamps of the first and last message in the data packet, and the frame rate of the message group is determined based on the time stamps of the first and last message in the statistical interval and the number of message in the statistical interval.

For example, taking the acquisition duration of each data packet as 300 seconds as an example, the verification module calculates the total duration according to the time stamps of the first message packet and the last message packet in the data packet, checks whether the total duration is 300 seconds, and sets a certain tolerance, for example, the total duration can be regarded as normal when the total duration is up and down floating for 1 second, otherwise, the total duration of the verification data packet is regarded as abnormal.

The checking module also judges whether the rosbag data packet is abnormal or not by checking whether the message types corresponding to the topics in the rosbag data packet are consistent with the types of the sensors.

For example, taking the acquisition duration of each data packet as 300 seconds as an example, the verification module takes the whole message packet or a time interval in the message packet as a statistics interval, then determines the duration of the statistics interval according to the time stamps of the first message packet and the last message packet in the statistics interval, counts the number of the message packets in the statistics interval, calculates the frame rate of the message packets in the statistics interval according to the duration of the statistics interval and the number of the message packets in the period, and further compares the frame rate with the acquisition frame rate of the sensor corresponding to the message packet to judge the frame rate stability.

In some of these embodiments, the frame rate stability of the message packet includes a total frame rate stability of the data packet and a segment frame rate stability of each segment after an average segmentation of the data packet. That is, the statistical interval may be a message group collected in the whole preset duration, or may be a message group segment collected in a certain average time segment of the preset duration.

When calculating the frame rate of the message packet, for the whole data packet, dividing the total message number of a certain topic of the data packet by the total duration of the data packet to obtain the total frame rate, if the total frame rate of the topic is the same as the acquisition frequency of a sensor corresponding to the topic, indicating that the data packet is normal, otherwise, considering that the data packet is abnormal. For the data packet with normal total frame rate corresponding to each topic, the data packet can be further segmented, for example, the data packet with 300 seconds total duration is segmented for 30 seconds on average, 10 segments of data packets are obtained, and then whether the segmented frame rate corresponding to each topic in each segmented data packet is the same as the acquisition frequency of the sensor corresponding to the topic is further checked. Thereby checking whether the data packet has abnormal conditions of abrupt frequency change of acquisition in a short time.

In this embodiment, a certain tolerance may be set for packet segmentation anomalies, for example, if the number of packet segmentation anomalies is less than 1/5, the packet may be considered normal, otherwise, the packet is classified as anomalous.

After verification, the data packet is divided into a valid data packet and an invalid data packet, the invalid data packet is rejected or stored in a specific folder, and only the valid data packet is sent to the next module of the data processing system to be processed.

In addition, for all data packets acquired in the whole data acquisition task, the verification module of the embodiment further calculates the accuracy of the data acquisition task and records the accuracy in verification data for subsequent processing and reference.

The accuracy of the data acquisition task comprises acquisition time effective rate and data packet effective rate. The effective acquisition time efficiency refers to the proportion of the effective acquisition time to the duration of the total acquisition time; the effective acquisition time refers to the time when all the acquired topic frames are stable in rate, free of frame loss and correct in data. The data packet effective rate refers to the proportion of the effective data packet to the total data packets; the effective data packet is a data packet with stable frame rate, no frame loss and correct data of all acquired topic frames in the data packet.

The check data generated by the check module adopts json format, and the content of the check data is shown in table 1, so that the database processing is convenient.

Table 1 check field contents of data

The marking module provides a graphical user interface for visualization and playback control of the original data packet data. For example, the video data is fast-forwarded or double-fast played by moving, rotating and zooming the point cloud and the video data with the left, right, and middle keys of the mouse. Meanwhile, the marking module also provides a label panel for interacting with a user, and when the point cloud data and/or the video data are played on the graphical user interface, the user can mark the original data in the data packet by operating a label control on the label panel. A schematic diagram of the graphical user interface provided by the marking module is shown in fig. 2, and the image interface of the video data and the point cloud data is not shown in fig. 2.

Referring to fig. 2, in some embodiments thereof, the label panel includes a label control for editing at least the following labels: road type, special targets, driving behavior, special scenes, buildings, abnormal conditions, time periods, weather, and scenes.

The road types include: crossing, culvert, overpass, tunnel, return lane, bridge and roundabout.

Specific objects include: multiple objects, large vehicles, very near large vehicles, engineering vehicles, special vehicles, very near small objects, animals, special obstacles, special people and vehicles.

The driving behavior includes: uphill, downhill, turning, waiting for traffic lights, decelerating, stopping, and sudden braking.

The special scene includes: congestion to slow travel, congestion to standstill, traffic accident, airport, port, tidal lane, parking lot, railway gate.

The building comprises: gas stations, toll booths, service areas, energy booths, checkpoints.

The abnormal conditions include: camera anomaly, point cloud anomaly, upper ghost point cloud, mirror point cloud, ghost point cloud, among others.

The time includes: morning (0-6), morning (6-12), afternoon (12-18), evening (18-24).

Weather includes: sunny days, cloudy days, rainy days, foggy days, snowy days, and sand storm.

The scene comprises: high speed, urban, rural, campus.

It should be noted that, the above-mentioned partial labels may be labeled by computer automation, for example, driving behavior is determined and labeled based on inertial navigation data and own vehicle data of the vehicle, time labels are determined and labeled based on time stamps, and the like. The user may edit each label of each segment of the data packet.

The tag data generated by the tagging module is also in json format, the contents of which are shown in table 2, for processing by the database.

Table 2 field content of tag data

In some of these embodiments, the data packets are in a ROSbag format, and the marking module includes ROS nodes and creates a visual graphical interface based on PyQt, tkiner, or Django. The ROS node is used for subscribing to the data of the data packet and mapping the data of the data packet to the visual graphical interface. Taking PyQt5+ROS as an example, the method of visualizing ROS data using PyQt5 is as follows: installation of PyQt5: the PyQt5 package is installed using pip or conda. Creating a PyQt5 GUI application: a graphical interface is created using a Qt Designer, or code is written manually. Creating a ROS node: an ROS node is created to subscribe to data from ROS topics. Mapping ROS data to the PyQt5 interface: in the callback function of the ROS node, the signal and slot mechanism of PyQt5 is used to map ROS data into the PyQt5 interface. Running a program: the ROS system and the PyQt5 application program are started, and the visual display of the ROS data on the PyQt5 interface can be seen.

In the labeling module, point cloud labeling can be performed on the data packet segments screened according to the labels through a labeling platform. Wherein only the data packet fragments with object labels may be screened out. And for the point cloud data in the data packet segment with the object label, the point cloud is marked by adding a marking frame and the label and giving different point colors to the point clouds belonging to different objects. In order to test the labeling quality, the labeling module also performs quality inspection on the labeling data in a sampling inspection mode, for example, 10% -30% of the labeling data are sampled and manually inspected, and if the qualification rate is greater than a qualification threshold (for example, 95%), the labeling data of the current batch are considered to be qualified, and quality inspection data of the batch are generated.

In some embodiments, the database includes a web application framework with a visual graphical interface, a conversion unit, and a database management system, the conversion unit is configured to combine query parameters input by the visual graphical interface into query statements of the database management system, and provide query results returned by the database management system to the web application framework to visualize the query results through the visual graphical interface. The web application framework is created based on Django, the conversion unit is written based on Python, and the database management system comprises a mysql database management system.

This embodiment implements all functions by using python+django+mysql in combination, with python as the primary language. django is a Web application framework written by Python and mysql is a common database management system. The visual page is built through django, and then database operation is connected with the visual page through python, so that the function of database management is achieved. The database management tool in the embodiment is used for judging and splicing the information into a complete sql statement by acquiring the input information, transmitting the information to the mysql through python, realizing the return of a result, solving the difficulty of directly inputting the sql statement by a user and realizing the function directly through Chinese.

In addition, because the web application framework and the conversion unit in the database are written based on python and based on the characteristics of automated processing of python, batch data processing can be added into the database, the result is downloaded into excel, and functions such as remote data downloading are integrated into a database management system, so that the convenient use is realized for users.

The embodiment also provides an automatic driving data processing method. The method is applied to the automatic driving data processing system. Fig. 3 is a flowchart of the automatic driving data processing method of the present embodiment, as shown in fig. 3, the flowchart including the steps of:

in step S301, the data interface acquires a data packet acquired by executing a data acquisition task, where the data packet includes data acquired by multiple sensors within a preset duration.

In step S302, the verification module generates verification data according to the validity of the data packet verification data acquisition task, and divides the data packet into a valid data packet and an invalid data packet.

In step S303, the marking module visualizes the valid data packet, displays a label panel for editing labels of the valid data packet, and generates label data of the valid data packet based on the user operation on the label panel, the label data including the object label.

In step S304, the labeling module segments and labels the laser radar point cloud data in the valid data packet with the object tag, generates labeling data, and performs quality inspection on the labeling data to generate quality inspection results.

In step S305, the database stores the data packet, the verification data, the tag data, the label data, and the quality inspection result.

In some of these embodiments, the data packet includes: the laser radar point cloud data and the laser radar calibration parameters acquired by the laser radar, the video data and the camera calibration parameters acquired by the camera, the positioning data acquired by the positioning system, the inertial navigation data acquired by the inertial navigation system and the vehicle data at least comprise: wheel speed data and steering wheel angle data.

In some embodiments, the data packet includes multiple message sets corresponding to the multiple sensors one by one, where each message set includes multiple message sets arranged according to an acquisition sequence, and the message sets carry a timestamp; the verification module verifies the validity of the data acquisition task according to the data packet, and comprises the following steps: verifying the validity of a data acquisition task by verifying the total duration of a data packet and the frame rate stability of a message group in the data packet; the total duration is determined based on the time stamps of the first and last message in the data packet, and the frame rate of the message group is determined based on the time stamps of the first and last message in the statistical interval and the number of message in the statistical interval.

In some of these embodiments, the frame rate stability of the message packet includes a total frame rate stability of the data packet and a segment frame rate stability of each segment after an average segmentation of the data packet.

In some embodiments, the data packet is a data packet in a ROSbag format, and in step S303, the visualized valid data packet includes: creating a visual graphical interface based on PyQt, tkiner or Django; subscribing to the data of the data package through the ROS node and mapping the data of the data package to the visual graphical interface.

In some embodiments, the database includes a web application framework with a visual graphical interface, a conversion unit, and a database management system, the method may further include: and combining the query parameters input by the visual graphical interface into a query statement of the database management system through the conversion unit, and providing the query result returned by the database management system to the web application framework so as to visualize the query result through the visual graphical interface.

By the above examples or preferred embodiments, the following beneficial effects are achieved:

1. in the data production process, the possible problems in various uses are solved, and the simplification and the high efficiency are achieved by using a stable mode and a simplified flow. The problems of complex and low-efficiency data production flow are solved.

2. The database management tools in the embodiments are realized by using open-source and free technology fusion, and the visual interface can be personalized and customized, so that the problems of poor visual effect and payment of the traditional management tools can be solved.

3. The table operation of the database does not need the SQL language to be used by the user, has lower threshold, is suitable for common users to use, and reduces the use difficulty of the database.

4. The django framework can be combined with python to realize various customizing functions, such as automatic data processing, automatic data cleaning and the like, so that the technical problem of poor customizing function of the traditional management tool is solved.

It should be noted that the term "comprising" and its variants as used in the embodiments of the present application are open-ended, i.e. "including but not limited to". The term "based on" is based at least in part on. The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments. References to "one or more" modifications in the examples of the application are intended to be illustrative rather than limiting, and it will be understood by those skilled in the art that "one or more" is intended to be interpreted as "one or more" unless the context clearly indicates otherwise.

User information (including but not limited to user equipment information, user personal information and the like) and data (including but not limited to data for analysis, stored data, presented data and the like) according to the embodiment of the application are information and data authorized by a user or fully authorized by all parties, and the collection, use and processing of related data are required to comply with related laws and regulations and standards of related countries and regions, and are provided with corresponding operation entrances for users to select authorization or rejection.

The steps described in the method embodiments provided in the embodiments of the present application may be performed in a different order and/or performed in parallel. Furthermore, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the application is not limited in this respect.

The term "embodiment" in this specification means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive. The various embodiments in this specification are described in a related manner, with identical and similar parts being referred to each other. In particular, for apparatus, devices, system embodiments, the description is relatively simple as it is substantially similar to method embodiments, see for relevant part of the description of method embodiments.

The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the patent claims. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.

Claims

1. An autopilot data processing system characterized by comprising:

2. The system of claim 1, wherein the data packet comprises: the system comprises laser radar point cloud data and laser radar calibration parameters collected by a laser radar, video data and camera calibration parameters collected by a camera, positioning data collected by a positioning system, inertial navigation data collected by an inertial navigation system and vehicle data, wherein the vehicle data at least comprises: wheel speed data and steering wheel angle data.

3. The system of claim 1, wherein the data packet comprises a plurality of message sets in one-to-one correspondence with the plurality of sensors, each message set comprising a plurality of message messages arranged in a collection order, the message messages carrying a time stamp; the verification module verifies the validity of the data acquisition task by verifying the total duration of the data packet and the frame rate stability of the message packet in the data packet; the total duration is determined based on the time stamps of the first message and the last message of the data Bao Nadi, and the frame rate of the message group is determined based on the time stamps of the first message and the last message in the statistical interval and the number of message in the statistical interval.

4. The system of claim 3, wherein the frame rate stability of the message packet comprises a total frame rate stability of the data packet and a segment frame rate stability of each segment after averaging segments of the data packet.

5. The system of claim 1, wherein the label panel includes a label control for editing at least the following labels: road type, special targets, driving behavior, special scenes, buildings, abnormal conditions, time periods, weather, and scenes.

6. The system of claim 1, wherein the data packets are data packets in a ROSbag format, the marking module includes ROS nodes and creates a visual graphical interface based on PyQt, tkter, or Django; the ROS node is used for subscribing to the data of the data packet and mapping the data of the data packet to the visual graphical interface.

7. The system of claim 1, wherein the database comprises a web application framework having a visualization graphical interface, a conversion unit, and a database management system, the conversion unit configured to combine query parameters entered by the visualization graphical interface into query statements for the database management system, and provide query results returned by the database management system to the web application framework for visualization of the query results via the visualization graphical interface.

8. The system of claim 7, wherein the web application framework is created based on Django, the conversion unit is written based on Python, and the database management system comprises a mysql database management system.

9. An automatic driving data processing method applied to the system according to any one of claims 1 to 8, characterized by comprising:

10. The method of claim 9, wherein the data packet includes a plurality of message sets in one-to-one correspondence with the plurality of sensors, each message set including a plurality of message messages arranged in a collection order, the message messages carrying a time stamp; checking the validity of the data acquisition task according to the data packet comprises the following steps:

11. The method of claim 10, wherein the frame rate stability of the message packet comprises a total frame rate stability of the data packet and a segment frame rate stability of each segment after averaging segments of the data packet.

12. The method of claim 9, wherein the data packet is a ROSbag format data packet, and visualizing the valid data packet comprises:

creating a visual graphical interface based on PyQt, tkiner or Django;

13. The method of claim 9, wherein the database comprises a web application framework with a visual graphical interface, a conversion unit, and a database management system, the method further comprising: