CN115567600A - Message processing system and method based on vehicle-end computing resources - Google Patents
Message processing system and method based on vehicle-end computing resources Download PDFInfo
- Publication number
- CN115567600A CN115567600A CN202211059185.4A CN202211059185A CN115567600A CN 115567600 A CN115567600 A CN 115567600A CN 202211059185 A CN202211059185 A CN 202211059185A CN 115567600 A CN115567600 A CN 115567600A
- Authority
- CN
- China
- Prior art keywords
- message
- distribution
- vehicle
- interval
- computing resources
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/54—Interprogram communication
- G06F9/546—Message passing systems or structures, e.g. queues
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a message processing system and a method based on vehicle-end computing resources, which comprises the following steps: the real-time message module is used for processing the vehicle data in real time to obtain a first message; and the controllable message module is used for determining the distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first message according to the distribution interval to obtain a second message. The method can control the running of the associated automatic driving algorithm according to the requirement, avoid the additional consumption of the vehicle end computing resource and reduce the influence of the shadow mode on the normal running of the automatic driving algorithm.
Description
Technical Field
The invention relates to the technical field of automatic driving, in particular to a message processing system and method based on vehicle-side computing resources.
Background
The automatic driving, also called unmanned driving, acquires environmental information through sensors such as radar and controls the vehicle to run according to the environmental information, thereby reducing the time for the driver to control the vehicle. However, unmanned driving also faces many challenges, such as bad weather, complex traffic environments, etc., and thus may not perform as well as human drivers in certain scenarios. To solve this problem, more tests and iterative updates need to be performed on the autopilot model to improve the model throughput in complex scenarios.
In the test updating of the automatic driving model, the automatic driving model needs to be mounted on a vehicle to acquire and collect automatic driving data. How to efficiently screen out the extreme working condition data from the massive automatic driving data and only transmit the part of data back to the background is a new problem. With the introduction of the "shadow mode", the above problem is effectively solved. Specifically, the "shadow mode" means that in a manned driving state, the automatic driving system including the sensor still runs but does not participate in vehicle control, only the decision algorithm is verified, namely the algorithm of the system is subjected to continuous decision simulation in the "shadow mode", and the simulated decision is compared with the behavior of the driver, and once the two are inconsistent, the scene is judged to be an "extreme working condition", and then data return is triggered.
Under the "shadow mode", an autonomous controllable shadow vehicle still exists in the automatic driving system, and at this moment, a whole set of associated automatic driving algorithm needs to be configured on a vehicle-end computing chip to support the computation of the behavior of the shadow vehicle. However, since vehicle-end computing resources are scarce resources, and the operation of the associated automatic driving algorithm cannot affect the operation of the normal automatic driving algorithm, how to ensure the operation of the normal automatic driving algorithm and provide support for the operation of the associated automatic driving algorithm under the condition of limited computing resources becomes an urgent problem to be solved.
Disclosure of Invention
The invention provides a message processing system and method based on vehicle-end computing resources, which are used for solving the problems.
The invention provides a message processing system based on vehicle-end computing resources, which comprises:
the real-time message module is used for processing the vehicle data in real time to obtain a first message;
and the controllable message module is used for determining the distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first message according to the distribution interval to obtain a second message.
According to the message processing system based on the vehicle-end computing resources, the controllable message module comprises a message driving unit and an algorithm running unit, and the algorithm running unit comprises a plurality of algorithm running subunits; wherein,
the message driving unit determines the distribution interval at the current moment by utilizing a predefined dynamic distribution interval adjustment rule according to the current vehicle-end computing resource and the number of the first messages, and distributes the first messages to corresponding algorithm operation subunits according to the distribution interval at the current moment based on the association relationship between the algorithm operation subunits and the first messages;
and the algorithm operation subunit acquires the corresponding first message, and processes the corresponding first message by using a pre-stored test algorithm to obtain the second message.
According to the message processing system based on the vehicle-end computing resources, the message driving unit is further used for determining the sequence of distributing the first messages to the corresponding algorithm operation sub-units according to the topological relation among the first messages.
According to the message processing system based on the vehicle-end computing resources, the controllable message module further comprises a message queue forming unit and a message distributing unit;
the message queue forming unit is used for acquiring a plurality of first messages and forming a message queue;
correspondingly, the message driving unit determines the order of distributing to the corresponding algorithm operation sub-units according to the topological relation among the first messages, and the method comprises the following steps:
the message driving unit determines the sequence distributed to the corresponding algorithm operation subunit according to the topological relation among the first messages in the message queue, and sends the first messages in the message queue to the corresponding algorithm operation subunit according to the sequence and the distribution interval of the current moment;
the message distribution unit places the second message in the message queue according to a predefined distribution rule to form a new message queue.
According to the message processing system based on the vehicle-end computing resources, the system further comprises a data caching module, and the controllable message module further comprises a data reading unit; wherein,
the data caching module is respectively connected with the controllable message module and the real-time message module and is used for caching the first message;
the data reading unit is used for reading a plurality of first messages from the data cache module according to the data reading interval time and pushing the plurality of first messages to the message queue forming unit; and the data reading interval time is determined according to the queue length information fed back by the message queue forming unit.
According to the message processing system based on the vehicle-end computing resource provided by the invention, the controllable message module comprises:
and the resource detection unit is used for detecting the hardware resources of the vehicle end so as to obtain the current vehicle end computing resources.
The invention also provides a message processing method based on the vehicle-end computing resource, which is realized based on the message processing system based on the vehicle-end computing resource and comprises the following steps:
processing vehicle data in real time to obtain a first message;
and determining the distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first message according to the distribution interval to obtain a second message.
According to the message processing method based on the vehicle-end computing resource provided by the invention, the step of determining the distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first message according to the distribution interval to obtain the second message comprises the following steps:
according to the current vehicle end computing resource and the number of the first messages, determining the distribution interval at the current moment by utilizing a predefined distribution interval dynamic adjustment rule;
distributing the first message according to the distribution interval of the current moment based on the incidence relation between a test algorithm and the first message;
and processing the distributed first message by using a test algorithm to obtain a second message.
According to the message processing method based on the vehicle-end computing resource, the method for determining the distribution interval at the current moment by utilizing the predefined dynamic distribution interval regulation rule comprises the following steps:
acquiring an initial time interval, current vehicle-end computing resources and a distribution interval calibration table, wherein the distribution interval calibration table is composed of message distribution time intervals and historical computing resources and is used for representing the corresponding relation between the message distribution time intervals and the historical computing resources;
inquiring whether historical computing resources matched with target computing resources exist in the distribution interval calibration table or not;
under the condition that historical computing resources matched with the target computing resources exist, determining the distribution interval of the current moment according to the message distribution time interval corresponding to the historical computing resources obtained through matching;
comparing the size relationship between the target computing resource and the current vehicle-end computing resource under the condition that no historical computing resource matched with the target computing resource exists;
under the condition that the target computing resource is larger than the current vehicle-end computing resource, the initial time interval is prolonged, and the prolonged initial time interval is used as the distribution interval of the current time;
and under the condition that the target computing resource is not larger than the current vehicle-end computing resource, shortening the initial time interval, and taking the shortened initial time interval as the distribution interval of the current time.
According to the message processing method based on the vehicle-end computing resource, provided by the invention, under the condition that the historical computing resource matched with the target computing resource exists, the distribution interval of the current moment is determined according to the message distribution time interval corresponding to the historical computing resource obtained by matching, and the method comprises the following steps:
under the condition that one historical computing resource is matched with the target computing resource, taking the time interval of message distribution corresponding to the matched historical computing resource as the distribution interval of the current moment;
and under the condition that the number of the historical computing resources matched with the target computing resource is two, carrying out average value calculation on the message distribution time intervals corresponding to the two matched historical computing resources, and taking the average value calculation result as the distribution interval of the current moment.
According to the message processing method based on the vehicle-end computing resources, the distribution interval calibration table is a distribution interval calibration table which is sorted according to the size of the message distribution time interval;
accordingly, after determining the distribution interval at the current time using the predefined distribution interval dynamic adjustment rule, the method further comprises:
calculating the difference between the distribution interval of the current moment and the distribution time interval of each message in the distribution interval calibration table;
judging whether the difference value exceeds a preset interval threshold value or not;
and correspondingly inserting the distribution interval of the current moment and the current vehicle-end computing resource into the distribution interval calibration table according to the size of the distribution interval of the current moment to form a new distribution interval calibration table under the condition that the difference exceeds a preset interval threshold.
According to the message processing method based on vehicle-end computing resources, before the distribution interval calibration table is obtained, the method further comprises the following steps:
judging whether the storage time of each message distribution time interval in the distribution interval calibration table and the corresponding historical computing resource in the distribution interval calibration table exceeds a preset storage time threshold value or not;
and deleting the corresponding message distribution time interval and the corresponding historical computing resource under the condition that the message distribution time interval exceeds a preset storage time threshold value to form a new distribution interval calibration table.
According to the message processing method based on the vehicle-end computing resource, provided by the invention, the current vehicle-end computing resource is obtained, and the method comprises the following steps:
under the condition of stable resource consumption, obtaining current vehicle-end computing resources; and the resource consumption is stable, namely the fluctuation of the vehicle-end computing resources is within a preset interval range.
According to the message processing method based on the vehicle-end computing resource provided by the invention, before the step of determining the distribution interval at the current moment by utilizing the predefined distribution interval dynamic adjustment rule according to the current vehicle-end computing resource and the number of the first messages, the method further comprises the following steps:
acquiring a plurality of first messages and forming a message queue;
correspondingly, before the first message is distributed according to the distribution interval of the current time based on the incidence relation between the test algorithm and the first message, the method further includes:
determining a distribution sequence according to a topological relation among the first messages in the message queue;
the distributing the first message according to the distribution interval of the current time based on the incidence relation between the test algorithm and the first message comprises the following steps:
distributing the first message in the message queue according to the sequence and the distribution interval of the current moment;
after the processing the distributed first message by using the test algorithm to obtain the second message, the method further includes:
and placing the second message in the message queue according to a predefined distribution rule to form a new message queue.
According to the message processing method based on the vehicle-end computing resource provided by the invention, the step of placing the second message in the message queue according to the predefined distribution rule comprises the following steps:
under the condition that the triggering mechanism of the first message is clock triggering, predicting an expected clock of the next first message according to a clock triggering rule, and correspondingly placing the second message in the message queue according to the expected clock;
and under the condition that the triggering mechanism of the first message is event triggering, acquiring the topological relation between the first message and the second message according to an event triggering rule, and correspondingly placing the second message in the message queue based on the topological relation between the first message and the second message.
The invention also provides an electronic device, which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein when the processor executes the program, the message processing method based on the vehicle-end computing resource is realized.
The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, wherein the computer program, when executed by a processor, implements any of the above-mentioned message processing methods based on vehicle-side computing resources.
The message processing system based on the vehicle-end computing resources determines the distribution interval of the first message through the controllable message module, distributes and processes the first message according to the distribution interval, and can control the operation of the associated automatic driving algorithm as required under the condition of ensuring the normal operation of the automatic driving algorithm in the message module, thereby avoiding the additional consumption of the vehicle-end computing resources and reducing the influence of the shadow mode on the normal operation of the automatic driving algorithm.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic structural diagram of a message processing system based on vehicle-side computing resources according to an embodiment of the present invention;
FIG. 2 is a second schematic structural diagram of a message processing system based on vehicle-end computing resources according to an embodiment of the present invention;
fig. 3 is a schematic flowchart of a message processing method based on vehicle-side computing resources according to an embodiment of the present invention;
fig. 4 is a flowchart illustrating a process of determining a distribution interval at a current time according to an embodiment of the present invention;
fig. 5 is a schematic physical structure diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Currently, the operation of autopilot systems generally relies on a message middleware framework. Generally, there are two message triggering mechanisms of the message middleware: the method comprises the steps of clock triggering and event triggering, wherein message middleware drives continuous messages to flow in the whole automatic driving system based on two message pushing strategies of clock triggering and event triggering.
Since the autopilot system is targeted for optimization in "real time" and each frame trigger opportunity is deterministic and clock dependent. Under the condition of flexible and variable consumption of single frame resources, frequent jitter of the use state of the whole system resources is brought. Based on the background, the invention provides a message processing system and method based on vehicle-end computing resources, which can meet the real-time requirement of normal algorithm operation in an automatic driving system by controlling the consumption of system resources by an associated automatic driving algorithm according to the use state of the system resources. The following describes the message processing system and method based on vehicle-end computing resources provided by the present invention.
FIG. 1 is a schematic structural diagram of a message processing system based on vehicle-side computing resources according to an embodiment of the present invention; FIG. 2 is a second schematic structural diagram of a message processing system based on vehicle-side computing resources according to an embodiment of the present invention; as shown in fig. 1 and fig. 2, the message processing system based on vehicle-side computing resources includes a real-time message module 110, a data caching module 130, and a controllable message module 120. Among other things, the controllable message module 120 includes:
the real-time message module 110 is configured to process the vehicle data in real time to obtain a first message. Specifically, an autopilot system algorithm is run in the real-time message module 110, and the autopilot system algorithm calculates and processes vehicle data to form a first message. Wherein, the vehicle data are obtained through different sensor perceptions.
And the controllable message module 120 is configured to determine a distribution interval of the first message according to the current vehicle-end computing resource and the number of the first message, and distribute and process the first message according to the distribution interval to obtain a second message.
The current vehicle-end computing resource is obtained under the condition that the algorithm of the automatic driving system normally runs, and the current vehicle-end computing resource is a hardware resource of a vehicle end, such as a resource of a CPU, a memory (directly displayable through a memory display program MEM), an I/O and the like. The method comprises the steps that a distribution interval of a first message is preliminarily determined according to current vehicle-end computing resources, specifically, if the current vehicle-end computing resources are idle, the distribution time interval of the first message is shortened a bit, the distribution frequency is improved, the message processing speed is increased, and the utilization rate of the computing resources is improved; if the current vehicle-end computing resource is busy, the distribution time interval of the first message is prolonged, the distribution frequency is reduced, and the vehicle-end computing resource is stabilized.
The distribution interval of the first message needs to be adjusted according to the number of the first messages in addition to the current vehicle-end computing resource, for example, if the number of the first messages sent by the real-time message module 110 is too large, the distribution interval of the first messages is further shortened, the message processing rate is increased, and excessive accumulation is avoided. And if the first message does not form excessive accumulation, determining the distribution interval of the first message according to the current vehicle-end computing resource. After the distribution interval is determined, the first message is distributed according to the determined distribution interval, and the distributed first message is processed by using an automatic driving test algorithm to obtain a second message.
The data caching module 130 is connected to the controllable message module 120 and the real-time message module 110, respectively, and is configured to cache the first message.
In this embodiment, since the real-time message module 110 processes the vehicle data in real time, and the controllable message module 120 does not process the first message in real time, a data cache module 130 is needed between the two modules to cache the first message obtained by real-time processing, so that the controllable message module 120 can obtain the first message from the data cache module 130 according to the current vehicle-end computing resource and the operation requirement. The real-time message module 110 will continuously push the first message to the data caching module 130.
In the message processing system based on vehicle-side computing resources provided by the embodiment of the present invention, the controllable message module 120 determines the distribution interval of the first message, and distributes and processes the first message according to the distribution interval, so that the operation of the associated autopilot algorithm can be controlled as needed under the condition that the normal operation of the autopilot algorithm in the message module 110 is guaranteed, thereby avoiding additional consumption of vehicle-side computing resources and reducing the influence of the shadow mode on the normal operation of the autopilot algorithm.
Further, the controllable message module 120 includes a data reading unit 121, a message queue forming unit 122, a resource detecting unit 123, a message driving unit 124, an algorithm executing unit 125, and a message allocating unit 126. Wherein the algorithm execution unit 125 includes a plurality of algorithm execution sub-units.
The data reading unit 121 reads a plurality of first messages from the data caching module 130 according to the data reading interval time, and pushes the plurality of first messages to the message queue forming unit 122; the data reading interval time is determined according to the queue length information fed back by the message queue forming unit 122.
Specifically, the data reading unit 121 generally delays reading the first message in the data buffer module 130 in batches for several seconds or in stages, where the specific delay time (i.e. the data reading interval time) depends on the length information of the current processing queue fed back by the message queue forming unit 122, and if the current processing queue is longer, the delay time is longer; accordingly, if the current processing queue is short, the delay time is short. After the data reading unit 121 reads the plurality of first messages, the first messages are pushed to the message queue forming unit 122.
The message queue forming unit 122 is configured to obtain a plurality of first messages and form a message queue. Specifically, the message queue forming unit 122 forms the message queue after obtaining the first message pushed by the data reading unit 121.
Further, the message driving unit 124 determines a distribution interval at the current time by using a predefined dynamic distribution interval adjustment rule according to current vehicle-end computing resources and the number of the first messages, determines a sequence of distributing to a corresponding algorithm running subunit according to a topological relation between the first messages in the message queue, and finally sends the first messages in the message queue to the corresponding algorithm running subunit according to the sequence and the distribution interval at the current time.
And the algorithm operation subunit acquires the corresponding first message and processes the corresponding first message by utilizing a pre-stored test algorithm to obtain the second message.
Specifically, the message driving unit 124 obtains the first message from the formed message queue, and determines the order of sending to each algorithm operation subunit according to the topological relation between each first message in the message queue, for example, there are a first message a, a first message B, a first message C, and a first message D, where the first message a depends on the first message B, the first message B depends on the first message C, and the first message C depends on the first message D, then the first message D is taken as the message to be processed first, and so on, and then the first message C, the first message B, and the first message a are processed next. Through the processing, the first messages can be ensured to be aligned, and the topological relation among the messages cannot be damaged because the subsequent algorithm operation subunits process the first messages at first due to different processing time of the first messages.
In addition to the above-mentioned "alignment" processing of the first message, the "flow control" processing of the first message is also performed according to the current vehicle-end computing resource and the number of the first message. Specifically, the current end-of-vehicle computing resources and the number of the first messages determine the frequency of distribution of the first messages by the message driving unit 124. For example, when the current vehicle-end computing resource feedback "the hardware resource is relatively free", the sending frequency of the message driving unit 124 may be increased a little bit, that is, the message distribution time interval is shortened. If the data reading unit 121 feeds back "the current data buffer module accumulates too much data", the sending frequency of the message driving unit 124 is increased. The specific determination process of the distribution interval at the current time is described below.
In addition, the current vehicle-end computing resource is detected by the resource detecting unit 123.
After the distribution interval and the distribution sequence of the first message at the current moment are obtained, the first message is distributed to the corresponding algorithm operation sub-units according to the sequence and the distribution interval at the current moment according to the incidence relation between the first message and the algorithm operation sub-units.
In this context, the association relationship between the first message and the algorithm running subunit refers to a corresponding relationship between the test algorithm and the first message, for example, if the first message a is used for the sensing test, the message driving unit 124 sends the first message a to the algorithm running subunit in which the sensing algorithm is stored, so that the subsequent sensing algorithm can perform calculation processing based on the related first message, and resources are saved.
Further, the message distribution unit 126 places the second message in the message queue according to a predefined distribution rule, forming a new message queue.
And after each algorithm operation subunit utilizes different test algorithms to perform calculation processing on the corresponding first message, obtaining a second message. The message distributing unit 126 inserts the second message into the message queue according to different message triggering mechanisms, and forms a new message queue with the supplemented first message at the next time in the message queue, and the data driving unit determines the message distribution interval at the next time according to the new message queue, continuously calculates and processes the first message, and then generates a new second message to be inserted into the message queue, thereby forming a continuous closed loop of the consumption of the first message and the generation of the second message.
Wherein the predefined allocation rules are described below.
According to the message processing system based on the vehicle-end computing resources, the distribution interval of the current moment is determined by utilizing the predefined dynamic distribution interval adjustment rule, the frequency control of the first message distribution is realized, and the normal operation of an automatic driving system is not influenced by the operation of an automatic driving test algorithm (namely an associated automatic driving algorithm). And determining the distribution sequence of the first messages according to the topological relation among the first messages, ensuring the accuracy of the message processing on the sequence and ensuring that the processed messages still have the topological relation.
The message processing method based on the vehicle-side computing resource provided by the invention is described below, and the message processing method based on the vehicle-side computing resource described below is realized based on the message processing system based on the vehicle-side computing resource described above.
Fig. 3 is a schematic flow chart of a message processing method based on vehicle-side computing resources according to an embodiment of the present invention, and as shown in fig. 3, the message processing method based on vehicle-side computing resources includes the following steps:
s301, vehicle data are processed in real time to obtain a first message.
In this step, the vehicle data is calculated and processed by the autopilot system algorithm to form a first message. Wherein, the vehicle data are obtained through different sensor perceptions.
S302, determining a distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first messages according to the distribution interval to obtain a second message.
In this step, the current vehicle-end computing resource is obtained under the condition that the algorithm of the automatic driving system is normally running, and the current vehicle-end computing resource is a hardware resource of the vehicle end, such as a resource of a CPU, a memory (which can be directly displayed through a memory display program MEM), an I/O, and the like. The method comprises the steps that a distribution interval of a first message is preliminarily determined according to current vehicle-end computing resources, specifically, if the current vehicle-end computing resources are idle, the distribution time interval of the first message is shortened a little, distribution frequency is improved, the processing speed of the message is increased, and the utilization rate of the computing resources is improved; if the current vehicle-end computing resource is busy, the distribution time interval of the first message is prolonged, the distribution frequency is reduced, and the vehicle-end computing resource is stabilized.
The distribution interval of the first message is adjusted according to the number of the first messages except for considering the current vehicle-side computing resource, for example, if the number of the first messages sent by the real-time message module is too large, the distribution interval of the first messages is further shortened, the message processing rate is increased, and excessive accumulation is avoided. And if the first message does not form excessive accumulation, determining the distribution interval of the first message according to the current vehicle-end computing resource.
After the distribution interval is determined, the first message is distributed according to the determined distribution interval, and the distributed first message is processed by using an automatic driving test algorithm to obtain a second message.
According to the message processing method based on the vehicle-side computing resources, the distribution interval of the first message is determined according to the current vehicle-side computing resources and the number of the first messages, the first message is distributed and processed according to the distribution interval, the running of the associated automatic driving algorithm can be controlled according to needs under the condition that the normal running of the automatic driving algorithm is guaranteed, the extra consumption of the vehicle-side computing resources is avoided, and the influence of the shadow mode on the normal running of the automatic driving algorithm is reduced.
Further, the determining a distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first message according to the distribution interval to obtain a second message includes:
and determining the distribution interval at the current moment by utilizing a predefined distribution interval dynamic adjustment rule according to the current vehicle-end computing resource and the number of the first messages.
And distributing the first message according to the distribution interval of the current moment based on the incidence relation between the test algorithm and the first message.
And processing the distributed first message by using a test algorithm to obtain a second message.
The test algorithm is an algorithm applied to automatic driving in a shadow mode, and can be divided into a perception type test algorithm, a control type test algorithm and a planning type test algorithm. The association relationship between the test algorithm and the first message means that different first messages are applied to different types of test algorithms, for example, the first message a is used for perception test, and the first message a is used as the input of the perception test algorithm and is distributed to the corresponding perception test algorithm.
In this embodiment, after obtaining the current vehicle-end computing resource and the number of the first messages, the first messages are subjected to "flow control" processing. Specifically, if the current vehicle-end computing resource feeds back that "the hardware resource is relatively idle", the sending frequency of the message driving unit will be increased a little, that is, the distribution time interval of the message is shortened. If the data reading unit feeds back that "the current data buffer module accumulates too much data", the sending frequency of the message driving unit will be increased, that is, the distribution time interval of the message is prolonged. The specific determination process of the distribution interval at the current time is described below.
After the distribution interval of the current moment is determined, the first message is distributed to the corresponding test algorithm according to the distribution interval, and the test algorithm carries out calculation processing according to the distributed first message to obtain a second message.
According to the message processing method based on vehicle-side computing resources, the distribution interval of the current moment is determined by utilizing the predefined dynamic distribution interval adjustment rule, the frequency control of the first message distribution is realized, the normal operation of an automatic driving system is not influenced by the operation of an automatic driving test algorithm, and the resource influence caused by the operation of a shadow mode is reduced to the maximum extent.
Fig. 4 is a flowchart illustrating a process of determining a distribution interval at a current time according to an embodiment of the present invention; as shown in fig. 4, the determining the distribution interval at the current time by using the predefined distribution interval dynamic adjustment rule includes:
the method comprises the steps of obtaining an initial time interval, current vehicle-end computing resources and a distribution interval calibration table, wherein the distribution interval calibration table is composed of message distribution time intervals and historical computing resources and is used for representing the corresponding relation between the message distribution time intervals and the historical computing resources.
And inquiring whether historical computing resources matched with the target computing resources exist in the distribution interval calibration table or not.
And under the condition that the historical computing resources matched with the target computing resources exist, determining the distribution interval of the current moment according to the message distribution time interval corresponding to the historical computing resources obtained by matching.
And under the condition that no historical computing resource matched with the target computing resource exists, comparing the size relationship between the target computing resource and the current vehicle-side computing resource.
And under the condition that the target computing resource is larger than the current vehicle-end computing resource, prolonging the initial time interval, and taking the prolonged initial time interval as the distribution interval of the current time.
And under the condition that the target computing resource is not larger than the current vehicle-end computing resource, shortening the initial time interval, and taking the shortened initial time interval as the distribution interval of the current moment.
Wherein the initial time interval is obtained by initializing the message distribution time interval. The current vehicle-end computing resource is obtained by detecting hardware resources (such as a CPU, a memory, an I/O and the like) on the vehicle end. The target computing resource is a preset target value, so that the overall use condition of the hardware resource is expected to be stable at the target value, namely the vehicle-end computing resource at each moment is expected to fluctuate around the target computing resource.
It should be noted that, in addition to the target computing resource, a computing resource critical value is configured in advance, and the current vehicle-end computing resource may not exceed the computing resource critical value.
In addition, when the first message is too much accumulated, the distribution frequency can be further increased, so that the current vehicle-end computing resource is properly over the target computing resource, but cannot exceed the computing resource critical value.
The distribution interval calibration table stores the time intervals of historical message distribution and also stores historical computing resources corresponding to the time intervals of historical message distribution. Specifically, as shown in table 1, the time interval of message distribution and the historical computing resources are data pairs, and the data pairs in the distribution interval calibration table are sorted according to the size of the time interval of message distribution, that is, a > b >. N, or a < b >. N.
TABLE 1 Dispenser Interval calibration schematic
| Time interval of message distribution | Historical computing resources |
| a seconds | Vehicle-end computing resource at time Ta |
| b second | Vehicle end computing resource at time Tb |
| ... | ... |
| n seconds | Tn time vehicle end computing resource |
In this embodiment, an initial time interval, current vehicle-end computing resources, and a distribution interval calibration table are obtained first; then, the historical computing resources matched with the target computing resources are inquired in the distribution interval calibration table, and if the historical computing resources matched with the target computing resources can be inquired, the distribution interval of the current moment is determined according to the message distribution time interval corresponding to the inquired historical computing resources. If no matched historical computing resource is inquired in the distribution interval calibration table, determining the distribution interval at the current moment by the following method:
and comparing the magnitude relation of the two values of the target computing resource and the current vehicle-end computing resource.
If the target computing resource is larger than the current vehicle-end computing resource, it indicates that the current vehicle-end computing resource cannot meet the requirement of the target computing resource, and the distribution frequency of the first message needs to be reduced, the initial time interval is prolonged, and the pressure of the current hardware resource is relieved. The specific initial time interval extension is to multiply the initial time interval basis by a preset multiple, wherein the preset multiple can be 1.5, 2, 3 and the like, and the extended initial time interval is used as the distribution interval of the current time according to the setting of requirements or expert experience.
If the target computing resource is less than or equal to the current vehicle-end computing resource, it indicates that the current vehicle-end computing resource is still idle, so that the distribution frequency of the first message can be increased, the initial time interval can be shortened, and the pressure utilization rate of the current hardware resource can be increased. The specific initial time interval shortening is to divide the initial time interval by a preset multiple, the preset multiple may be 1.5, 2, 3, or the like, and may be different from or the same as the multiple mentioned in the above extension process, and the shortened initial time interval is set as the distribution interval at the current time according to the requirement or the expert experience.
According to the message processing method based on the vehicle-end computing resource, provided by the embodiment of the invention, whether the historical computing resource matched with the target computing resource exists in the distribution interval calibration table is inquired, and the distribution interval at the current moment is directly determined according to the historical computing resource under the condition of being obtained by inquiry, so that after the first message is distributed and processed according to the determined distribution interval at the current moment, the vehicle-end computing resource can approach the target computing resource, and the whole hardware resource can be in a stable state. Under the condition that historical computing resources are not inquired, how to determine the distribution interval at the current moment by adjusting the initial time interval is further considered, and whether to increase the distribution frequency or decrease the distribution frequency is determined according to the size relation between the target computing resource and the current vehicle-end computing resource, so that the operation of the shadow mode is not influenced on the operation of a normal algorithm, and the additional consumption of hardware resources is reduced.
Further, in the case that there is a historical computing resource matching the target computing resource, determining a distribution interval at the current time according to a message distribution time interval corresponding to the historical computing resource obtained by matching, including:
and under the condition that one historical computing resource is matched with the target computing resource, taking the time interval of message distribution corresponding to the matched historical computing resource as the distribution interval of the current moment.
And under the condition that the number of the historical computing resources matched with the target computing resource is two, carrying out average value calculation on the message distribution time intervals corresponding to the two historical computing resources obtained by matching, and taking the average value calculation result as the distribution interval at the current moment.
Specifically, if the value of exactly one historical computing resource in the distribution interval calibration table is the same as or very close to the value of the configured target computing resource, the time interval of message distribution corresponding to the historical computing resource is directly used as the distribution interval at the current time.
If the value of the target computing resource happens to fall between two historical computing resources and the difference between the two historical computing resources and the target computing resource does not exceed the predetermined threshold, then the two historical computing resources and the target computing resource are matched. For example, if the target computing resource is 80%, the two adjacent historical computing resources are 75% and 90%, respectively, and both of the two historical computing resources are closer to the target computing resource, indicating that the two historical computing resources match the target computing resource. If the target computing resource is 60%, the two adjacent historical computing resources are respectively 10% and 90%, and the two historical computing resources are more different from the target computing resource, which indicates that the two historical computing resources are not matched with the target computing resource.
And under the condition that the two historical computing resources are obtained through matching, taking the average value of the time intervals of message distribution corresponding to the two historical computing resources as the distribution interval at the current moment.
In addition, if the target computing resource is much larger or much smaller than any historical computing resource in the distribution interval calibration table, the unmatching success is also indicated.
According to the message processing method based on the vehicle-end computing resource, the distribution interval corresponding to the target computing resource is determined according to the distribution interval calibration table, so that the vehicle-end computing resource can reach the target state after the first message is distributed according to the determined distribution interval, and the stable operation of the automatic driving system is guaranteed.
Further, the distribution interval calibration table is a distribution interval calibration table sorted according to the size of the message distribution time interval.
Accordingly, after determining the distribution interval at the current time by using the predefined dynamic distribution interval adjustment rule, the method further comprises:
and calculating the difference between the distribution interval of the current moment and the distribution time interval of each message in the distribution interval calibration table.
And judging whether the difference value exceeds a preset interval threshold value.
And correspondingly inserting the distribution interval of the current moment and the current vehicle-side computing resource into the distribution interval calibration table according to the size of the distribution interval of the current moment to form a new distribution interval calibration table under the condition that the difference exceeds a preset interval threshold.
After the distribution interval at the current moment is determined according to the predefined dynamic distribution interval adjustment rule, the distribution interval at the current moment and the current vehicle-side computing resource are bound into a pair of data pairs, and the data pairs are inserted into a distribution interval calibration table according to the size of the distribution interval at the current moment.
It should be noted that, if the value of the distribution interval at the current time is very close to the existing distribution interval in the distribution interval calibration table, the distribution interval at the current time and the current vehicle-side computing resource do not contribute much to the distribution interval calibration table, so that no insertion is performed.
Further, before obtaining the distribution interval calibration table, the method further includes:
judging whether the storage time of each message distribution time interval in the distribution interval calibration table and the corresponding historical computing resource in the distribution interval calibration table exceeds a preset storage time threshold value or not;
and deleting the corresponding message distribution time interval and the corresponding historical computing resource under the condition that the preset storage time threshold is exceeded, and forming a new distribution interval calibration table.
In this embodiment, the data pairs in the distribution interval calibration table have a certain timeliness, each data pair cannot exceed a preset storage time (for example, 3 minutes, 5 minutes, and the like) in the distribution interval calibration table, and the exceeded data pair is eliminated, so that the distribution interval calibration table is guaranteed to have timeliness.
Further, obtaining current vehicle-end computing resources includes:
and under the condition of stable resource consumption, acquiring the current vehicle-end computing resource.
And the resource consumption is stable, namely the fluctuation of the vehicle-end computing resources is within a preset interval range.
In this embodiment, at the previous time, after the test algorithm processes the distributed first message, the resource consumption may fluctuate for several seconds to several tens of seconds, and when the fluctuation exceeds the preset interval range, the acquired vehicle-side calculation resource at the current time is inaccurate and needs to be acquired in a stable state of the resource consumption, that is, after waiting for several seconds or several tens of seconds, the vehicle-side calculation resource at the current time is acquired.
Further, before the determining, according to the current vehicle-end computing resource and the number of the first messages, the distribution interval at the current time by using a predefined dynamic distribution interval adjustment rule, the method further includes:
a plurality of first messages are obtained, and a message queue is formed.
Correspondingly, before the first message is distributed according to the distribution interval of the current time based on the incidence relation between the test algorithm and the first message, the method further includes:
determining a distribution sequence according to a topological relation among the first messages in the message queue;
the distributing the first message according to the distribution interval of the current time based on the incidence relation between the test algorithm and the first message comprises the following steps:
and distributing the first message in the message queue according to the sequence and the distribution interval of the current moment.
After the processing the distributed first message by using the test algorithm to obtain the second message, the method further comprises:
and placing the second message in the message queue according to a predefined distribution rule to form a new message queue.
In this embodiment, in addition to the "flow control" processing of the first message, the "alignment" processing is also performed. Specifically, a distribution sequence is determined according to a topological relation between first messages in the message queue, and the first messages are distributed to corresponding test algorithms according to the distribution sequence and the distribution interval of the current time.
And processing the first messages respectively distributed by using different test algorithms to obtain second messages.
The second message is placed in the correct position of the message queue according to predefined allocation rules, thereby forming a new message queue.
According to the message processing method based on the vehicle-end computing resources, the distribution sequence is determined according to the topological relation among the first messages, and the second messages are placed at the correct positions of the message queues according to the predefined distribution rule, so that the topological relation among the processed messages is not damaged.
Further, the placing the second message in the message queue according to the predefined allocation rule includes:
under the condition that the triggering mechanism of the first message is clock triggering, predicting to obtain an expected clock of the next first message according to a clock triggering rule, and correspondingly placing the second message in the message queue according to the expected clock;
and under the condition that the triggering mechanism of the first message is event triggering, acquiring the topological relation between the first message and the second message according to an event triggering rule, and correspondingly placing the second message in the message queue based on the topological relation between the first message and the second message.
Specifically, for a first message triggered based on a clock, an expected clock of the next first message triggered is inferred according to a clock rule, and a second message is placed on a correct clock position of a message queue.
And calculating the dependency topology relation of the first message and the second message based on the event triggering rule aiming at the first message triggered based on the event, thereby determining the correct position of the second message in the message queue.
According to the message processing method based on the vehicle-end computing resources, provided by the embodiment of the invention, different strategies are adopted to place the second message at the correct position in the message queue aiming at the first message triggered under different triggering mechanisms.
Fig. 5 is a schematic entity structure diagram of an electronic device according to an embodiment of the present invention, and as shown in fig. 5, the electronic device may include: a processor (processor) 510, a communication Interface (Communications Interface) 520, a memory (memory) 530 and a communication bus 540, wherein the processor 510, the communication Interface 520 and the memory 530 communicate with each other via the communication bus 540. Processor 510 may invoke logic instructions in memory 530 to perform a vehicle-end computing resource based message processing method comprising: processing vehicle data in real time to obtain a first message; and determining the distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first message according to the distribution interval to obtain a second message.
Furthermore, the logic instructions in the memory 530 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention or a part thereof which substantially contributes to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a non-transitory computer readable storage medium, on which a computer program is stored, where the computer program is implemented by a processor to execute the message processing method based on vehicle-end computing resources provided by the foregoing methods, and the message processing method based on vehicle-end computing resources includes: processing vehicle data in real time to obtain a first message; and determining the distribution interval of the first message according to the current vehicle-end computing resource and the number of the first messages, and distributing and processing the first message according to the distribution interval to obtain a second message.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on the understanding, the above technical solutions substantially or otherwise contributing to the prior art may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method of various embodiments or some parts of embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (17)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211059185.4A CN115567600B (en) | 2022-08-31 | 2022-08-31 | Message processing system and method based on vehicle-side computing resources |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211059185.4A CN115567600B (en) | 2022-08-31 | 2022-08-31 | Message processing system and method based on vehicle-side computing resources |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115567600A true CN115567600A (en) | 2023-01-03 |
| CN115567600B CN115567600B (en) | 2025-10-14 |
Family
ID=84739399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211059185.4A Active CN115567600B (en) | 2022-08-31 | 2022-08-31 | Message processing system and method based on vehicle-side computing resources |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115567600B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101470012A (en) * | 2007-12-29 | 2009-07-01 | 厦门雅迅网络股份有限公司 | Method for real-time vehicle driving mileage statistics based on wireless network and GPS position information |
| CN110460534A (en) * | 2019-07-26 | 2019-11-15 | 腾讯云计算（北京）有限责任公司 | A request message reporting method, device, equipment and storage medium |
| DE102019210559A1 (en) * | 2019-04-15 | 2020-10-15 | Technische Universität Dresden | Method for controlling a convoy by means of vehicle-to-vehicle communication |
| CN112003949A (en) * | 2020-09-01 | 2020-11-27 | 中国银行股份有限公司 | JMS message process distribution control system and method under WAS cluster |
| CN112506808A (en) * | 2021-02-08 | 2021-03-16 | 南京吉拉福网络科技有限公司 | Test task execution method, computing device, computing system and storage medium |
| CN113411371A (en) * | 2020-11-12 | 2021-09-17 | 腾讯科技（深圳）有限公司 | Communication method and device applied to vehicle, electronic equipment and vehicle |
-
2022
- 2022-08-31 CN CN202211059185.4A patent/CN115567600B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101470012A (en) * | 2007-12-29 | 2009-07-01 | 厦门雅迅网络股份有限公司 | Method for real-time vehicle driving mileage statistics based on wireless network and GPS position information |
| DE102019210559A1 (en) * | 2019-04-15 | 2020-10-15 | Technische Universität Dresden | Method for controlling a convoy by means of vehicle-to-vehicle communication |
| CN110460534A (en) * | 2019-07-26 | 2019-11-15 | 腾讯云计算（北京）有限责任公司 | A request message reporting method, device, equipment and storage medium |
| CN112003949A (en) * | 2020-09-01 | 2020-11-27 | 中国银行股份有限公司 | JMS message process distribution control system and method under WAS cluster |
| CN113411371A (en) * | 2020-11-12 | 2021-09-17 | 腾讯科技（深圳）有限公司 | Communication method and device applied to vehicle, electronic equipment and vehicle |
| CN112506808A (en) * | 2021-02-08 | 2021-03-16 | 南京吉拉福网络科技有限公司 | Test task execution method, computing device, computing system and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 赵海涛;朱银阳;丁仪;朱洪波;: "车联网中基于移动边缘计算的内容感知分类卸载算法研究", 电子与信息学报, no. 01, 15 January 2020 (2020-01-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115567600B (en) | 2025-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7869354B2 (en) | Analyzing and generating network traffic using an improved Markov Modulated Poisson Process model with one bursty state and a plurality of idle states | |
| US11140089B2 (en) | Data writing method, system, device and computer-readable storage medium | |
| KR100577659B1 (en) | Sender to receiver request retry method and apparatus | |
| CN114844843A (en) | Method and device for adjusting number of application instances | |
| CN115671716B (en) | Processing method, device, storage medium and electronic equipment for preloading instance application | |
| CN115361705B (en) | A NOMA network task processing method and system with information timeliness guarantee | |
| CN111183436A (en) | Information processing device, control method, and program | |
| CN115567600A (en) | Message processing system and method based on vehicle-end computing resources | |
| US20250023806A1 (en) | Communication device, round-trip time estimation method, and round-trip time estimation program | |
| EP3879794A1 (en) | Execution control method, information processing apparatus, and execution control program | |
| TWI736975B (en) | Method for adjusting packet length, mobile device and computer readable storage medium | |
| US20250023827A1 (en) | Communication device, line delay time estimation method, and line delay time estimation program | |
| CN111177194B (en) | A streaming data caching method and device | |
| CN112540836A (en) | Service scheduling management method and system | |
| CN113612624B (en) | Method and device for processing heartbeat between nodes | |
| CN112866130B (en) | Message packet interval adjusting method and device | |
| CN117956429A (en) | A communication method, device and electronic device for autonomous driving vehicle | |
| WO2024109602A1 (en) | Link determination method, apparatus, electronic device, and storage medium | |
| CN118095431A (en) | Model reasoning method, device, electronic device and storage medium | |
| JP2001352345A (en) | Data transmission device and data transmission control method | |
| CN113204411A (en) | Data processing method, intermediate processing equipment and storage medium | |
| CN112887219A (en) | Message packet interval adjusting method and device | |
| KR102884987B1 (en) | Apparatus, method and system for auto-scaling dynamic pods in a serverless computing environment | |
| CN118802542B (en) | Information distribution adjustment methods, devices, equipment, storage media and program products | |
| CN119728718B (en) | Data transmission methods, apparatus, equipment, storage media and software products |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |