WO2024009499A1

WO2024009499A1 - Vehicle-mounted task division system and vehicle-mounted task division method

Info

Publication number: WO2024009499A1
Application number: PCT/JP2022/027107
Authority: WO
Inventors: 宏樹前濱; 浩朗伊藤; 剛山田; 功治前田
Original assignee: 日立Astemo株式会社
Priority date: 2022-07-08
Filing date: 2022-07-08
Publication date: 2024-01-11

Abstract

This vehicle-mounted task division system comprises: a task information holding unit that holds information about a plurality of tasks to be executed by a plurality of computing units provided in a vehicle; an important task extracting unit that extracts an important task from the plurality of tasks held in the task information holding unit; and a task division unit. The task division unit confirms, with regard to the important task extracted by the important task extracting unit, whether there is another important task that has a complementary relationship with the important task, and divides the important task and the other important task having the complementary relationship so that the tasks are computed with different computing units.

Description

In-vehicle task division system and in-vehicle task division method

The present invention relates to an on-vehicle task division system and an on-vehicle task division method that divide tasks executed by an electronic control unit mounted on a vehicle.

Vehicles include electronic control devices that are connected to external recognition sensors and recognize vehicles and pedestrians, as well as electronic control devices that integrate recognition results obtained from multiple electronic control devices and perform calculations for autonomous driving and driving support. Many electronic control devices are installed, such as a control device and an electronic control device that controls the engine and actuators. Furthermore, in recent years, functions have been consolidated into one electronic control device in order to reduce the number of electronic control devices installed. However, an electronic control device having only one calculation component has a problem in that the calculation component becomes expensive in order to cope with the increasing processing load.

Functions (tasks) are distributed and arranged using multiple electronic control devices. Examples of such conventional techniques include those described in Patent Document 1 and Patent Document 2, for example.

Patent Document 1 describes a core allocation device that allocates tasks to be executed by multiple types of cores included in a processor. Patent Document 1 discloses a feature acquisition unit that obtains feature information indicating the characteristics of a task from design information that defines the task, and a feature acquisition unit that allocates tasks to cores based on the feature information and the configuration of the cores in the processor. A technique having an allocation unit is described.

Furthermore, Patent Document 2 describes a plurality of servers each having a database storing information of different contents necessary for executing a business process, and a business process that executes a business process by accessing each database held by the server. A technique for maintaining and managing a database in a client-server system having a client and a server is described. In addition, a database monitoring device that extracts the information necessary for continuous execution of business processing from among the information stored in the database held by each server, and creates a standalone degraded operation database on the business processing client based on the extracted information. have. Furthermore, a technique is described in which, when a business processing client cannot access at least one server, the business processing client continues to execute business processing by switching the access destination to a stand-alone degraded operation database.

JP2021-039666A Japanese Patent Application Publication No. 2000-181770

However, the technology described in Patent Document 1 does not take into account safety when any one of the plurality of calculation units fails. Further, Patent Document 2 discloses a database maintenance management system that generates a database for degenerate operation in order to continue processing even when at least one server goes down.

However, in the technique described in Patent Document 2, it is necessary to prepare a database for generating a database for degenerate operation on a separate server. Therefore, it is difficult to apply the technique described in Patent Document 2 to a vehicle-mounted electronic control device that has severe resource constraints.

An object of the present invention is to take the above-mentioned problems into consideration and to ensure that safety is not lost even if at least one computing unit fails or malfunctions when assigning multiple tasks to multiple computing units. The purpose of the present invention is to provide an in-vehicle task division system that can divide tasks into multiple tasks.

In order to solve the above problems and achieve the purpose of the present invention, an in-vehicle task division system includes a task information holding section that holds information on a plurality of tasks executed by a plurality of calculation sections provided in a vehicle, and a task information holding section The present invention includes an important task extraction section that extracts important tasks from a plurality of tasks held in the section, and a task division section. The task dividing unit checks whether there is another important task that has a mutually complementary relationship with the important task extracted by the important task extraction unit, and performs different operations on the important task and another important task that has a complementary relationship. Divide it so that it is calculated in parts.

Furthermore, the in-vehicle task division method includes the processes shown in (1) to (3) below.
(1) A process of acquiring information on multiple tasks executed by multiple calculation units provided in a vehicle.
(2) Processing of extracting important tasks from the acquired multiple tasks.
(3) Check whether there is another important task that has a complementary relationship with the extracted important task, and make sure that the important task and another important task that has a complementary relationship are calculated in different calculation units. The process of dividing.

According to the in-vehicle task division system and the in-vehicle task division method configured as described above, when assigning multiple tasks to multiple computing units, safety is not lost even if at least one computing unit malfunctions or malfunctions. Tasks can be divided so that there are no problems.

FIG. 1 is a block diagram showing an in-vehicle task division system according to a first embodiment. 7 is a flowchart showing task division processing in the in-vehicle task division system according to the first embodiment. It is a figure which shows the example of division of the recognition result integration processing task performed by an automatic driving system. FIG. 2 is a diagram illustrating an example of dividing a recognition result integration processing task performed by the automatic driving system using the in-vehicle task dividing system according to the first embodiment. 7 is a flowchart illustrating task division processing in the in-vehicle task division system according to the second embodiment. FIG. 7 is a diagram showing a flow of a specific example of importance extraction and task allocation using priorities in the in-vehicle task division system according to the second embodiment. FIG. 7 is a block diagram showing an in-vehicle task division system and a vehicle system according to a third embodiment. 12 is a flowchart showing task division processing in an in-vehicle task division system and a vehicle system according to a third embodiment. FIG. 7 is a block diagram showing the flow of processing within a vehicle in an in-vehicle task division system and a vehicle system according to a third embodiment. FIG. 7 is a block diagram showing an in-vehicle task division system according to a fourth embodiment. 12 is a flowchart showing task division processing in the on-vehicle task division system according to the fourth embodiment. FIG. 12 is a block diagram showing the flow of processing within a vehicle in an in-vehicle task division system and a vehicle system according to a fourth embodiment. FIG. 7 is a block diagram showing an in-vehicle task division system according to a fifth embodiment. 13 is a flowchart showing task division processing in the on-vehicle task division system according to the fifth embodiment.

Hereinafter, embodiments of an in-vehicle task division system and an in-vehicle task division method will be described with reference to FIGS. 1 to 14. Note that common members in each figure are given the same reference numerals.

1. First embodiment example 1-1. Configuration of an on-vehicle task division system First, the configuration of an on-vehicle task division system according to a first embodiment (hereinafter referred to as "this example") will be described with reference to FIGS. 1 to 4.
FIG. 1 is a block diagram showing an in-vehicle task division system.

As shown in FIG. 1, the in-vehicle task division system 100 includes a task division device 101 and a calculation unit assignment device 102. Furthermore, the in-vehicle task division system 100 is connected to an in-vehicle system (ECU) 103 mounted on the vehicle so as to be able to transmit information. The vehicle system 103 includes a plurality of on-board calculation units (hereinafter simply referred to as calculation units) 121, 122, and 123.

Note that in this example, an example has been described in which the vehicle system 103 is provided with

arithmetic units

121, 122, and 123, which are a plurality of in-vehicle electronic control devices (automatic driving electronic control devices), but the present invention is not limited to this. The vehicle system 103 includes, for example, multiple electronic control units, multiple chips in one in-vehicle electronic control unit (e.g. System on a Chip, hereinafter referred to as SoC), or multiple CPU cores of a SoC in an in-vehicle electronic control unit. or an arithmetic unit such as an AI accelerator.

Further, the task dividing device 101 includes a task information holding section 104, an important task extracting section 105, and a task dividing section 106.

The task information holding unit 104 stores all the task information and task attribute information that the plurality of

calculation units

121, 122, and 123 of the vehicle system 103 are in charge of calculation. As for task attribute information, for example, if the task is related to automated driving, which phase of the task from external world recognition to vehicle control, and if it is a function of the vehicle's external world recognition, which phase is the task, such as forward recognition or backward recognition, etc. be. The task information holding unit 104 outputs task information to the important task extraction unit 105.

The important task extraction unit 105 acquires a plurality of pieces of task information stored in the task information storage unit 104 and their attribute information. Then, the important task extraction unit 105 extracts important tasks that may cause a serious accident if they fail, from the plurality of task information and attribute information. Further, the important task extraction unit 105 outputs the extracted important task information to the task division unit 106.

The task division unit 106 checks whether there is another important task that is complementary to the important task extracted by the important task extraction unit 105. Then, the task division unit 106 assigns complementary relationship attribute information to the important task that is determined to have a complementary relationship with another important task (complementary important task). Here, the task division unit 106 assigns attribute information of a complementary relationship so that important tasks that complement each other can be assigned to different calculation units in the calculation unit allocation device 102, which will be described later.

Further, the task dividing unit 106 divides an important task for which it is determined that there is no other important task in a complementary relationship (an important task without complementation) into two or more tasks that complement each other, and divides the important task into two or more tasks that complement each other. Assign attribute information. An example of an important task without complementation is a task such as white line recognition in an automatic driving system. Then, the task dividing unit 106 divides the white line recognition task into right side white line recognition and left side white line recognition, if the automatic driving system can recognize the white line on either the right or left side of the vehicle and control the vehicle along the white line. do.

The calculation unit allocation device 102 refers to the complementary relationship attribute information assigned by the task division device 101 to the tasks to be handled by the

calculation units

121, 122, and 123 in the vehicle system 103. Then, the calculation unit allocation device 102 allocates tasks so that tasks that are complementary to each other are not calculated by the

same calculation units

121, 122, and 123. Then, the calculation unit allocation device 102 outputs the allocated task to the vehicle system 103.

1-2. Task Division Processing Next, task division processing in the in-vehicle task division system 100 having the above-described configuration will be described with reference to FIG. 2.
FIG. 2 is a flowchart showing task division processing.

As shown in FIG. 2, the in-vehicle task division system 100 acquires all tasks executed by the vehicle system 103 and stores them in the task information holding unit 104 (step S11). At this time, the task information holding unit 104 gives the task attribute information for extracting important tasks.

For example, the process from external world recognition to vehicle control is divided into four phases: recognition, cognition (integration), judgment, and control, and the phase in which the task is included is given as task attribute information. Further, for a task that uses an external world recognition sensor, information on which direction the external world recognition sensor is used and information on what it is recognizing is added as attribute information. Furthermore, for tasks related to actuators, information on which actuator, such as brake or suspension, is used is given as attribute information.

Next, the task information holding unit 104 outputs the stored task information and attribute information to the important task extraction unit 105. Then, the important task extraction unit 105 extracts important tasks by referring to the attribute information of each task (step S12). For example, in an automated driving system, a serious accident can be avoided by recognizing an object that may cause a collision and stopping or avoiding it. In other words, it is necessary to decide which function to use and how to generate a trajectory based on the integration of recognition results from other sensors and past recognition results, as in the recognition phase, or on the integration results of recognition information, as in the judgment phase. Serious accidents can sometimes be avoided even without such calculations.

Therefore, tasks in the recognition phase and control phase are extracted as more important tasks than in the recognition phase and judgment phase. In addition, even if the task is in the same recognition phase, if only the front can be recognized, it is possible to avoid a rear-end collision with the vehicle in front or contact with a pedestrian crossing the street on a straight road, then the recognition function using the front sensor is It is extracted as a more important task than recognition using front and rear sensors.

In this way, the important task extraction unit 105 performs pedestrian recognition using the front camera, white line recognition using the front camera, vehicle recognition using the front camera, pedestrian recognition using the front Radar, vehicle recognition using the front Radar, vehicle deceleration control, and vehicle recognition. Steering control is extracted as an important task.

Next, the important task extraction unit 105 outputs the extracted important tasks to the task division unit 106. Then, the task dividing unit 106 refers to the attribute information of the extracted important tasks and determines whether there are any tasks that are complementary to each other (step S13). For example, tasks in which pedestrian recognition by a front camera and pedestrian recognition by front Radar, vehicle recognition by a front camera and vehicle recognition by front Radar, vehicle deceleration control and vehicle steering control are complementary to each other (important tasks with complementation). It is judged that.

In the process of step S13, if it is determined that there is a task with a complementary relationship (Yes determination in step S13), the task dividing unit 106 divides the important task with complementation into a task that has a complementary relationship with the important task with complementation. Complementary relationship attribute information is given to both tasks (step S14). For example, information such as "vehicle recognition by the front camera is in a complementary relationship" is given to vehicle recognition by the front camera, and information that "vehicle recognition by the front camera is complementary" is given to vehicle recognition by the front Radar. Note that here, a task may be expressed using another expression such as an ID.

Further, in the process of step S13, if it is determined that there is no task that has a complementary relationship (No determination in step S13), the task division unit 106 divides the non-complementary important task into multiple tasks that have a complementary relationship with each other. Divide (step S15). In the process of step S15, the task division unit 106 divides the non-complemented important task with a certain overlapping area so that it can be computed in a complementary manner by

different computing units

121, 122, and 123, respectively.

Next, the task dividing unit 106 determines whether all important tasks have complementary attribute information (step S16). In the process of step S16, if it is determined that all important tasks have it (Yes determination in step S16), the task division process ends.

Further, if it is determined that there is an important task that does not have complementary attribute information (No determination in step S16), the task dividing unit 106 determines whether all important tasks have undergone the determination in step S13 ( Step S17). If it is determined in the process of step S17 that there is an important task that has not been performed, the task dividing unit 106 returns to the process of step S13.

Furthermore, if it is determined that the determination in step S13 has been performed for all important tasks (Yes determination in step S16), there are important tasks that cannot be divided into multiple tasks in the process of step S15. In this case, the task dividing unit 106 assigns attribute information to the non-complement important task indicating that it cannot be divided (step S18). This completes the task division process.

The important tasks divided by the above-described processing are output to each of the

calculation units

121, 122, and 123 of the vehicle system 103 by the calculation unit allocation device 102.

In this way, the task dividing unit 106 ensures that the important task extracted by the important task extraction unit 105 is always executed in parallel with another task that has a mutually complementary relationship. This makes it possible to perform a degenerate operation when an abnormality occurs in one calculation unit without adding a new calculation unit to the vehicle. For example, even if an abnormality occurs in one of the plurality of

calculation units

121, 122, 123, the camera or Radar recognizes the vehicle ahead and the lane on the right or left side of the vehicle. As a result, degenerate operation that performs at least emergency braking or emergency steering to avoid a collision when there is a risk of collision becomes possible without adding a new on-vehicle electronic control device.

1-3. Example of division of recognition result integration processing task Next, an example of division of recognition result integration processing task performed by the automatic driving system will be described with reference to FIGS. 3 and 4.
FIG. 3 is a diagram illustrating an example of division of recognition result integration processing tasks. FIG. 4 is a diagram showing an example of dividing the recognition result integration processing task performed by the automatic driving system using the in-vehicle task division system of this example.

As shown in FIG. 3, the vehicle M1 is equipped with a plurality of

calculation units

121, 122, 123, and 124 in order to recognize objects in all directions of the vehicle M1. Then, if the important task can be processed in parallel by spatially dividing it, the task dividing unit 106 divides the important task into a plurality of tasks that have overlapping areas and are complementary to each other. For example, the first calculation unit 121 is in charge of the right front area F1, and the second calculation unit 122 is in charge of the right rear area F2. The third calculation unit 123 is in charge of the left front area F3, and the fourth calculation unit 124 is in charge of the left rear area F4.

At this time, as shown in FIG. 4, even if either the first calculation unit 121 in charge of the right front area F1 or the calculation unit 123 in charge of the left front area F3 fails, the task division unit 106 Divide so that the integrated processing for the front of the vehicle is always performed. That is, the task dividing unit 106 divides the area F10 and the integrated task so that the right front area F1 and the left front area F3 overlap in front of the vehicle.

Note that it is not limited to integration tasks, but can be applied to any task related to space. For example, it can be applied to the task of extracting dangerous areas from the integrated vehicle surroundings after integrating external world information, or the task of generating multiple vehicle trajectories taking into account the extracted dangerous areas. It is possible.

In this way, the in-vehicle task division system 100 of this example divides important tasks that cannot be divided into a plurality of important tasks that have a mutually complementary relationship. Therefore, although the total computational load may increase, safety can be ensured with a smaller increase in computational load than adding completely redundant safety tasks.

2. Second Embodiment Next, a task division system according to a second embodiment will be described with reference to FIGS. 5 and 6.
FIG. 5 is a flowchart showing task division processing in the in-vehicle task division system. FIG. 3 is a diagram illustrating a flow of a specific example of importance extraction and task allocation using priorities in the in-vehicle task division system.

As shown in FIG. 5, the in-vehicle task division system 100 acquires all tasks to be executed by the vehicle system 103, and stores them in the task information holding unit 104 (step S31). At this time, the task information holding unit 104 gives the task attribute information for extracting important tasks.

Furthermore, the task information holding unit 104 sets the priority of important tasks by referring to the attribute information of each task (step S32). For example, a priority score is calculated for each category of task attribute information, and a weighted sum of the priority scores for each task is calculated. Or compare and prioritize important tasks.

For example, as tasks related to automatic driving, tasks in the recognition phase and control phase are given a priority score of 3 points, tasks in the judgment phase are given 2 points as a priority score, and tasks in the recognition phase are given 1 point as a priority score. Further, a process of assigning 3 points as a priority score to tasks related to the front of the vehicle and 1 point as a priority score to tasks related to other directions is performed based on the attribute information of the tasks.

Then, the important task extraction unit 105 extracts important tasks from the task priorities (step S33). For example, the important task extraction unit 105 extracts tasks for which the weighted sum of the priority scores of each task is equal to or greater than a predetermined threshold as important tasks. Alternatively, the important task extraction unit 105 extracts tasks higher than a certain priority level as important tasks from the tasks arranged in priority order.

Although priority is scored, it is not limited to priority, and may be other indicators related to safety, such as the ease of occurrence of an abnormality, controllability in the event of an abnormality, or fatality of an accident in the event of an abnormality. They may also be combined.

Next, the important task extraction unit 105 outputs the extracted important tasks to the task division unit 106. The task dividing unit 106 then performs the processes from step S34 to step S39. Note that the processing from step S34 to step S39 is the same as the processing from step S13 to step S18 of the in-vehicle task division processing according to the first embodiment, so a description thereof will be omitted.

A specific example of priority setting, importance extraction, and task allocation will be described with reference to FIG. 6. In the example shown in FIG. 6, functions of recognition, recognition, and judgment phases are assigned to a plurality of computing units as automatic driving functions.

As shown in FIG. 6, the task information holding unit 104 stores information on "recognition", "recognition", and "judgment" for each task. Regarding the priority level of the task, "recognition" > "judgment" > "recognition", "recognition by camera" > "recognition by Radar", "highly urgent judgment (app)" > "highly urgent" "Low judgment (app)", "Braking-related judgment" > "Steering-related judgment", "Front" > "Right front" > "Left front" > "Right side" > "Left side" > "Right rear" > A rule such as "left rear" > "rear" is established.

The above rules ensure that even if the vehicle is in the worst condition, it can monitor what's ahead and use the emergency brake to stop or use emergency steering to avoid something.

When priorities are assigned according to the above-mentioned rules, the priorities are "front camera recognition" > "front Radar recognition" > "AEB (Automatic Emergency Braking)" > "AES (Automatic Emergency Steering)"...

At this time, the important task extraction unit 105 extracts priority tasks of priority 4 or lower (1st to 4th) as important tasks. Then, the task division unit 106 divides the extracted important tasks and assigns them to the

calculation units

121, 122, and 123.

With regard to allocation of calculation units, for example, when allocating important tasks to two calculation units, the calculation unit allocation device 102 allocates the important tasks alternately because important tasks with similar attributes have similar priority values. As a result, two task lists are created in which important tasks that complement each other to some extent are executed in parallel. Then, the calculation unit allocation device 102 refers to the attribute information and replaces important tasks as necessary. As a result, two task lists in which important tasks that complement each other are executed in parallel can be obtained.

As described above, according to the in-vehicle task division system according to the second embodiment, by using the priority level, priorities, or another safety-related index in the important task extraction section, failures cannot be prevented. Task standards are clarified. As a result, more reliable security can be ensured and tasks can be divided efficiently.

3. Third Embodiment Next, an in-vehicle task division system and a vehicle system according to a third embodiment will be described with reference to FIGS. 7 to 9.
FIG. 7 is a block diagram showing an in-vehicle task division system and a vehicle system according to the third embodiment, and FIG. 8 is a flowchart showing task division processing in the in-vehicle task division system and vehicle system. FIG. 9 is a block diagram showing the flow of processing within the vehicle in the in-vehicle task division system and the vehicle system.

As shown in FIG. 7, the in-vehicle task division system 400 includes a task division device 401 and a calculation unit assignment device 402. The task division device 401 includes a driving situation storage section 404, a task information storage section 405, an important task extraction section 406, and a task division section 407.

The driving situation storage unit 404 stores information assuming the driving situation of the vehicle in a use case definition or a virtual simulation environment. The driving environment includes, for example, road type information such as a general road, expressway, or parking lot, driving path structure information such as a straight road or intersection, weather information such as dense fog or rain, presence or absence of nearby vehicles, and road surface condition information such as puddles and unevenness.

In this embodiment, an example has been described in which the driving situation storage unit 404 is implemented in the in-vehicle task division system 400, but the present invention is not limited to this. For example, the driving situation storage unit 404 may be implemented in a device different from the in-vehicle task division system 400, and the in-vehicle task division system 400 may receive information on the driving situation from an external device.

Next, task division processing will be described with reference to FIG. 8.
As shown in FIG. 8, the in-vehicle task division system 400 acquires all tasks to be executed by the vehicle system 403, and stores them in the task information holding unit 405 (step S41). Next, the use case and virtual simulation data in which the vehicle's driving situation is described are stored in the driving situation storage unit 404 (step S42). Here, the driving situation may be information about the driving environment of the vehicle, the traveling direction of the vehicle, or both.

Next, the important task extraction unit 406 extracts important tasks by referring to the attribute information of each task and the driving situation information stored in the driving situation storage unit 404 (step S43). For example, when referring to the use case of overtaking a low-speed vehicle on a straight road on a highway, in addition to recognizing the vehicle in front, recognizing the vehicle in the overtaking lane (in the Japanese environment, the right rear) is also extracted as an important task. be done.

Next, the task division unit 407 performs the processes from step S44 to step S49. Note that the processing from step S44 to step S49 is the same as the processing from step S13 to step S18 of the in-vehicle task division processing according to the first embodiment, so a description thereof will be omitted.

When the process of step S47 or the process of step S49 is completed, the calculation unit allocation device 402 refers to the attribute information of each task and the calculation processing capacity of the plurality of

calculation units

121, 122, 123 of the vehicle system, and assigns each task A task list for allocating the information to the plurality of

calculation units

121, 122, and 123 is created (step S50).

Next, the calculation unit allocation device 402 determines whether a task list has been created for all driving situations (step S51). In the process of step S51, if it is determined that there is a driving situation for which a task list has not been created (No determination in step S51), the in-vehicle task division system 400 returns to the process of step S43. Further, in the process of step S51, if it is determined that task lists have been created for all driving situations (Yes determination in step S51), the task division process ends.

As a more specific example of the relationship between the driving situation and task attributes, the direction in which the vehicle is traveling affects the "direction" of the task. When the vehicle moves forward, the importance of tasks related to the front increases, and when the vehicle changes lanes to the right, the importance of tasks related to the right side increases. In addition, regarding the driving environment of the vehicle, first of all, regarding the road type, since there are highways that do not require recognition, this will affect the "target" of the recognition task. For example, recognition of bicycles and crosswalks increases in importance on general roads, but decreases in importance on expressways.

Furthermore, in a parking lot, the importance of recognizing lanes (or the priority according to the second embodiment) decreases, but the importance of recognizing pedestrians on the side or rear and recognition of parking frame lines decreases. increases. Next, regarding road structures such as straight roads and intersections, the ``direction'' of the task and the ``type of actuator to be controlled'' have an impact. For example, at an intersection, the importance of recognizing left and right vehicles increases even if the vehicle is going straight.

In addition, in situations where it is not possible to avoid left or right (such as narrow straight roads or single-lane roads), the task of braking to a stop becomes more important than steering avoidance. Weather information such as dense fog or rainy weather is affected by the task's ``type of sensor used for recognition'' and ``type of actuator to be controlled.''

There are two types of external world recognition sensors used in autonomous driving: those that are easily affected by rain and snow, and those that are relatively less affected. Therefore, during rainy weather, the importance of recognition tasks using sensors that are less affected by rain increases. Furthermore, in a situation where the vehicle is being blown by strong winds, controlling the left and right movement of the vehicle using the steering wheel becomes more important than when there is no wind.

Road surface condition information such as puddles and unevenness is affected by the task's ``type of sensor used for recognition,'' ``recognition target,'' and ``type of actuator to be controlled.'' For example, puddles, snow accumulation, etc. change the reflectance of the road surface, which may impair the accuracy of some external world recognition sensors. Therefore, the importance of recognition tasks for sensors that are less affected by changes in reflectance increases. Furthermore, on a road surface where white lines cannot be recognized due to snow and it is necessary to recognize utility poles and road edges, the importance of the road edge recognition task increases and the importance of the white line recognition task decreases.

Furthermore, in order to avoid accidents due to slipping, it is necessary to appropriately select whether to use the brakes to stop, steer to avoid, or perform emergency operations depending on changes in the friction coefficient of the road surface. Therefore, the importance of "braking judgment," "emergency braking judgment," "steering judgment," and "emergency steering judgment" increases or decreases depending on the friction coefficient of the road surface. The relationship between the driving situation and the importance of the task is one example, and relationships other than the relationship may be used. Further, the importance may be determined by using the priority order or priority score in the second embodiment described above, or by using an index related to safety.

By extracting important tasks according to driving conditions, it is possible to divide tasks more safely and efficiently. For example, in the first embodiment, the task of recognizing a vehicle in front is extracted as an important task, but the task may include the vehicle moving backward or making a large turn to the left or right. Therefore, while reversing or changing direction to the left or right, the rear vehicle recognition task and the left and right vehicle recognition task are more important than the front recognition task. By using the vehicle traveling direction information, it is possible to extract the front vehicle recognition task when the vehicle is moving forward, and the rear vehicle recognition task when the vehicle is reversing, as important tasks.

The important tasks to be extracted also change depending on the driving environment information. For example, on a highway, the task of recognizing a crosswalk is not an important task, but on a general road, recognizing a crosswalk is an important task. By using the driving situation of the vehicle as described above, it is possible to prevent tasks that are important in that driving situation from being omitted from being extracted, and it is possible to prevent tasks that are not important in that situation from being extracted as important tasks.

FIG. 9 is a block diagram showing the flow of processing within the vehicle in the in-vehicle task division system and the vehicle system.
As shown in FIG. 9, each of the

calculation units

121, 122, 123, . Further, each

calculation unit

121, 122, 123, . . . 12n is connected to a task list storage unit 36 created by the calculation unit allocation device 402. Furthermore, each

calculation unit

121, 1, 122, 123, . . . 12n is connected to an external world recognition sensor 90.

Then, the driving situation recognition unit 31 recognizes the vehicle driving situation based on the information from the external world recognition sensor 90. The driving situation recognition unit 31 outputs the recognized driving situation information to the task list calling unit 32. Then, the task list calling section 32 calls a predetermined task list from the task list storage section based on the driving situation information. The called task list is input to the task execution unit 33. The task execution unit 33 then executes the task based on the input task list.

4. Fourth Embodiment Next, an on-vehicle task division system and a vehicle system according to a fourth embodiment will be described with reference to FIGS. 10 to 12.
FIG. 10 is a block diagram of the in-vehicle task division system according to the fourth embodiment, and FIG. 11 is a flowchart showing task division processing in the in-vehicle task division system according to the fourth embodiment.

As shown in FIG. 10, the in-vehicle task division system 500 includes a task division device 501 and a calculation unit assignment device 502. The task dividing device 501 includes a driving situation determining section 504, a task information holding section 505, an important task extracting section 506, and a task dividing section 507.

The driving situation determination unit 504 is connected to an external world recognition sensor mounted on the vehicle. Then, the driving situation determining unit 504 calculates the driving situation from the vehicle positioning information, the recognition result of the external world recognition sensor, the map information, and the vehicle motion information such as the steering angle and the release of the accelerator.

Further, as shown in FIG. 12, the task division device 501 and the calculation unit allocation device 502 are provided in the first calculation unit 121 in the vehicle system 503. That is, in the fourth embodiment, the first calculation unit 121, as a manager, determines the driving situation, extracts important tasks, and assigns the tasks to itself (the first calculation unit 121) and

other calculation units

122 and 123. Make assignments. However, there may be a plurality of managers, or they may be implemented in a different arithmetic unit from the arithmetic unit to which processing is assigned.

Next, task division processing will be described with reference to FIG. 11.
As shown in FIG. 11, the in-vehicle task division system 500 acquires all tasks to be executed by the vehicle system 503, and stores them in the task information holding unit 505 (step S61). Next, the driving situation determining unit 504 calculates (recognizes) the driving situation from the recognition results of the vehicle's positioning sensor and external world recognition sensor, map information, and vehicle movement information such as steering angle and accelerator release (step S62). Here, the driving situation may be the driving environment of the vehicle, the traveling direction information of the vehicle, or both.

Next, the important task extraction unit 506 extracts important tasks by referring to the attribute information of each task and the driving situation information determined by the driving situation determining unit 504 (step S63). The task dividing unit 507 then performs the processes from step S64 to step S69. Note that the processing from step S64 to step S69 is the same as the processing from step S13 to step S18 of the in-vehicle task division processing according to the first embodiment, so a description thereof will be omitted.

When the process of step S67 or the process of step S69 is completed, the calculation unit allocation device 502 refers to the attribute information of each task and the calculation processing capacity of the plurality of

calculation units

121, 122, 123 of the vehicle system, and assigns each task is assigned to the plurality of

calculation units

121, 122, and 123 (step S70). This completes the task division process.

In the task division process described above, important tasks depending on the driving situation may be determined again within the vehicle system 503, or a predetermined task list may be output to the vehicle system 503.

5. Fifth Embodiment Next, an in-vehicle task division system and a vehicle system according to a fifth embodiment will be described with reference to FIGS. 13 and 14.
FIG. 13 is a block diagram of an in-vehicle task division system according to the fifth embodiment, and FIG. 14 is a flowchart showing task division processing in the in-vehicle task division system according to the fifth embodiment.

As shown in FIG. 13, the in-vehicle task division system 800 includes a task division device 801 and a calculation unit assignment device 802. The task division device 801 includes a task information holding section 804 , a task classification section 805 , and a task division section 807 .

The task classification unit 805 tags each task according to attribute information. For example, tags such as "recognition," "recognition (integration)," "judgment," and "control" are given to each task as an autonomous driving task. Additionally, tags such as recognition target "vehicle", "white line", "pedestrian", etc. are added to the "recognition" task. Tasks with directionality are given tags such as "forward", "right side", "left side", and "backward".

The task dividing unit 806 divides tasks that do not have a task attached to one tag into multiple tasks that complement each other. If the only task tagged with "recognition," "white line," and "front" is white line recognition using the front camera, multiple tasks that complement each other, such as right white line recognition using the front camera and left white line recognition using the front camera, can be created. To divide.

Note that tagging and task division may be performed only on important tasks by extracting important tasks, or may be performed on all tasks. When important tasks are extracted, it is also possible to combine this with the embodiment described above. In addition, although it was described in the form of adding tags, it is also possible to group tasks (for example, a group of tasks that recognize the white line ahead, a group of tasks that recognize the vehicle in front), and tasks of the same group to be integrated into a single calculation unit. It may be divided so that it is not implemented.

Next, task division processing will be described with reference to FIG. 14.
As shown in FIG. 14, the in-vehicle task division system 800 acquires all tasks executed by the vehicle system 803, and stores them in the task information holding unit 804 (step S81). Next, the task classification unit 805 assigns a tag to each task based on the attribute information of each task (step S82).

Next, the task dividing unit 806 refers to the tags assigned in step S82 and determines whether there are tasks that are complementary to each other, that is, tasks with the same tag (step S83). Further, in the process of step S83, if it is determined that there is no task with a complementary relationship (No determination in step S83), the task division unit 806 divides the non-complementary important task into multiple tasks that have a complementary relationship with each other. Divide (step S84). Next, the task dividing unit 806 attaches a tag indicating that it cannot be divided to the non-complemented important task (step S85). Then, the process moves to step S86.

Further, in the process of step S83, if it is determined that there are tasks that have a complementary relationship, that is, there are tasks that have the same tag (Yes determination of step S83), the task dividing unit 806 divides all tasks except for non-divisible tasks. It is determined whether a task with the same tag exists for (step S86).

In the process of step S86, if it is determined that there is a task for which the process of step S83 has not been performed (No determination in step S86), the task dividing unit 806 returns to the process of step S83. If it is determined that the process of step S83 has been performed for all tasks (Yes determination of step S86), the task division process ends.

Note that the present invention is not limited to the embodiments described above and shown in the drawings, and various modifications can be made without departing from the gist of the invention as set forth in the claims.

For example, the task list may be generated by predicting the driving situation using the in-vehicle task division system 500 according to the fourth embodiment described above, the destination and route information of the vehicle, and the map information. For example, if the route from the current location to the destination includes local roads and expressways, a task list for driving on general roads and a task list for driving on expressways are created and stored at the time of route search. Then, when the driving situation changes from an ordinary road to an expressway, the task list to be read may be changed from an ordinary road to an expressway. This eliminates the need to extract important tasks or divide tasks every time the driving situation changes, making it possible to reduce the computational load.

Additionally, in all embodiments, the important task extraction unit may extract important tasks using results learned from past driving data. For example, a database is prepared in advance that collects driving conditions during past traffic accidents. Then, it may be possible to analyze which functions in the database are essential to operate in order to avoid traffic accidents, and use this to extract important tasks in similar driving situations.

Furthermore, as shown in the fourth embodiment, the driving situation is the traveling direction and driving environment, and is a combination of information such as road type and weather. Therefore, important tasks may be extracted in unknown driving situations by learning from important task extraction results in driving situations experienced in the past. This eliminates the need to assume all possible driving conditions during design, making it possible to reduce design and verification man-hours. Furthermore, it is possible to ensure safety even in unknown driving situations caused by changes in the times and environment.

Furthermore, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Further, it is possible to add, delete, or replace a part of the configuration of each embodiment with other configurations. Further, each of the above-mentioned configurations, functions, processing units, processing means, etc. may be partially or entirely realized in hardware by designing, for example, an integrated circuit. Furthermore, each of the above configurations, functions, etc. may be realized by software by a processor interpreting and executing a program for realizing each function. Information such as programs, tables, files, etc. that implement each function can be stored in a memory, a recording device such as a hard disk, an SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

31... Driving situation recognition unit, 32... Task list calling unit, 33... Task execution unit, 36... Task list storage unit, 90... External world recognition sensor, 100, 400, 500, 800... In-vehicle task division system, 101, 401, 501, 801...Task division device, 102...Computation unit allocation device, 103...Vehicle system, 104...Task information holding unit, 105...Important task extraction unit, 106...Task division unit, 121, 122, 123...Computation unit (in-vehicle) arithmetic unit), 404...driving status storage unit, 504... driving status determining unit, 805... task classification unit

Claims

a task information holding unit that holds information on a plurality of tasks executed by a plurality of calculation units provided in the vehicle;
an important task extraction unit that extracts important tasks from the plurality of tasks held in the task information storage unit;
It is checked whether there is another important task that has a complementary relationship with the important task extracted by the important task extraction unit, and the important task and another important task that has a complementary relationship are processed in different calculation units. a task dividing unit that divides the task so as to be calculated;
An in-vehicle task division system with
According to claim 1, when the task dividing unit determines that there is no other important task that is complementary to the important task, the task dividing unit divides the important task so that each important task can be calculated by a different calculation unit. In-vehicle task division system described.
The in-vehicle task division system according to claim 2, wherein the task division section divides the important tasks determined to be non-complementary so that a certain overlapping area is provided so that each of the important tasks can be computed in a complementary manner by different computing sections 1.
comprising a driving situation determination unit that determines the driving situation of the vehicle;
The in-vehicle task division system according to claim 1, wherein the important task extraction unit extracts the important task based on the driving situation.
comprising a driving situation storage section in which the driving situation of the vehicle is stored in advance;
The in-vehicle task division system according to claim 1, wherein the important task extraction unit extracts the important task based on the driving situation.
The task information holding unit sets priorities for the plurality of tasks,
The in-vehicle task division system according to claim 1, wherein the important task extraction unit extracts the important tasks based on the priority.
The in-vehicle task division system according to claim 6, wherein the priority is set in the order of recognition of external world information, judgment based on the integrated result of recognition, and recognition that integrates the recognition results.
The in-vehicle task division system according to claim 1, further comprising a calculation unit allocation device that allocates the important task divided by the task division unit to a plurality of calculation units.
The in-vehicle task division system according to claim 1, wherein the in-vehicle task division system is installed in any one of the plurality of arithmetic units of the vehicle.
A process of acquiring information on a plurality of tasks executed by a plurality of calculation units provided in the vehicle;
a process of extracting important tasks from the plurality of acquired tasks;
Confirm whether there is another important task that has a complementary relationship with the extracted important task, and divide the important task and another important task that has a complementary relationship so that they are calculated in different calculation units. processing and
In-vehicle task division method including.