JP7250651B2 - working machine - Google Patents

Description

The present invention relates to work machines.

A working machine is equipped with a dedicated arithmetic unit to perform complex arithmetic operations such as matrix arithmetic at high speed, and the operation is controlled by performing various kinds of arithmetic processing with this arithmetic unit. Further, in recent years, the amount of information processing such as arithmetic processing tends to increase with the sophistication of the control content required for working machines, and the load on arithmetic units is increasing.

As a technique for reducing the load of such information processing, for example, Patent Document 1 discloses an in-vehicle multi-application execution device that executes a plurality of applications within a specified cycle, and dynamically predicts the processing time of each application. a multi-application control unit that schedules each application based on the processing time dynamically predicted by the dynamic processing time prediction unit; and based on the scheduling result of the multi-application control unit: A vehicle-mounted multi-application execution device is disclosed that includes a multi-application execution unit that executes processing of the plurality of applications.

JP 2011-100338 A

In the conventional technology described above, the processing time of an application is dynamically predicted, and scheduling of each application is performed based on the predicted processing time. As a result of scheduling, when it is determined that there is an application whose processing is not completed within a specified cycle, the application is aborted or the function of the application is lowered based on a preset priority. However, in the conventional technology described above, when a plurality of applications use the arithmetic unit at the same time, even high-priority applications are subject to discontinuation or functional deterioration, and the arithmetic processing of important applications is hindered. It is conceivable that

SUMMARY OF THE INVENTION It is an object of the present invention to provide a working machine capable of performing arithmetic processing by a plurality of applications without interfering with arithmetic processing of important applications.

The present application includes a plurality of means for solving the above problems . A controller for executing an operation and an operation for which a delay is allowed is provided, and the controller is different from the exclusive area for performing the priority operation processing set as the operation processing for preferentially performing the operation and the priority operation processing. an arithmetic device in which a shared area for performing non-prioritized arithmetic processing is set; and an arithmetic management device that manages execution of the preferential arithmetic processing in the exclusive region and execution of the non-prioritized arithmetic processing in the shared region. wherein, when an arithmetic processing is requested, the arithmetic management device performs an arithmetic operation relating to the attitude of the working machine, in which the requested arithmetic processing is the preferential arithmetic, and an arithmetic operation in which the delay is allowed. If it is determined that the calculation is related to the attitude of the work machine, the processing is executed as the priority calculation processing using the exclusive area of the calculation device, and the delay is allowed. If it is determined that the calculation is a non-prioritized calculation, it is executed as the non-prioritized calculation processing using the shared area left for the exclusive area of the arithmetic unit .

According to the present invention, arithmetic processing can be performed by a plurality of applications without interfering with arithmetic processing of important applications.

1 is a diagram schematically showing the appearance of a hydraulic excavator, which is an example of a working machine; FIG. FIG. 4 is a diagram schematically showing part of processing functions of a controller mounted on the hydraulic excavator; It is a figure which shows the hardware constitutions of a controller. It is a figure which shows typically the mode of each arithmetic processing of a controller. FIG. 4 is a diagram showing details of processing functions of a GPU operation management unit; 7 is a flow chart showing the processing contents of application registration to the GPU operation management unit; 5 is a flow chart showing the processing contents of application execution by a GPU operation management unit; 4 is a timing chart of execution processing in the GPU operation management unit; FIG. 11 is a flow chart showing processing contents of application registration to the GPU operation management unit according to the second embodiment; FIG. FIG. 11 is a diagram schematically showing how each arithmetic processing of the controller according to the third embodiment is performed; FIG. 11 is a timing chart of processing in a GPU usage management unit according to the third embodiment; FIG. FIG. 14 is a timing chart of processing in the GPU operation management unit according to the fourth embodiment; FIG. FIG. 14 is a flow chart showing the processing contents of a coordinate conversion application according to the fourth embodiment; FIG. FIG. 11 is a timing chart of processing in a GPU usage management unit according to the fifth embodiment; FIG.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings. In this embodiment, a hydraulic excavator will be described as an example of a working machine, but the present invention can also be applied to other working machines.

<First embodiment>
A first embodiment of the present invention will be described with reference to FIGS. 1 to 8. FIG.

FIG. 1 is a diagram schematically showing the appearance of a hydraulic excavator, which is an example of a working machine according to the present embodiment. Moreover, FIG. 2 is a diagram schematically showing a part of the processing functions of the controller mounted on the hydraulic excavator.

In FIG. 1, a hydraulic excavator 100 is a multi-joint type front working machine 1 configured by connecting a plurality of driven members (a boom 4, an arm 5, and a bucket (working tool) 6) that rotate in the vertical direction. , a lower running body 3 , and an upper revolving body (vehicle body) 2 provided so as to be able to turn relative to the lower running body 3 . The base end of the boom 4 of the front working machine 1 is supported by the front part of the upper revolving body 2 so as to be rotatable in the vertical direction, and one end of the arm 5 is located at an end (tip) different from the base end of the boom 4. A bucket 6 is supported at the other end of the arm 5 so as to be vertically rotatable. The boom 4, the arm 5, the bucket 6, the upper rotating body 2, and the lower traveling body 3 are hydraulic actuators such as a boom cylinder 4a, an arm cylinder 5a, a bucket cylinder 6a, a swing motor 2a, and left and right traveling motors 3a (however, (only one travel motor is shown).

An operator's cab 9 is provided with operation levers (operation devices) 9a and 9b for outputting operation signals for operating the hydraulic actuators 2a to 6a. Although not shown, the operating levers 9a and 9b can be tilted forward, backward, leftward, and rightward. The amount of lever operation is output to the controller 19, which is a control device, via electrical wiring. That is, the operations of the hydraulic actuators 2a to 6a are assigned to the front-rear direction or left-right direction of the operation levers 9a and 9b, respectively.

The boom cylinder 4a, the arm cylinder 5a, the bucket cylinder 6a, the swing motor 2a, and the left and right traveling motors 3a are controlled by hydraulic actuators 2a to 6a from a hydraulic pump device 7 driven by a prime mover such as an engine or an electric motor (not shown). A control valve 8 controls the direction and flow rate of the hydraulic oil supplied to . The control valve 8 is driven by a drive signal (pilot pressure) output from a pilot pump (not shown) through an electromagnetic proportional valve. The operation of each of the hydraulic actuators 2a to 6a is controlled by controlling the electromagnetic proportional valves with the controller 19 based on the operation signals from the operation levers 9a and 9b.

The operation levers 9a and 9b may be of a hydraulic pilot type, and the pilot pressure corresponding to the operation direction and operation amount of the operation levers 9a and 9b operated by the operator is supplied to the control valve 8 as a drive signal, It may be configured to drive each of the hydraulic actuators 2a to 6a.

Inertial Measurement Units (IMUs) 12, 14 to 16 are arranged as attitude sensors on the upper rotating body 2, boom 4, arm 5, and bucket 6, respectively. Hereinafter, when it is necessary to distinguish these inertial measurement devices, they are referred to as a vehicle body inertial measurement device 12, a boom inertial measurement device 14, an arm inertial measurement device 15, and a bucket inertial measurement device 16, respectively.

The inertial measurement devices 12, 14-16 measure angular velocity and acceleration. Considering the case where the upper rotating body 2 on which the inertial measurement devices 12, 14 to 16 are arranged and the driven members 4 to 6 are stationary, the IMU coordinate system set for each inertial measurement device 12, 14 to 16 is direction of gravitational acceleration (that is, vertically downward direction) and the mounting state of each inertial measurement device 12, 14 to 16 (that is, each inertial measurement device 12, 14 to 16, upper revolving body 2, and each driven member 4 to 6), the orientation of the upper rotating body 2 and the driven members 4 to 6 can be detected. Here, the inertial measurement devices 14 to 16 constitute an attitude information detection section that detects information about the attitude of each of the plurality of driven members.

Note that the posture information detection unit is not limited to the inertial measurement device, and for example, a tilt angle sensor may be used. In addition, a potentiometer is arranged at the connecting portion of each of the driven members 4 to 6 to detect the relative orientation (information about the attitude) of the upper revolving body 2 and each of the driven members 4 to 6, and each driven member is detected from the detection result. The postures of the members 4-6 may be obtained. Stroke sensors are arranged in the boom cylinder 4a, the arm cylinder 5a, and the bucket cylinder 6a, respectively. information), and the posture of each of the driven members 4 to 6 may be obtained from the result.

In FIG. 2, the controller 19 has various functions for controlling the operation of the hydraulic excavator 100. As a part of it, the controller 19 includes an attitude calculation unit 19a, a monitor display control unit 19b, a hydraulic system control unit 19c, and a construction control unit 19c. It has each functional part of the target surface calculation part 19d.

The attitude calculation unit 19a performs attitude calculation processing for calculating the attitude of the front working machine 1 based on the detection results from the inertial measurement devices 12, 14-16.

The construction target plane calculation unit 19d is based on construction information 17 such as a three-dimensional construction drawing pre-stored by the construction manager in a storage device (not shown) and the construction target plane calculated by the construction target plane calculation unit 19d. to calculate a construction target plane that defines the target shape of the construction target.

The monitor display control unit 19b controls the display of a monitor (not shown) provided in the driver's cab 9, and displays the construction target surface calculated by the construction target surface calculation unit 19d and the front surface calculated by the posture calculation unit 19a. Based on the attitude of the work machine 1, the contents of instructions for operation support to the operator are calculated and displayed on the monitor in the operator's cab 9. FIG. That is, the monitor display control unit 19b displays, for example, the attitude of the front working machine 1 having driven members such as the boom 4, the arm 5, and the bucket 6, and the tip position and angle of the bucket 6 on the monitor, thereby enabling operator's operation. It plays a part of the function as a machine guidance system that supports the

The hydraulic system control unit 19c controls the hydraulic system of the hydraulic excavator 100 including the hydraulic pump device 7, the control valve 8, and the hydraulic actuators 2a to 6a. Based on the surface and the attitude of the front work implement 1 calculated by the attitude calculation unit 19a, the operation of the front work implement 1 is calculated, and the hydraulic system of the hydraulic excavator 100 is controlled so as to realize the operation. That is, the hydraulic system control unit 19c, for example, limits the operation so that the tip of the work tool such as the bucket 6 does not come closer to the target construction surface than a certain amount, or restricts the movement of the work tool (for example, the tip of the bucket 6) to the target construction surface. It is responsible for part of the function as a machine control system that controls movement along the surface.

FIG. 3 is a diagram showing the hardware configuration of the controller.

3, the controller 19 includes an input interface 91, a central processing unit (CPU) 92 and an image processing unit (GPU) 93, which are processors, a read only memory (ROM) 94 and a random access memory (RAM) which are storage devices. ) 95 and an output interface 96 . An input interface 91 inputs signals from the inertial measurement devices 12, 14 to 16 and signals such as the construction information 17, and performs A/D conversion. The ROM 93 is a recording medium that stores a control program for executing a flowchart to be described later and various information necessary for executing the flowchart. The GPU 94 is a GPU used for performing so-called GPGPU (general-purpose computing on graphics processing units: general-purpose computation by GPU). The CPU 92 and GPU 94 perform predetermined arithmetic processing on signals received from the input interface 91 and memories 94 and 95 according to control programs stored in the ROM 93 . The output interface 96 creates a signal for output according to the calculation results of the CPU 92 and the GPU 93, and outputs the signal to a display device, an electromagnetic proportional valve that constitutes the hydraulic system, and the like, so that the hydraulic actuators 3a, 3b, 3c, and displays information related to the hydraulic excavator 100 on the display device. The controller 19 shown in FIG. 3 exemplifies a case in which semiconductor memories such as ROM 94 and RAM 95 are provided as storage devices, but any device having a storage function can be substituted. A configuration including a storage device may be employed.

FIG. 4 is a diagram schematically showing how each arithmetic processing of the controller is performed. FIG. 5 is a diagram showing the details of the processing functions of the GPU operation management unit.

In this embodiment, a case of using one GPU 93 for processing in order to execute a plurality of applications will be described as an example. Assume that GPU 93 has several properties. The GPU 93 can execute multiple applications in parallel. This is expressed as having multiple cores. If you have 4 cores, you can run up to 4 applications in parallel. It is also possible for one application to use four cores. In this case, the calculation speed is improved by approximately four times. Also, while one application is running, its operation cannot be interrupted to start another application. This property is called single task hardware. Note that the CPU 92 is multitasking hardware that has a context switch and can switch to another application during execution of one application. Hereinafter, an application may be simply referred to as an app.

FIG. 4 illustrates a case where the CPU 92 is running three applications: the attitude measurement application 102, the terrain section calculation application 103, and the coordinate conversion application 104. In FIG. Of these, the posture measurement application 102 is defined as an important application (priority application) that should not be delayed and should be executed preferentially. On the other hand, the terrain cross-section calculation application 103 and the coordinate conversion application 104 are defined as applications (non-priority applications) that allow some delay.

For example, the attitude measurement application 102 calculates the tilt angle of the upper rotating body 2 and the tilt of each part of the front work machine 1 (boom 4, arm 5, bucket 6) based on the acceleration information from the inertial measurement devices 12, 14-16. The angle can be calculated, and the vehicle body overturn angle at which the vehicle body is likely to overturn is obtained from this information. Hereinafter, the inclination angle of the upper rotating body 2 and the inclination angles of the respective parts (the boom 4, the arm 5, and the bucket 6) of the front working machine 1 are collectively referred to as the vehicle body inclination angle. In posture calculation processing by the posture measurement application 102, the vehicle body tilt angle and the vehicle body overturn angle are always calculated in each work situation, and based on these calculation results, if there is a risk of the hydraulic excavator 100 overturning, it is prevented. Issue a control command to That is, the posture measurement application 102 is an important application (priority application) that needs to be executed preferentially in order to prevent the excavator 100 from overturning.

The hydraulic excavator 100 includes a terrain information acquisition device (not shown) for acquiring terrain information around the vehicle body. The terrain information acquisition device is, for example, a stereo camera. For example, the terrain section calculation application 103 performs terrain section calculation processing for calculating the terrain section in front of the hydraulic excavator 100 based on the terrain information acquired by the terrain information acquisition device. The hydraulic excavator 100 can make a work plan based on the result of the terrain cross-section calculation process. That is, the calculation result of the terrain cross-section calculation application 103 is neither information that is required immediately nor information that needs to be calculated frequently, so the terrain cross-section calculation application 103 is a non-prioritized application.

For example, when an operation using the GPU 93 is requested, the posture measurement application 102 issues a priority application registration request 105 to the GPU operation management unit 92a. Inside the GPU 93, a GPU exclusive area 93a and a GPU shared area 93b are secured. The GPU operation management unit 92a issues a priority application execution request 107 to the GPU exclusive area 93a secured inside the GPU 93 to execute operations on the GPU 93 . Since the GPU exclusive area 93a is reserved exclusively for the orientation measurement application 102, which is a priority application, the calculation processing of the orientation measurement application 102 is not delayed by other applications.

Further, when the calculation using the GPU 93 is requested, the terrain section calculation application 103 and the coordinate conversion application 112 issue a non-prioritized application registration request 106 to the GPU calculation management unit 92a. The GPU computation management unit 92 a issues a non-prioritized application execution request 108 to the GPU shared area 93 b secured inside the GPU 93 to perform computation in the GPU 93 . For example, when the coordinate transformation application 104 requests an operation using the GPU 93 (that is, the GPU shared area 93b), the terrain section calculation application 103 may already be using the GPU 93 (the GPU shared area 93b). . In that case, the calculation of the coordinate conversion application 104 is suspended, and after the calculation of the terrain section calculation application 103 is completed and the GPU 93 (GPU shared area 93b) is released, the calculation processing of the coordinate conversion application 104 is executed. Therefore, the execution of non-priority apps may be delayed.

As shown in FIG. 5, arithmetic processing corresponding to each application is registered in advance in the GPU arithmetic management unit 92a. That is, the GPU calculation management unit 92a includes calculation 202 for the attitude measurement application corresponding to the attitude measurement application 102, calculation 203 for the terrain section calculation application corresponding to the terrain section calculation application 103, and coordinates corresponding to the coordinate conversion application 104. A calculation 204 for the conversion application is registered.

FIG. 6 is a flow chart showing the processing contents of application registration to the GPU operation management unit.

In FIG. 6, the priority application first issues a priority application registration request 105 to the GPU operation management unit 92a (step S100). This registration request contains request information as to how many cores the priority app needs to use.

Since it is necessary to allocate cores in the GPU-owned area 93a to the priority application, the GPU operation management unit 92a determines whether or not there are remaining cores that can be allocated to the GPU-owned area 93a (step S110).

If the determination result in step S110 is NO, that is, if there is no free space in the GPU-owned area 93a, a notification of no free-owned area is issued to the priority application that issued the priority application registration request 105 (step S121). , terminate the process.

Further, if the determination result in step S110 is YES, that is, if the GPU-owned area 93a has a free space, the core used for the calculation is secured in the GPU-owned area 93a, and priority application registration is executed (step S120).

In step S120, when the registration of all the priority applications that issued the priority application registration request 105 is completed, the non-priority application subsequently issues the non-priority application registration request 106 (step S130). Any number of non-prioritized applications can be registered because they use the GPU shared area 93b and their execution can be suspended. Therefore, non-prioritized application registration is immediately executed (step S140).

When the registration of all the non-prioritized applications that issued the non-prioritized application registration request 106 is completed, the determination of the GPU operation management unit is executed (step S150). This corresponds, for example, to compiling a program.

FIG. 7 is a flowchart showing the processing contents of application execution by the GPU operation management unit.

In FIG. 7, the GPU operation management unit 92a first prepares to execute an application (step S200). Specifically, this processing is activation of a thread corresponding to each application.

Subsequently, the GPU calculation management unit 92a waits for an execution request from the application (step S210).

Subsequently, when the priority application issues a priority application execution request 107 (step S220), the GPU calculation management unit 92a executes the priority application in the GPU exclusive area 93a (step S230).

Subsequently, when the non-prioritized application issues the non-prioritized application execution request 108 (step S240), the GPU operation management unit 92a executes the non-prioritized application in the GPU shared area 93b (step S250).

FIG. 8 is a timing chart of execution processing in the GPU operation management unit.

In FIG. 8 , the GPU operation management unit 92 a first activates the main thread 301 . The main thread 301 prepares for application execution (step S200 in FIG. 7), and starts a thread corresponding to each registered application. In the present embodiment, the attitude measurement application thread 302 corresponding to the orientation measurement application 102 which is the priority application, and the terrain section calculation application corresponding to the terrain section calculation application 103 and the coordinate transformation application 104 which are non-priority applications. Start the thread 303 and the thread 304 corresponding to the coordinate conversion application.

The posture measurement application support thread 302 waits for the priority application execution request 107 from the posture measurement application 102 (step S210 in FIG. 7). When the priority application execution request 107 is issued from the posture measurement application 102 (step S220 in FIG. 7), the computation 202 for the posture measurement application registered in advance is executed in the GPU exclusive area 93a (step S230 in FIG. 7). When the execution is completed, a priority application completion notification is issued to the posture measurement application 102, and the application execution standby state is entered again.

The terrain cross-section calculation application corresponding thread 303 waits for application execution immediately after being started, and waits for a non-prioritized application execution request 108 from the terrain cross-section calculation application 103 (step S210 in FIG. 7). When the terrain cross-section calculation application 103 issues a non-prioritized application execution request 108 (step S240 in FIG. 7), the previously registered terrain cross-section calculation application operation 203 is executed in the GPU shared area 93b. When the execution is completed, a non-prioritized application completion notification is issued to the terrain section calculation application 103, and the application execution standby state is entered again.

The coordinate conversion application support thread 304 waits for application execution immediately after being started, and waits for a non-prioritized application execution request 108 from the coordinate conversion application 104 (step S210 in FIG. 7). When the non-prioritized application execution request 108 is issued from the coordinate transformation application 104 (step S240 in FIG. 7), the coordinate transformation application calculation 204 registered in advance is executed in the GPU shared area 93b. Here, when the calculation 203 for the terrain section calculation application is being executed in the GPU shared area 93b, the execution of the calculation 204 for the coordinate transformation application is delayed until its completion. When the execution is completed, a non-prioritized application completion notification is issued to the coordinate conversion application 104, and the application execution standby state is entered again.

In the work machine (hydraulic excavator 100) according to the present embodiment configured as described above, the exclusive area for performing the preferential arithmetic processing which is set as the preferential arithmetic processing is different from the preferential arithmetic processing. an arithmetic device in which a shared area for performing non-prioritized arithmetic processing is set; and an arithmetic management device that manages execution of the preferential arithmetic processing in the exclusive region and execution of the non-prioritized arithmetic processing in the shared region. , wherein when the priority operation processing and the non-priority operation processing are requested, the operation management device distinguishes between the operation with priority and the operation for which delay is allowed, and the operation with priority is the above An exclusive area is used to execute the priority arithmetic processing, and an arithmetic operation for which a delay is allowed is executed as the non-prioritized arithmetic processing using the shared area left for the exclusive area. . For this reason, it becomes possible to use the arithmetic device for a plurality of applications without delaying the operation of the priority application. That is, arithmetic processing can be performed by a plurality of applications without interfering with arithmetic processing of important applications.

<Second Embodiment>
A second embodiment of the present invention will be described with reference to FIG.

This embodiment makes the size of the GPU exclusive area 93a of the GPU 93 variable. In a construction site where a work machine (for example, a hydraulic excavator 100) operates, there is an increasing number of applications (priority apps) for preventing contact with people when there is a possibility that people may be present in the work area of the work machine. , the amount of computation required also increases. On the other hand, when the possibility of contact with a person or the like is low, the execution efficiency of other applications (non-priority applications) should be increased. Therefore, in the present embodiment, the size of the GPU shared area 93b is changed by changing the size of the GPU exclusive area 93a according to the number of registered priority applications.

FIG. 9 is a flow chart showing the processing contents of application registration to the GPU operation management unit. In the figure, the same reference numerals are given to the same members as in the first embodiment, and the description thereof is omitted.

In FIG. 9, the priority application first issues a priority application registration request 105 to the GPU operation management unit 92a (step S300).

Subsequently, the GPU operation management unit 92a determines whether or not there is a free space in the GPU exclusive area 93a (step S310).

If the determination result in step S310 is NO, it is further determined whether or not there is free space in the entire GPU 93 (step S311). In this case, a no-GPU-unoccupied notification is issued to the priority application that issued the priority application registration request 105 (step S313), and the process ends.

If the determination result in step S311 is YES, that is, if there is space in the GPU shared area 93b, the GPU exclusive area 93a is expanded according to the computation amount of the priority application (that is, the GPU shared area 93b is expanded). to shrink).

If the determination result in step S310 is YES, or if the process of step S312 is completed, that is, if an area for executing the priority application is secured in the GPU exclusive area 93a, the priority application registration is executed ( step S320).

In step S320, when the registration of all the priority applications that issued the priority application registration request 105 is completed, it is then determined whether or not there is space in the GPU exclusive area 93a. The area 93a is reduced to a predetermined size (step S340).

If the determination result in step S330 is NO, or if the process of step S340 is finished, then the non-prioritized application issues the non-prioritized application registration request 106 (step S350), and immediately registers the non-prioritized application. (step S360). When the registration of all the non-prioritized applications that issued the non-prioritized application registration request 106 is completed, the determination of the GPU operation management unit is executed (step S370).

Other configurations are the same as those of the first embodiment.

The same effects as those of the first embodiment can be obtained in the present embodiment configured as described above.

Also, the GPU exclusive area 93a and the GPU shared area 93b can be optimized according to the state of the working area of the work machine.

<Third Embodiment>
A third embodiment of the present invention will be described with reference to FIGS. 10 and 11. FIG.

In this embodiment, calculations are performed by the GPU 93 not from the GPU calculation management unit 92a in the first embodiment, but from each application. In order to manage the direct use of the GPU 93 by each application, the CPU 92 in this embodiment has a GPU usage manager 92b instead of the GPU operation manager 92a.

FIG. 10 is a diagram schematically showing how each arithmetic processing of the controller is performed. Also, FIG. 11 is a timing chart of processing in the GPU usage management unit. In the figure, the same reference numerals are given to the same members as in the first embodiment, and the description thereof is omitted.

In FIG. 10, the CPU 92A is equipped with an orientation measurement application 102 as a priority application, and a terrain section calculation application 103 and a coordinate conversion application 104 as non-priority applications. The GPU 92A also has a GPU use management section 92b.

Since the posture measurement application 102 is a priority application, the GPU exclusive area 93a is allocated in advance. In other words, it is set so that calculations can be performed directly using the GPU 93 (GPU exclusive area 93a). On the other hand, since the terrain section calculation application 103 is a non-priority application, it is set so as not to directly use the GPU 93 (GPU shared area 93b). Therefore, when an operation on the GPU 93 is requested by a non-priority application, first, a usable core count request 403 is issued to the GPU usage management unit 92b. The GPU usage management unit 92b responds to this with the number of cores assigned to the GPU shared area 93b. The terrain cross-section calculation application 103 performs calculation using the GPU shared area 93b based on this response.

Since the coordinate conversion application 104 is also a non-prioritized application, it first issues a usable core number request 404 to the GPU usage management unit 92b. The GPU usage management unit 92b responds to this with the number of cores assigned to the GPU shared area 93b. At this time, it is not considered whether or not another non-prioritized application (that is, the terrain section calculation application 103) is using the GPU 93 (that is, the GPU shared area 93b). The coordinate conversion application 104 performs calculation using the GPU shared area 93b based on this response.

As shown in FIG. 11, since the orientation measurement application 102 is a priority application, the orientation measurement calculation can be executed at any time.

Since the terrain section calculation application 103 is a non-prioritized application, it first issues a request for the number of usable cores to the GPU usage management unit 92b. The GPU usage management unit 92b issues a usable core number response in response to this request, and notifies the number of usable cores. The landform cross-section calculation application 103 executes the landform cross-section calculation based on this.

The coordinate conversion application 104 is similar to the terrain section calculation application 103 . That is, since the coordinate conversion application 104 is a non-prioritized application, it first issues a request for the number of usable cores to the GPU usage management unit 92b. The GPU usage management unit 92b issues a usable core number response in response to this request, and notifies the number of usable cores. The coordinate conversion application 104 executes coordinate conversion calculation based on this.

Other configurations are the same as those of the first embodiment.

The same effects as those of the first embodiment can be obtained in the present embodiment configured as described above.

<Fourth Embodiment>
A fourth embodiment of the present invention will be described with reference to FIGS. 12 and 13. FIG.

This embodiment shows a case of notifying that another application is using the GPU in the first embodiment. In the present embodiment, when a non-prioritized application uses the GPU shared area 93b, the current usage information of the GPU shared area 93b (that is, the usage status of the GPU 93 by other applications) is acquired from the GPU operation management unit 92a. do.

FIG. 12 is a timing chart of processing in the GPU operation management unit. Further, FIG. 13 is a flowchart showing the processing contents of the coordinate conversion application. In the figure, the same reference numerals are given to the same members as in the first embodiment, and the description thereof is omitted. Note that FIG. 12 omits illustration of the thread corresponding to the posture measurement application, which is a priority application, for the sake of simplicity of explanation.

In FIG. 12 , the GPU operation management unit 92 a first activates the main thread 301 . The main thread 301 prepares for execution of the application, and a thread 303 corresponding to the terrain section calculation application and a thread 304 corresponding to the coordinate conversion application are started.

The terrain section calculation application 103 always uses the GPU to perform calculations regardless of the state of the GPU shared area 93b, but some delay is allowed. Therefore, the non-prioritized application execution request 108 is issued without checking the state of the GPU shared area 93b. In response to this, the GPU operation management unit 92a executes the terrain section calculation operation, and issues a non-prioritized application completion notice when the execution is completed.

On the other hand, in the coordinate conversion application 104, calculation is performed without using the GPU 93 when the GPU shared area 93b is not available. Therefore, first, a request for the number of usable cores is issued. The GPU operation management unit 92a responds with the number of usable cores at this stage as a usable core number response. In this case, since the GPU shared area 93b is used by the thread 303 corresponding to the terrain section calculation application, the number of cores excluding this area is returned. Upon receiving this, the coordinate conversion application 104 stops issuing the non-prioritized application execution request.

In FIG. 13, the coordinate conversion application 104 first issues a request for the number of usable cores for the GPU shared area 93b (step S400). Subsequently, it is determined whether or not there are enough usable cores (step S410), and if the determination result is YES, a non-prioritized application execution request is issued to the GPU operation management unit 92a (step S420). , the non-prioritized application completion notification is received (step S430), and the process ends. On the other hand, if the determination result in step S410 is NO, that is, if there are not enough available cores in the GPU shared area 93b, the coordinates are calculated only by the CPU 92 without using the GPU 93 (that is, the GPU shared area 93b). A conversion calculation process is executed (step S411), and the process ends.

Other configurations are the same as those of the first embodiment.

The same effects as those of the first embodiment can be obtained in the present embodiment configured as described above.

Also, the GPU shared area 93b can be utilized more efficiently.

<Fifth Embodiment>
A fifth embodiment of the present invention will be described with reference to FIG.

In the present embodiment, the size (the number of cores) of the GPU exclusive area 93a can be dynamically changed in the first embodiment. When the hydraulic excavator 100 enters an area where there are workers around, the operator manually or the system automatically activates a surrounding obstacle detection application for detecting workers, etc., existing around the hydraulic excavator 100. may start up. The surrounding obstacle detection application is an important application (priority application) that should not be delayed and should be executed preferentially. On the other hand, since it is not always running, it is not always registered as a priority application, but is dynamically registered and deleted.

FIG. 14 is a timing chart of processing in the GPU usage management unit. In the figure, the same reference numerals are given to the same members as in the first embodiment, and the description thereof is omitted.

In FIG. 14, when the surrounding obstacle detection application 601 is activated, it issues a priority application registration request to the GPU operation management unit 92a. In response to this, the GPU operation management unit 92a returns a priority application registration response. There are three types of this response. One is a registration completion response indicating that there is sufficient free space in the core and registration has been completed. The other is a registration wait response that allows registration when the operation currently being executed in the shared area is completed although there is no free space. Yet another is a registration failure response that is completely empty. If normal registration is completed, a priority application execution request is issued, the priority application is executed, and a priority application completion notice is received.

Other configurations are the same as those of the first embodiment.

The same effects as those of the first embodiment can be obtained in the present embodiment configured as described above.

Moreover, since priority applications can be added at the time of execution, it is possible to suppress a decrease in GPU usage efficiency compared to the case where all priority applications are always registered.

Next, features of each of the above embodiments will be described.

(1) In the above embodiment, a work machine (for example, the hydraulic excavator 100) that operates at a construction site is a calculation related to the attitude of the work machine, which is a calculation that does not allow delay and is prioritized. A controller 19 for executing allowable calculations, wherein the controller includes an exclusive area (for example, a GPU exclusive area 93a) for performing preferential arithmetic processing set as an arithmetic processing for preferentially performing an arithmetic operation, and the preferential arithmetic operation. A computing device (for example, GPU 93) set with a shared area (for example, GPU shared area 93b) for performing non-prioritized arithmetic processing different from processing, execution of the prioritized arithmetic processing in the exclusive area, and the non-prioritized arithmetic processing and an operation management device (for example, a GPU operation management unit 92a) that manages the execution of priority operation processing in the shared area, and the operation management device , when the operation processing is requested , It is determined whether the computation processing is computation related to the posture of the work machine, which is the computation with priority, or computation for which the delay is allowed , and if it is determined that the computation is related to the posture of the work machine. is executed as the priority arithmetic processing using the exclusive area of the arithmetic unit , and if it is determined that the delay is an allowable arithmetic operation, the operation is left for the exclusive area of the arithmetic unit The non-prioritized arithmetic processing is executed using the shared area.

By configuring in this way, arithmetic processing can be performed by a plurality of applications without interfering with arithmetic processing of important applications.

(2) In the above embodiment, in the working machine (eg, hydraulic excavator 100) of (1), the computing device (eg, GPU 93) is single-task hardware that does not have a task switching function during execution. was assumed to be

(4) Further, in the above embodiment, in the working machine (for example, the hydraulic excavator 100) of (2), the calculations for which the delay is allowed are calculations relating to the terrain around the working machine. bottom.

(5) In the above embodiment, in the work machine (eg, hydraulic excavator 100) of (1), the computation management device (eg, GPU computation management unit 92a) performs to expand the exclusive area (for example, GPU exclusive area 93a) of the arithmetic unit (for example, GPU 93).

(6) In the above embodiment, in the work machine (eg, hydraulic excavator 100) of (1), the operation management device (eg, GPU operation management unit 92a) It is assumed that the usage status is managed and the usage status of the arithmetic unit is returned in response to the request.

(7) In the above embodiment, in the work machine (eg, hydraulic excavator 100) of (1), the operation management device (eg, GPU operation management unit 92a) is activated according to the state of the work area. It was assumed that the proprietary part of the arithmetic unit was expanded according to the demand of the application.

<Appendix>
It should be noted that the present invention is not limited to the above-described embodiments, and includes various modifications and combinations within the scope of the invention. Moreover, the present invention is not limited to those having all the configurations described in the above embodiments, and includes those having some of the configurations omitted. Further, each of the above configurations, functions, etc. may be realized by designing a part or all of them, for example, with an integrated circuit. Moreover, each of the above configurations, functions, etc. may be realized by software by a processor interpreting and executing a program for realizing each function.

REFERENCE SIGNS LIST 1 front work machine 2 upper revolving body (vehicle body) 2a revolving motor 3 lower traveling body 3a traveling motor 4 boom 4a boom cylinder 5 arm 5a arm cylinder 6 Bucket 6a Bucket cylinder 7 Hydraulic pump device 8 Control valve 9 Driver's cab 9a, 9b Operating lever (operating device) 12, 14 to 16 Inertial Measurement Unit (IMU) ), 17... Construction information, 19... Controller, 19a... Posture calculation unit, 19b... Monitor display control unit, 19c... Hydraulic system control unit, 19d... Construction target surface calculation unit, 91... Input interface, 92... Central processing unit ( CPU), 92a... GPU operation management unit, 92b... GPU use management unit, 93... Image processing unit (GPU), 93a... GPU exclusive area, 93b... GPU shared area, 94... Read only memory (ROM), 95... Random Access memory (RAM), 96... Output interface, 100... Hydraulic excavator, 102... Attitude measurement application, 103... Terrain section calculation application, 104... Coordinate conversion application, 105... Priority application registration request, 106... Non-priority application registration request, 107 Priority application execution request, 108 Non-priority application execution request, 202 Calculation for attitude measurement application, 203 Calculation for terrain cross-section calculation application, 204 Calculation for coordinate transformation application, 301 Main thread, 302 Attitude measurement application Corresponding thread 303 Thread corresponding to terrain section calculation application 304 Thread corresponding to coordinate conversion application 403 Request for number of usable cores 404 Request for number of usable cores 601 Surrounding obstacle detection application

Claims (6)

A working machine that operates at a construction site,
a controller for executing calculations relating to the attitude of the work machine, which are calculations that are not allowed for delay and are prioritized, and calculations for which delay is allowed;
The controller is
an arithmetic device having an exclusive area for performing preferential arithmetic processing set as an arithmetic processing for preferential arithmetic processing and a shared region for performing non-prioritized arithmetic processing different from the preferential arithmetic processing;
an operation management device that manages execution of the priority operation processing in the exclusive area and execution of the non-priority operation processing in the shared area;
The operation management device is
When computational processing is requested,
determining whether the requested arithmetic processing is a calculation relating to the posture of the working machine, which is the priority calculation , or a calculation for which the delay is allowed;
if it is determined that the calculation is related to the posture of the working machine, executing as the priority calculation process using the exclusive area of the calculation device ;
When it is determined that the operation is allowed for the delay, the operation is executed as the non-prioritized operation processing using the shared area left for the exclusive area of the operation device . machine. The work machine according to claim 1,
A working machine, wherein the computing device is single-task hardware that does not have a function of switching tasks during execution. The work machine according to claim 2,
A working machine according to claim 1, wherein the delay-allowable calculation is a calculation relating to terrain surrounding the working machine. The work machine according to claim 1,
A working machine, wherein the arithmetic management device expands an exclusive area of the arithmetic device according to the number of the priority arithmetic processes. The work machine according to claim 1,
A working machine, wherein the operation management device manages the usage status of the operation device and returns the usage status of the operation device in response to a request. The work machine according to claim 1,
The operation management unit expands the exclusive portion of the operation unit according to the request of the application activated according to the state of the work area.

Images