Simulation task allocation method and system for unmanned system simulation test

Abstract

The invention discloses a simulation task allocation method and system for unmanned system simulation testing, and relates to the technical field of unmanned system simulation testing, and the method comprises the steps: building a corresponding relation between each calculation node in a calculation cluster and each task allocation node in a control cluster; the access end sends a simulation task request to the control cluster; a task management node in the control cluster manages a simulation task request list based on the simulation task request; each task distribution node in the cluster is controlled to distribute simulation tasks to corresponding computing nodes based on the simulation task request list, and the simulation tasks are distributed according to a preset sequence; the preset sequence is as follows from front to back: the computing nodes which are not allocated to the simulation tasks currently, the computing nodes of which the number of the allocated simulation tasks is smaller than the recommended allocation number, and the computing nodes of which the number of the allocated simulation tasks is smaller than the maximum allocation number. According to the method, the processing overhead is balanced, and the smooth operation of the large-scale unmanned system simulation test can be promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned system simulation testing, and in particular to a simulation task allocation method and system for unmanned system simulation testing. Background Art

[0002] With the development of intelligent technology, unmanned systems (such as unmanned vehicles, drones, unmanned ships, robots, etc.) have been applied to various fields. In order to ensure the safety, stability, and reliability of the operation of unmanned systems, it is necessary to fully test and verify the performance of unmanned systems.

[0003] Since the testing of unmanned systems in real scenarios has defects such as long test cycle, high cost and low efficiency, simulation test methods based on virtual scenarios have been proposed and continuously studied and applied. The relevant technology controls the behavior of the unmanned system in the simulation test through the access end; in addition, the relevant technology generates the operating environment of the simulation test, NPC, simulation nodes corresponding to the unmanned system, etc. through the simulation engine, and controls the simulation node to execute the control signaling of the access end in the operating environment. The simulation engine can be implemented as UE (Unreal Engine), Unity, etc. During the simulation test, the simulation engine is responsible for simulation calculation to generate simulation data, and then the access end determines the control amount of the unmanned system based on the simulation data. For large-scale unmanned system simulation testing, in order to ensure the smooth operation of the simulation test, the relevant technology also proposes a distributed simulation architecture, which consists of a computing cluster composed of multiple computing nodes. The simulation engine runs on each computing node in the computing cluster. Each computing node is responsible for the simulation calculation of part of the unmanned system, sharing the resource processing pressure of large-scale unmanned system simulation testing.

[0004] Based on this, how to reasonably allocate the simulation tasks of the unmanned system to each computing node in the computing cluster to balance the processing overhead of each computing node and avoid the situation where individual computing nodes are overloaded and cause simulation test failures has become a problem that needs to be solved urgently. Summary of the invention

[0005] Purpose of the invention: In view of the problems existing in the prior art, the present invention provides a simulation task allocation method and system for unmanned system simulation testing that can balance processing overhead.

[0006] Technical solution: The simulation task allocation method for unmanned system simulation test of the present invention is implemented based on a system including an access terminal, a control cluster and a computing cluster. The method includes:

[0007] Establish the corresponding relationship between each computing node in the computing cluster and each task allocation node in the control cluster;

[0008] The access end sends a simulation task request to the control cluster;

[0009] The task management node in the control cluster manages the simulation task request list based on the simulation task request;

[0010] Each task allocation node in the control cluster allocates simulation tasks to the corresponding computing nodes based on the simulation task request list; wherein the simulation tasks are allocated in a preset order, and the preset order is from front to back: computing nodes that are not currently allocated to simulation tasks, computing nodes that have been allocated with a number of simulation tasks less than the recommended allocation number, and computing nodes that have been allocated with a number of simulation tasks less than the maximum allocation number.

[0011] Optionally, the correspondence between each computing node and each task allocation node is a many-to-many relationship.

[0012] Optionally, each computing node is only responsible for the simulation task of one type of unmanned system, and the number of computing nodes in the computing cluster that are responsible for the simulation task of the same type of unmanned system is greater than or equal to 2.

[0013] Optionally, the task management node in the control cluster manages a simulation task request list based on the simulation task request, including: the task management node determines a sorting priority of the simulation task request based on at least one of the waiting time during which the simulation task request is not received, the number of times the simulation task request is released again after being received, and the waiting time during which the simulation task request is not received after being released again; and the task management node updates the simulation task request list according to the sorting priority of the simulation task request when receiving a simulation task request, a simulation task request is received, or a simulation task request is released.

[0014] Optionally, each task allocation node in the control cluster allocates simulation tasks to the corresponding computing node based on the simulation task request list, including: the task allocation node reads the simulation task request from the simulation task request list; the task allocation node determines whether to accept the current simulation task request based on the simulation task type and / or the number of simulation tasks; when determining to accept the current simulation task request, the task allocation node obtains k computing nodes and the allocation priorities of the k computing nodes; wherein the k computing nodes refer to computing nodes among the multiple computing nodes corresponding to the task allocation node, and the unmanned system type corresponding to the simulation task for which the task allocation node is responsible is the same as the unmanned system type included in the simulation task request; in the tth round of allocation, the task allocation node sends a task allocation request RQ to computing node i according to the allocation priority. i ; Where t is a positive integer less than or equal to 3; the computing node i sends a task allocation response RE to the task allocation node based on the current simulation state and the value of t i; Among them, the current simulation state of the computing node includes: whether it has been assigned to a simulation task, the recommended number of simulation tasks to be assigned, and the maximum number of simulation tasks to be assigned; the task assignment node responds to the task assignment RE based on the task assignment response i , calculate the number of simulation tasks to be assigned, and if the number of simulation tasks is not zero, keep t unchanged and set ii+1, and send a task assignment request RQ to computing node i according to the assignment priority i The step is executed again until the number of simulation tasks is zero or until i+1 is greater than k; if the number of simulation tasks to be assigned by the task assignment node is still not zero when i+1 is greater than k, then let t=t+1 and let i=1, and send a task assignment request RQ to the computing node i according to the assignment priority. i The steps are executed again starting from the step i+1 until the number of simulation tasks is zero or until i+1 is greater than k.

[0015] Optionally, the computing node i sends a task allocation response RE to the task allocation node based on the current simulation state and the value of t. i , including: when t is equal to 1, if the computing node i has been assigned to a simulation task, a first task assignment response is sent to the task assignment node, and the first task assignment response is used to indicate the rejection of the task assignment; if the computing node i has not been assigned to a simulation task, a second task assignment response is sent to the task assignment node, and the second task assignment response is used to indicate the acceptance of the task assignment and the number of simulation tasks currently accepted and assigned; when t is equal to 2, if the number of simulation tasks currently assigned to the computing node i is equal to the recommended number of simulation tasks that the computing node i is responsible for, a first task assignment response is sent to the task assignment node, and the first task assignment response is used to indicate the rejection of the task assignment; if the number of simulation tasks currently assigned to the computing node i is less than the number of simulation tasks currently assigned to the computing node i, If the recommended number of simulation tasks that the computing node is responsible for is the same as the recommended number of simulation tasks that the computing node is responsible for, a second task allocation response is sent to the task allocation node, and the second task allocation response is used to indicate the acceptance of the task allocation and the number of simulation tasks that are currently accepted for allocation; when t is equal to 3, if the number of simulation tasks that the computing node i has currently been allocated is equal to the maximum number of simulation tasks that the computing node i is responsible for, a first task allocation response is sent to the task allocation node, and the first task allocation response is used to indicate the rejection of the task allocation; if the number of simulation tasks that the computing node i has currently been allocated is less than the maximum number of simulation tasks that the computing node i is responsible for, a second task allocation response is sent to the task allocation node, and the second task allocation response is used to indicate the acceptance of the task allocation and the number of simulation tasks that are currently accepted for allocation.

[0016] Optionally, the method for determining the recommended allocation quantity of the simulation task and the maximum allocation quantity of the simulation task includes: obtaining hardware indicator reference parameters and their corresponding reference quantity configurations; the hardware indicator reference parameters refer to hardware indicator parameters of the reference computer device; based on the hardware indicator reference parameters and the hardware indicator target parameters, determining the quantity configuration ratio; the hardware indicator target parameters refer to the hardware indicator parameters of the computer device where the computing node is located; based on the quantity configuration ratio and the reference quantity configuration, obtaining the recommended allocation quantity and the maximum allocation quantity.

[0017] The simulation task allocation system for unmanned system simulation test of the present invention comprises an access terminal, a control cluster and a computing cluster, wherein:

[0018] The access terminal includes a first communication node, the first communication node is used to establish a communication connection with the control cluster and send a simulation task request to the control cluster;

[0019] The control cluster includes a second communication node, a task management node and a plurality of task allocation nodes, wherein the second communication node is used to establish a communication connection with the access terminal and the computing cluster, the task management node is used to manage a simulation task request list based on the simulation task request, and the task allocation node is used to allocate simulation tasks to corresponding computing nodes based on the simulation task request list; wherein the simulation tasks are allocated in a preset order, and the preset order is from front to back: computing nodes that are not currently allocated to simulation tasks, computing nodes that have been allocated with a number of simulation tasks less than the recommended allocation number, and computing nodes that have been allocated with a number of simulation tasks less than the maximum allocation number;

[0020] The computing cluster includes a plurality of computing nodes, each computing node includes a third communication node and a simulation computing node, the third communication node is used to establish a communication connection with the control cluster, and the simulation computing node is used to perform simulation computing according to the assigned simulation tasks.

[0021] Optionally, the correspondence between each computing node and each task allocation node is a many-to-many relationship.

[0022] Optionally, each computing node is only responsible for the simulation task of one type of unmanned system, and the number of computing nodes in the computing cluster that are responsible for the simulation task of the same type of unmanned system is greater than or equal to 2.

[0023] The present invention has at least the following innovative features and beneficial effects:

[0024] (1) Add a control cluster, and perform multiple rounds of progressive task allocation to the computing cluster through the control cluster. Beneficial effect: In the present invention, the task allocation node in the control cluster preferentially allocates simulation tasks to computing nodes in the computing cluster that are not currently allocated to simulation tasks, and secondly allocates simulation tasks to computing nodes that are currently allocated to simulation tasks but the number of simulation tasks allocated is less than the recommended allocation number, and finally allocates simulation tasks to computing nodes that are currently allocated to simulation tasks but the number of simulation tasks allocated is less than the maximum allocation number. Through such three rounds of progressive allocation process, the processing overhead of each computing node is effectively balanced, considering the current simulation status and hardware resource occupancy rate of each computing node in the computing cluster, and realizing the reasonable allocation of simulation tasks, which is helpful to promote the smooth operation of large-scale unmanned system simulation testing.

[0025] (2) The task management node in the control cluster maintains a simulation task list, and the task allocation node reads the simulation task request from the simulation task list in sequence. Beneficial effect: In the present invention, the simulation task list is maintained by the task management node. The longer the waiting time of the simulation task request that has not been received, the more the number of releases after being received, the more the number of releases under the same waiting time, and the longer the waiting time under the same number of releases, the higher the sorting priority of the simulation task request in the simulation task list, so that when the task allocation node reads the simulation task request in sequence, it has priority to read the simulation task request, which helps to balance the response time of each simulation task request and avoid individual simulation task requests not being read and received for a long time. In addition, each task allocation node maintains a release list of simulation task requests to record the simulation task requests that are released again after being received. When reading the simulation task list later, it is directly judged not to receive these simulation task requests to avoid repeatedly receiving the same simulation task request that cannot be allocated by itself, avoid affecting the successful response of the simulation task request by other task allocation nodes, and improve the response success rate of the simulation task request.

[0026] (3) The simulation tasks of the same type of unmanned system are in charge of at least two computing nodes, and there is a many-to-many correspondence between the task allocation nodes and the computing nodes. Beneficial effects: In the present invention, in order to facilitate the management and allocation of simulation tasks and improve the efficiency of simulation task allocation, one computing node is only responsible for the simulation tasks of one type of unmanned system; but the simulation tasks of the same type of unmanned system are in charge of at least two computing nodes, which can avoid the normal operation of the simulation test being affected by the disconnection or crash of individual computing nodes. In addition, the present invention sets a plurality of task allocation nodes in the control cluster, and the simulation task allocation of one computing node is managed by at least two task allocation nodes, and one task allocation node is responsible for managing the simulation task allocation of at least two computing nodes, which effectively avoids the disconnection or crash of a single task allocation node or a single computing node affecting the normal operation of task allocation and simulation testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart of a simulation task allocation method for unmanned system simulation test provided by the present invention;

[0028] Figure 2 It is a schematic diagram of the architecture of a simulation task allocation system for unmanned system simulation testing provided by the present invention;

[0029] Figure 3 It is a schematic diagram of the simulation task allocation execution flow in the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects rather than to describe a specific order. The reference to "embodiment" in this article means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0032] Embodiment 1

[0033] This embodiment provides a simulation task allocation method for unmanned system simulation test, the steps of the method are as follows: Figure 1 As shown, the method is implemented based on a system including an access terminal, a control cluster and a computing cluster. The architecture of the system is as follows Figure 2 The method of this embodiment is described in detail below. The method of this embodiment specifically includes the following steps (steps 1 to 4).

[0034] Step 1: Establish a corresponding relationship between each computing node in the computing cluster and each task allocation node in the control cluster.

[0035] The computing cluster is composed of m computing nodes, where m is an integer greater than 1. To ensure that the simulation task requests from each access terminal can be quickly responded to and quickly assigned to a specific computing node, n task allocation nodes are set in the control cluster, where n is an integer greater than 1. At the beginning of the simulation test, the control cluster and the computing cluster are started, and each computing node in the computing cluster logs in to the control cluster and establishes a communication connection with the control cluster. Among them, each computing node is pre-set with the IP address of the control cluster. The third communication node in each computing node sends a communication connection request to the second communication node in the control cluster according to the IP address. The second communication node in the control cluster establishes an exclusive communication channel with the third communication node in each computing node according to the communication connection request. The subsequent data and information between the control cluster and each computing node are transmitted through the corresponding exclusive communication channel.

[0036] The second communication node of the control cluster also establishes a corresponding relationship between each computing node and each task allocation node according to the received communication connection request, and stores it in a corresponding relationship table, wherein the corresponding relationship is specifically a many-to-many relationship. The corresponding relationship table can be stored in the second communication node, or sent to the task allocation node for storage. The many-to-many relationship means that a task allocation node is responsible for managing the simulation task allocation of multiple computing nodes, and the simulation task allocation of a computing node is managed by multiple task allocation nodes, which can effectively avoid the situation that when a one-to-many or many-to-one corresponding relationship is adopted, a single task allocation node or a single computing node is disconnected or crashed, etc., which affects the normal operation of task allocation and simulation test. Optionally, multiple computing nodes corresponding to a task allocation node can be responsible for the simulation tasks of the same type of unmanned system, or they can be responsible for the simulation tasks of different types of unmanned systems respectively.

[0037] In the present invention, in order to facilitate the allocation and management of simulation tasks and improve the efficiency of simulation task allocation, a computing node is only responsible for the simulation tasks of one type of unmanned system. Optionally, the classification basis of the unmanned system includes but is not limited to: the attributes of the unmanned system (such as unmanned aerial vehicles, unmanned vehicles, etc.), the initial parameters of the unmanned system (such as initial speed, initial position, initial attitude, etc.). It should be understood that for unmanned systems with different power types, the unmanned systems are also considered to have different attributes, so that the unmanned systems are considered to belong to different types. For example, for unmanned aerial vehicles with fixed-wing and quad-rotor wing types, they are considered to have different attributes and thus belong to different types of unmanned systems.

[0038] In addition, in order to alleviate the impact of computing node disconnection or crash on the normal operation of the simulation test, the simulation tasks of the same type of unmanned system are handled by at least two computing nodes. Optionally, the unmanned system type corresponding to the simulation task responsible for each computing node is pre-configured by the user, so that the communication connection request sent by each computing node to the control cluster can also include the unmanned system type; or, the unmanned system type corresponding to the simulation task responsible for each computing node is allocated by the control cluster, so that after each computing node logs in to the control cluster, the control cluster cyclically allocates all unmanned system types involved in the simulation test to the computing node according to the login order of the computing node.

[0039] Step 2: The access end sends a simulation task request to the control cluster.

[0040] Each access terminal is pre-set with the IP address of the control cluster. The first communication node in each access terminal sends a communication connection request to the second communication node in the control cluster according to the IP address. The second communication node in the control cluster establishes an exclusive communication channel with the first communication node in each access terminal according to the communication connection request. The subsequent data and information between the control cluster and each access terminal are transmitted through the corresponding exclusive communication channel.

[0041] Taking one of the access terminals as an example, after the access terminal establishes a communication connection with the control cluster, the access terminal determines the required simulation task type and simulation task quantity, and sends a simulation task request to the second communication node of the control cluster. The simulation task request includes the simulation task type (unmanned system type) and the simulation task quantity; for example, if the unmanned system type included in the simulation task request is a fixed-wing UAV and the simulation task quantity is 20, then the access terminal requests to generate simulation nodes corresponding to 20 fixed-wing UAVs and requests to execute simulation tasks corresponding to 20 fixed-wing UAVs.

[0042] Step 3: The task management node in the control cluster manages the simulation task request list based on the simulation task request.

[0043] After receiving the simulation task request, the second communication node in the control cluster forwards the simulation task request to the task management node in the control cluster. The task management node maintains and manages a simulation task list, which includes simulation task requests that have not been successfully responded to. In the present invention, the simulation task request that has not been successfully responded to includes a simulation task request that has not been received by the task allocation node, and a simulation task request that has been received by the task allocation node and released again.

[0044] The present invention does not limit the specific sorting basis of the simulation task request in the simulation task list. Optionally, the basis for determining the sorting priority of the simulation task request includes but is not limited to: the waiting time of the simulation task request not being received, the number of releases after the simulation task request is received, and the waiting time of the simulation task request not being received after being released again. Among them, the longer the waiting time, the more releases, the more releases under the same waiting time, and the longer the waiting time under the same release number, the higher the sorting priority of the simulation task request.

[0045] Exemplarily, assume that there are 6 simulation task requests in the simulation task list, and assume that simulation task request 1 has not been received and the waiting time for not being received is 1s, simulation task request 2 has been released again after being received twice and the waiting time for not being received after being released again is 1s, simulation task request 3 has not been received and the waiting time for not being received is 2s, simulation task request 4 has been released again after being received once and the waiting time for not being received after being released again is 2s, simulation task request 5 has not been received and the waiting time for not being received is 0.5s, simulation task request 6 has been released again after being received once and the waiting time for not being received after being released again is 0.5s, then the simulation task requests in the simulation task list are sorted in order of priority from high to low as follows: simulation task request 2, simulation task request 4, simulation task request 6, simulation task request 3, simulation task request 1, and simulation task request 5.

[0046] The present invention does not limit the specific update time of the simulation task list. Optionally, the task management node updates the status of each simulation task request in the simulation task list in real time (such as the waiting time that has not been received, etc.); in addition, the task management node adds the simulation task request to the simulation task list every time it receives a simulation task request from the access terminal, adds the simulation task request to the simulation task list every time it receives a simulation task request released by the task allocation node, and removes the simulation task request from the simulation task list every time the task allocation node receives the simulation task request; on this basis, the task management node can also detect whether there is an unresponsive simulation task request every period of time (such as 5s, 10s, etc.), and remove the simulation task request if it exists to update the simulation task list. Wherein, the unresponsive simulation task request includes at least one of the following characteristics: the waiting time that has not been received exceeds the preset time (such as 10s, 15s, etc.), and the number of releases after being received and released again reaches a preset number of times (such as 3 times, 5 times, etc.). Optionally, when removing a simulation task request that cannot be responded to, the task management node may, in cooperation with the second communication node, send a request failure indication to the access terminal that sent the simulation task request. The access terminal may subsequently send the simulation task request again after modifying the simulation task request or after waiting for a period of time.

[0047] Step 4: Each task allocation node in the control cluster allocates simulation tasks to corresponding computing nodes based on the simulation task request list.

[0048] Among them, the simulation tasks are allocated in a preset order, and the preset order is from front to back: computing nodes that are currently not allocated to simulation tasks, computing nodes that have been allocated a number of simulation tasks that is less than the recommended allocation number, and computing nodes that have been allocated a number of simulation tasks that is less than the maximum allocation number.

[0049] In one embodiment, Figure 3 As shown, the above step 4 includes the following sub-steps (steps 4.1 to 4.7).

[0050] Step 4.1: The task allocation node reads the simulation task request from the simulation task request list.

[0051] Each task allocation node can read the simulation task list from the task management node at intervals of a certain period (such as 0.5s, 1s, etc.) in an idle state. The idle state means that the task allocation node does not perform specific simulation task allocation. When reading the simulation task list, the task allocation node reads each simulation task request in the order of the sorting priority from high to low in step 3 above.

[0052] Step 4.2: The task allocation node determines whether to receive the current simulation task request based on the simulation task type and / or the number of simulation tasks.

[0053] When the task allocation node reads a simulation task request, it determines whether the simulation task request can be received; if it is determined to receive the current simulation task request, it will no longer read the remaining simulation task requests; if it is determined not to receive the current simulation task request, it will continue to read the remaining simulation task requests. That is, if the judgment result of step 4.2 is yes, it will continue to execute step 4.3; if the judgment result of step 4.2 is no, it will return to execute step 4.1.

[0054] When judging whether to receive the current simulation task request, the task allocation node can obtain the unmanned system type corresponding to the simulation task that the multiple computing nodes corresponding to itself are responsible for respectively. If there is an unmanned system type corresponding to the simulation task that at least one computing node is responsible for, which is the same as the unmanned system type (simulation task type) included in the current simulation task request, then the task allocation node determines to receive the current simulation task request, otherwise it determines not to receive the current simulation task request; or, the task allocation node can obtain the number of computing nodes corresponding to itself, if the ratio between the number and the number of simulation tasks included in the current simulation task request is greater than or equal to a preset threshold (such as 0.5, 0.25, etc.), then the task allocation node determines to receive the current simulation task request, otherwise it determines not to receive the current simulation task request. Of course, the task allocation node can also combine the above two methods to determine whether to receive the current simulation task request. If the receiving conditions in the above two methods are met, it is determined to receive the current simulation task request; if any of the receiving conditions in the above two methods is not met or all of the receiving conditions are not met, it is determined not to receive the current simulation task request.

[0055] Optionally, in the present invention, each task allocation node maintains a release list of simulation task requests, and the release list is used to record the simulation task requests that are released again after being received by the task allocation node. Therefore, after releasing a simulation task request, when the task allocation node reads the simulation task list from the task management node again, it directly determines not to receive the simulation task request when reading the simulation task request, so as to avoid repeatedly receiving the same simulation task request that it cannot allocate, avoid affecting the successful response of the simulation task request by other task allocation nodes, and improve the response success rate of the simulation task request.

[0056] Step 4.3: The task allocation node obtains k computing nodes and the allocation priorities of the k computing nodes.

[0057] Among them, k computing nodes refer to computing nodes among the multiple computing nodes corresponding to the task allocation node, whose unmanned system type corresponding to the simulation task is the same as the unmanned system type included in the simulation task request; k is a positive integer less than or equal to m. Optionally, the allocation priority order of the k computing nodes is preset, for example, the allocation priority order is configured and stored according to the order in which the k computing nodes log in to the control cluster; or, the allocation priority order of the k computing nodes is randomly generated, for example, after the task allocation node obtains the k computing nodes, the allocation priority order of the k computing nodes is generated in real time through a random algorithm.

[0058] Step 4.4: In the tth round of allocation, the task allocation node sends a task allocation request RQ to the computing node i according to the allocation priority. i .

[0059] In the present invention, in order to balance the processing overhead of each computing node, the task allocation node preferentially allocates the simulation task to the computing node that is not currently allocated to the simulation task, and secondly allocates the simulation task to the computing node that is currently allocated to the simulation task but the number of the allocated simulation tasks is less than the recommended allocation number, and finally allocates the simulation task to the computing node that is currently allocated to the simulation task but the number of the allocated simulation tasks is less than the maximum allocation number. Based on this, the task allocation node needs to perform at most three rounds of allocation process when allocating simulation tasks, and t is a positive integer less than or equal to 3.

[0060] Since one computing node corresponds to multiple task allocation nodes, there may be situations where other task allocation nodes have already allocated simulation tasks to the computing node. Based on this, each task allocation node in the present invention does not record whether a simulation task is allocated to a specific computing node and the number of simulation tasks allocated. Instead, each time a task allocation is executed, the specific computing node is asked whether it currently accepts the task allocation through a "request-response" mode. Therefore, in each round of allocation, the task allocation node must send task allocation requests to the k computing nodes in turn in the order of priority allocation in the above step 4.3. For computing node i (i is a positive integer less than k), the task allocation node sends a task allocation request RQ to computing node i. i Among them, the task allocation request RQ i Included in the task allocation node is the number of simulation tasks currently to be allocated, which is less than or equal to the number of simulation tasks included in the simulation task request (when t is equal to 1 and i is equal to 1, the number of simulation tasks is equal to the number of simulation tasks included in the simulation task request).

[0061] Step 4.5: Based on the current simulation state and the value of t, computing node i sends a task assignment response RE to the task assignment node. i .

[0062] After receiving the task allocation request, each computing node needs to determine whether to accept the task allocation and the number of simulation tasks currently accepted based on the current allocation round (i.e., the value of t) and its own current simulation state. The current simulation state of the computing node includes: whether it has been currently allocated a simulation task, the recommended number of simulation tasks to be allocated, and the maximum number of simulation tasks to be allocated.

[0063] When t is equal to 1 (i.e., the current allocation round is the first round), if computing node i has been assigned to a simulation task, a first task allocation response (i.e., task allocation response RE ) is sent to the task allocation node. i is the first task assignment response), the first task assignment response is used to indicate the rejection of task assignment; if the computing node i is not assigned to the simulation task, a second task assignment response (ie, task assignment response RE) is sent to the task assignment node. i The second task allocation response is used to indicate the acceptance of the task allocation and the number of simulation tasks currently accepted for allocation. It should be understood that when t is equal to 1, the task allocation response RE i The number of simulation tasks currently assigned in is less than or equal to the recommended number of simulation tasks that computing node i is responsible for.

[0064] When t is equal to 2 (i.e., the current allocation round is the second round), it can be determined that all k computing nodes have been currently allocated to simulation tasks. Therefore, if the number of simulation tasks currently allocated to computing node i is equal to the recommended number of simulation tasks that computing node i is responsible for, a first task allocation response (i.e., task allocation response RE) is sent to the task allocation node. i The first task assignment response is used to indicate the rejection of task assignment; if the number of simulation tasks currently assigned to computing node i is less than the recommended number of simulation tasks that computing node i is responsible for, a second task assignment response (i.e., task assignment response RE) is sent to the task assignment node. i The second task allocation response is used to indicate the acceptance of the task allocation and the number of simulation tasks currently accepted for allocation. It should be understood that when t is equal to 2, the task allocation response RE i The number of simulation tasks currently assigned to computing node i is less than the recommended number of simulation tasks that computing node i is responsible for, or in other words, less than or equal to the difference between the recommended number of simulation tasks that computing node i is responsible for and the number of simulation tasks currently assigned to computing node i.

[0065] When t is equal to 3 (i.e., the current allocation round is the third round), it can be determined that all k computing nodes have been currently allocated to simulation tasks and the number of simulation tasks that have been allocated is equal to or greater than the recommended allocation number. Therefore, if the number of simulation tasks that has been currently allocated to computing node i is equal to the maximum allocation number of simulation tasks that computing node i is responsible for, a first task allocation response (i.e., task allocation response RE ) is sent to the task allocation node. i The first task assignment response is used to indicate the rejection of task assignment; if the number of simulation tasks currently assigned to computing node i is less than the maximum number of simulation tasks that computing node i is responsible for, a second task assignment response (i.e., task assignment response RE) is sent to the task assignment node. i The second task allocation response is used to indicate the acceptance of the task allocation and the number of simulation tasks currently accepted for allocation. It should be understood that when t is equal to 3, the task allocation response RE i The number of simulation tasks currently assigned to computing node i is less than the maximum number of simulation tasks that computing node i is responsible for, or in other words, less than or equal to the difference between the maximum number of simulation tasks that computing node i is responsible for and the number of simulation tasks currently assigned to computing node i.

[0066] Step 4.6: The task allocation node responds to the task allocation RE based on the task allocation response i , calculate the number of simulation tasks to be assigned, and if the number of simulation tasks is not zero, keep t unchanged and set ii+1, and execute again from the above step 4.4 until the number of simulation tasks is zero or until i+1 is greater than k.

[0067] Among the k computing nodes, the allocation priority of computing node i is higher than the allocation priority of computing node i+1. After each task allocation node receives a task allocation response, it updates the current number of simulation tasks to be allocated based on the task allocation response. If the current number of simulation tasks to be allocated is zero, it means that all simulation tasks included in the simulation task request have been allocated and the simulation task request has been successfully responded to; if the current number of simulation tasks to be allocated is not zero, the task allocation node continues to send task allocation requests to the next computing node (i.e., ii+1) in the current allocation round (i.e., keep t unchanged) in the order of allocation priority.

[0068] It should be understood that the task allocation request RQ i+1 and task allocation request RQ i The number of simulation tasks currently to be assigned to the task assignment nodes included in may be the same or different, depending on the task assignment response RE i If the task assignment response RE iUsed to indicate rejection of task assignment (i.e. task assignment response RE i is the first task allocation response), then the task allocation request RQ i+1 and task allocation request RQ i The number of simulation tasks currently to be assigned to the task assignment nodes included in the task assignment is the same; if the task assignment response RE i Used to indicate acceptance of task assignment (i.e. task assignment response RE i For the second task allocation response), the task allocation request RQ i+1 and task allocation request RQ i The number of simulation tasks currently to be assigned to the task assignment nodes included in the task assignment nodes is different, and the difference in the number of simulation tasks between them is equal to the task assignment response RE i Indicates the number of simulation tasks currently assigned.

[0069] Step 4.7: If the number of simulation tasks to be assigned by the task allocation node is still not zero when i+1 is greater than k, set t=t+1 and i=1, and execute again from step 4.4 above until the number of simulation tasks is zero or until i+1 is greater than k.

[0070] If in the current allocation round, the task allocation node has completed the task allocation query for all k computing nodes (that is, when i+1 is greater than k), and the current number of simulation tasks to be allocated is still not zero, then continue to the next allocation round, set t=t+1 and i=1, and start again from the above step 4.4.

[0071] based on Figure 3 In the execution process of steps 4.1 to 4.7 above, in each allocation round, the task allocation node asks k computing nodes in turn whether to accept the task allocation. If the task allocation node completes three rounds of allocation, that is, when t+1>3, the number of simulation tasks to be allocated is still not zero, then the task allocation node releases the simulation task request to the task management node again, and records the simulation task request in the release list to avoid repeated reading.

[0072] One thing that needs to be explained is that different types of unmanned systems require different computing resources to execute simulation tasks, so the specific number of simulation tasks that can be executed varies depending on the hardware level of the computing node or the type of unmanned system corresponding to the simulation task. Based on this, before each computing node is assigned a simulation task, it is necessary to determine the number of simulation tasks it can execute, including the recommended allocation number and the maximum allocation number, to provide a basis for subsequent simulation task allocation, avoid being assigned too many simulation tasks, resulting in system crashes, and ensure the smooth operation of the simulation test.

[0073] Based on this, in one example, the simulation task allocation method provided by the present invention further includes the following steps (steps 011 to 013).

[0074] Step 011: Obtain hardware indicator reference parameters and their corresponding reference quantity configurations.

[0075] The hardware index reference parameter refers to the hardware index parameter of the reference computer device, and the reference computer device can be any computer device. Optionally, the hardware index includes but is not limited to at least one of the following: memory, CPU (Central Processing Unit), GPU (Graphic Processing Unit), etc.

[0076] The present invention is directed to various types of unmanned systems, and respectively runs simulation tasks of the type of unmanned systems on reference computer devices, so as to obtain the running number of the simulation tasks when the hardware resource occupancy ratio is a first value and a second value, thereby obtaining the recommended allocation number and the maximum allocation number of the reference computer device when executing the simulation task of the type of unmanned system. The first value is less than the second value, for example, the first value is 80% and the second value is 100%. The reference quantity configuration includes the recommended allocation number and the maximum allocation number of the simulation tasks of various types of unmanned systems.

[0077] Optionally, each computing node in the computing cluster can obtain hardware indicator reference parameters and their corresponding reference quantity configurations from the cloud via the Internet; or, can obtain hardware indicator reference parameters and their corresponding reference quantity configurations from the control cluster through a corresponding dedicated communication channel; or, the hardware indicator reference parameters and their corresponding reference quantity configurations can be pre-configured in each computing node.

[0078] Step 012: Determine the quantity configuration ratio based on the hardware indicator reference parameters and the hardware indicator target parameters.

[0079] The hardware indicator target parameters refer to the hardware indicator parameters of the computer equipment where each computing node is located. After obtaining the hardware indicator reference parameters, each computing node determines the quantity configuration ratio based on the ratio of the target parameters and reference parameters of each hardware indicator. Among them, the quantity configuration ratio is the minimum ratio of the target parameters and reference parameters of all hardware indicators.

[0080] Step 013: Based on the quantity configuration ratio and the reference quantity configuration, the target quantity configuration is obtained.

[0081] The target quantity configuration includes the recommended allocation quantity and maximum allocation quantity of the simulation tasks that the computing node is responsible for. At the beginning of the simulation test, each computing node needs to clarify the type of unmanned system corresponding to the simulation task it is responsible for; then obtain the recommended allocation quantity and maximum allocation quantity of the reference computer equipment for the simulation tasks of the corresponding type of unmanned system from the obtained reference quantity configuration, and then increase or decrease the recommended allocation quantity and maximum allocation quantity in the reference quantity configuration according to the quantity configuration ratio to obtain the recommended allocation quantity and maximum allocation quantity of the simulation task it is responsible for.

[0082] It should be understood that in addition to obtaining the recommended allocation number and the maximum allocation number of the simulation tasks that each computing node is responsible for through the above steps 011 to 013, the recommended allocation number and the maximum allocation number can also be obtained through actual testing. For example, each computing node runs the simulation task that it is responsible for on its computer device, so as to obtain the running number of the simulation task when the hardware resource occupancy ratio is the first value and the second value, respectively, so as to obtain the recommended allocation number and the maximum allocation number.

[0083] Embodiment 2

[0084] This embodiment provides a simulation task allocation system for unmanned system simulation testing. The system can be configured to implement the method of embodiment 1. Figure 2 As shown, the system includes an access terminal, a control cluster and a computing cluster. Each part is introduced in detail below.

[0085] The access terminal is used to calculate the control amount of the unmanned system based on the simulation data of the unmanned system. The access terminal includes a first communication node and a control amount calculation node. The first communication node is used to establish a communication connection with the control cluster and send a simulation task request to the control cluster. Optionally, the first communication node is also used to receive the simulation data forwarded by the control cluster and send the control amount to the control cluster; the control amount calculation node is used to calculate the control amount based on the simulation data.

[0086] Optionally, the access terminal may be implemented as a computer device, and the first communication node and the control amount calculation node may be implemented as different functional modules in the computer device.

[0087] It should be understood that during the simulation test of the present invention, there may be multiple access terminals that establish communication connections with the control cluster. Figure 2 Only one access terminal is used as an example for illustration, which does not constitute a limitation of the present invention. For example, for a large-scale unmanned system simulation test, multiple formations are divided, each formation includes multiple unmanned systems, and there are multiple access terminals connected to the control cluster, each access terminal corresponds to a formation, and is used to obtain simulation data of each unmanned system in the formation and calculate the control amount of each unmanned system in the formation.

[0088] The control cluster is used to manage simulation task requests from the access terminal and assign simulation tasks to the computing cluster; optionally, it is also responsible for data forwarding between the access terminal and the control cluster. The control cluster includes a second communication node, a task management node and a plurality of task allocation nodes. Among them, the second communication node is used to establish communication connections with the access terminal and the computing cluster respectively, and receive simulation task requests. Optionally, the second communication node is also used to forward simulation data from the computing cluster to the access terminal, and to forward control quantities from the access terminal to the computing cluster; the task management node is used to manage the simulation task request list based on the simulation task request; the task allocation node is used to allocate simulation tasks to corresponding computing nodes based on the simulation task request list. Among them, the simulation tasks are allocated in a preset order, and the preset order is from front to back: computing nodes that are not currently allocated to simulation tasks, computing nodes that have been allocated with a number of simulation tasks less than the recommended allocation number, and computing nodes that have been allocated with a number of simulation tasks less than the maximum allocation number.

[0089] Optionally, the control cluster can be implemented as multiple computer devices, the second communication node and the task management node can be implemented as different functional modules in one of the computer devices, and each task allocation node can be implemented as a computer device respectively; or, the second communication node, the task management node and each task allocation node are each implemented as a computer device.

[0090] The computing cluster is used to execute the simulation task of the unmanned system and generate the simulation data of the unmanned system. The computing cluster includes multiple computing nodes, each of which includes a third communication node and a simulation computing node. The third communication node is used to establish a communication connection with the control cluster. Optionally, the third communication node is also used to receive the control amount forwarded by the control cluster and send simulation data to the control cluster; the simulation computing node is used to perform simulation calculation according to the assigned simulation task, specifically: generate a simulation node corresponding to the unmanned system, control the simulation node according to the control amount, and calculate the simulation data.

[0091] Optionally, the computing cluster may be implemented as multiple computer devices, each computing node may be implemented as a computer device, and the third communication node and the simulation computing node in each computing node may be implemented as different functional modules in the corresponding computer device.

[0092] Optionally, the correspondence between each computing node and each task allocation node is a many-to-many relationship.

[0093] Optionally, each computing node is only responsible for the simulation task of one type of unmanned system, and the number of computing nodes in the computing cluster that are responsible for the simulation task of the same type of unmanned system is greater than or equal to 2.

[0094] The system provided in the embodiment of the present invention can be used to execute the method provided in the first embodiment of the present invention, and has the corresponding functions and beneficial effects of the execution method.

[0095] In summary, the embodiments of the present invention have at least the following innovative features and beneficial effects:

[0096] (1) Add a control cluster, and perform multiple rounds of progressive task allocation to the computing cluster through the control cluster. Beneficial effect: In the present invention, the task allocation node in the control cluster preferentially allocates simulation tasks to computing nodes in the computing cluster that are not currently allocated to simulation tasks, and secondly allocates simulation tasks to computing nodes that are currently allocated to simulation tasks but the number of simulation tasks allocated is less than the recommended allocation number, and finally allocates simulation tasks to computing nodes that are currently allocated to simulation tasks but the number of simulation tasks allocated is less than the maximum allocation number. Through such three rounds of progressive allocation process, the processing overhead of each computing node is effectively balanced, considering the current simulation status and hardware resource occupancy rate of each computing node in the computing cluster, and realizing the reasonable allocation of simulation tasks, which is helpful to promote the smooth operation of large-scale unmanned system simulation testing.

[0097] (2) The task management node in the control cluster maintains a simulation task list, and the task allocation node reads the simulation task request from the simulation task list in sequence. Beneficial effect: In the present invention, the simulation task list is maintained by the task management node. The longer the waiting time of the simulation task request that has not been received, the more the number of releases after being received, the more the number of releases under the same waiting time, and the longer the waiting time under the same number of releases, the higher the sorting priority of the simulation task request in the simulation task list, so that when the task allocation node reads the simulation task request in sequence, it has priority to read the simulation task request, which helps to balance the response time of each simulation task request and avoid individual simulation task requests not being read and received for a long time. In addition, each task allocation node maintains a release list of simulation task requests to record the simulation task requests that are released again after being received. When reading the simulation task list later, it is directly judged not to receive these simulation task requests to avoid repeatedly receiving the same simulation task request that cannot be allocated by itself, avoid affecting the successful response of the simulation task request by other task allocation nodes, and improve the response success rate of the simulation task request.

[0098] (3) The simulation tasks of the same type of unmanned system are in charge of at least two computing nodes, and there is a many-to-many correspondence between the task allocation nodes and the computing nodes. Beneficial effects: In the present invention, in order to facilitate the management and allocation of simulation tasks and improve the efficiency of simulation task allocation, one computing node is only responsible for the simulation tasks of one type of unmanned system; but the simulation tasks of the same type of unmanned system are in charge of at least two computing nodes, which can avoid the normal operation of the simulation test being affected by the disconnection or crash of individual computing nodes. In addition, the present invention sets a plurality of task allocation nodes in the control cluster, and the simulation task allocation of one computing node is managed by at least two task allocation nodes, and one task allocation node is responsible for managing the simulation task allocation of at least two computing nodes, which effectively avoids the disconnection or crash of a single task allocation node or a single computing node affecting the normal operation of task allocation and simulation testing.

Claims

1. A simulation task allocation method for unmanned system simulation test, characterized in that: The method is implemented based on a system including an access terminal, a control cluster and a computing cluster, and the method includes: Establish the corresponding relationship between each computing node in the computing cluster and each task allocation node in the control cluster; The access end sends a simulation task request to the control cluster; The task management node in the control cluster manages the simulation task request list based on the simulation task request; Each task allocation node in the control cluster allocates simulation tasks to the corresponding computing nodes based on the simulation task request list; wherein the simulation tasks are allocated in a preset order, and the preset order is from front to back: computing nodes that are not currently allocated to simulation tasks, computing nodes that have been allocated with a number of simulation tasks less than the recommended allocation number, and computing nodes that have been allocated with a number of simulation tasks less than the maximum allocation number.

2. The simulation task allocation method for unmanned system simulation test according to claim 1 is characterized in that: The corresponding relationship between each computing node and each task allocation node is a many-to-many relationship.

3. The simulation task allocation method for unmanned system simulation test according to claim 1 is characterized in that: Each computing node is only responsible for the simulation task of one type of unmanned system, and the number of computing nodes in the computing cluster that are responsible for the simulation task of the same type of unmanned system is greater than or equal to 2.

4. The simulation task allocation method for unmanned system simulation test according to claim 1 is characterized in that: The task management node in the control cluster manages the simulation task request list based on the simulation task request, including: The task management node determines the sorting priority of the simulation task request based on at least one of the waiting time during which the simulation task request is not received, the number of times the simulation task request is released again after being received, and the waiting time during which the simulation task request is not received after being released again; When a simulation task request is received, a simulation task request is received, or a simulation task request is released, the task management node updates the simulation task request list according to the sorting priority of the simulation task request.

5. The simulation task allocation method for unmanned system simulation test according to claim 1 is characterized in that: Each task allocation node in the control cluster allocates simulation tasks to corresponding computing nodes based on the simulation task request list, including: The task allocation node reads the simulation task request from the simulation task request list; The task allocation node determines whether to receive the current simulation task request based on the simulation task type and / or the number of simulation tasks; In the case of determining to receive the current simulation task request, the task allocation node obtains k computing nodes and the allocation priorities of the k computing nodes; wherein the k computing nodes refer to computing nodes among the multiple computing nodes corresponding to the task allocation node, the unmanned system type corresponding to the responsible simulation task being the same as the unmanned system type included in the simulation task request; In the tth round of allocation, the task allocation node sends a task allocation request RQ to the computing node i according to the allocation priority. i ; Wherein, t is a positive integer less than or equal to 3; Based on the current simulation state and the value of t, the computing node i sends a task allocation response RE to the task allocation node i ; Among them, the current simulation state of the computing node includes: whether it has been currently assigned to a simulation task, the recommended number of simulation tasks to be assigned, and the maximum number of simulation tasks to be assigned; The task allocation node responds to the task allocation RE based on the task allocation response i , calculate the number of simulation tasks to be assigned, and when the number of simulation tasks is not zero, keep t unchanged and set i=i+1, and send a task assignment request RQ to computing node i according to the assignment priority i The step of is executed again until the number of simulation tasks is zero or until i+1 is greater than k; If the number of simulation tasks to be assigned to the task assignment node is still not zero when i+1 is greater than k, then let t=t+1 and i=1, and send a task assignment request RQ to the computing node i according to the assignment priority. i The steps are executed again starting from the step i+1 until the number of simulation tasks is zero or until i+1 is greater than k.

6. The simulation task allocation method for unmanned system simulation test according to claim 5 is characterized in that: The computing node i sends a task allocation response RE to the task allocation node based on the current simulation state and the value of t. i ,include: In the case where t is equal to 1, if the computing node i has been assigned to the simulation task, a first task assignment response is sent to the task assignment node, and the first task assignment response is used to indicate the rejection of the task assignment; if the computing node i has not been assigned to the simulation task, a second task assignment response is sent to the task assignment node, and the second task assignment response is used to indicate the acceptance of the task assignment and the number of simulation tasks currently accepted for assignment; When t is equal to 2, if the number of simulation tasks currently assigned to computing node i is equal to the recommended number of simulation tasks that computing node i is responsible for, a first task assignment response is sent to the task assignment node, and the first task assignment response is used to indicate the rejection of task assignment; if the number of simulation tasks currently assigned to computing node i is less than the recommended number of simulation tasks that computing node i is responsible for, a second task assignment response is sent to the task assignment node, and the second task assignment response is used to indicate the acceptance of task assignment and the number of simulation tasks currently accepted for assignment; When t is equal to 3, if the number of simulation tasks currently assigned to computing node i is equal to the maximum number of simulation tasks that computing node i is responsible for, a first task assignment response is sent to the task assignment node, and the first task assignment response is used to indicate the rejection of task assignment; if the number of simulation tasks currently assigned to computing node i is less than the maximum number of simulation tasks that computing node i is responsible for, a second task assignment response is sent to the task assignment node, and the second task assignment response is used to indicate the acceptance of task assignment and the number of simulation tasks currently accepted for assignment.

7. The simulation task allocation method for unmanned system simulation test according to claim 5 is characterized in that: The method for determining the recommended allocation number of simulation tasks and the maximum allocation number of simulation tasks includes: Obtaining hardware indicator reference parameters and their corresponding reference quantity configurations; the hardware indicator reference parameters refer to hardware indicator parameters of reference computer equipment; Determine the quantity configuration ratio based on the hardware index reference parameter and the hardware index target parameter; the hardware index target parameter refers to the hardware index parameter of the computer device where the computing node is located; Based on the quantity configuration ratio and the reference quantity configuration, the recommended allocation quantity and the maximum allocation quantity are obtained.

8. A simulation task allocation system for unmanned system simulation testing, characterized in that: The system includes an access terminal, a control cluster and a computing cluster, wherein: The access terminal includes a first communication node, the first communication node is used to establish a communication connection with the control cluster and send a simulation task request to the control cluster; The control cluster includes a second communication node, a task management node and a plurality of task allocation nodes, wherein the second communication node is used to establish a communication connection with the access terminal and the computing cluster, the task management node is used to manage a simulation task request list based on the simulation task request, and the task allocation node is used to allocate simulation tasks to corresponding computing nodes based on the simulation task request list; wherein the simulation tasks are allocated in a preset order, and the preset order is from front to back: computing nodes that are not currently allocated to simulation tasks, computing nodes that have been allocated with a number of simulation tasks less than the recommended allocation number, and computing nodes that have been allocated with a number of simulation tasks less than the maximum allocation number; The computing cluster includes a plurality of computing nodes, each computing node includes a third communication node and a simulation computing node, the third communication node is used to establish a communication connection with the control cluster, and the simulation computing node is used to perform simulation computing according to the assigned simulation tasks.

9. The simulation task allocation system for unmanned system simulation test according to claim 8, characterized in that: The corresponding relationship between each computing node and each task allocation node is a many-to-many relationship.

10. The simulation task allocation system for unmanned system simulation test according to claim 8, characterized in that: Each computing node is only responsible for the simulation task of one type of unmanned system, and the number of computing nodes in the computing cluster that are responsible for the simulation task of the same type of unmanned system is greater than or equal to 2.

Images