Test evaluation scene construction method and system based on multi-target game

Abstract

The embodiment of the invention provides a test evaluation scene construction method and system based on a multi-target game. The method comprises the following steps: constructing a scene model; dividing an obstacle area, planning an initial path of the dynamic target by avoiding the obstacle area, dividing the initial path into path sections according to time intervals, performing obstacle area re-optimization on each path section, and generating a moving path; comparing whether the moving paths of the dynamic targets have conflicts or not, identifying dynamic target subsets with conflicts and conflict points, and generating non-conflict alternative paths of the dynamic targets at the conflict points; and establishing a cost function of the dynamic target subsets, integrating the cost functions of all the dynamic target subsets, establishing a game model, iteratively adjusting the evasion strategy of the dynamic target subsets by combining the initial evasion strategy of the dynamic target subsets until the game model converges, and generating an optimal path of the dynamic target subsets based on the optimal evasion strategy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle testing, in particular to a test evaluation scene construction method and system based on multi-target game. BACKGROUND

[0002] With the continuous progress of intelligent driving technology, efficient testing of intelligent driving technology is an important means to ensure the safety and reliability of intelligent driving systems. Intelligent driving evaluation technology constructs a simulated test environment, formulates test standards and processes, and comprehensively evaluates the performance of vehicle active safety systems, providing strong support for the further development of active safety technology.

[0003] However, in the prior art, the intelligent driving real vehicle test scene still uses single target or multiple static targets as test targets, and lacks a multi-target dynamic coordination test scene. Even in some scenes with multiple targets, the coordination between targets is not intelligent, or a large amount of cost is consumed. SUMMARY

[0004] In view of the problems in the prior art, the embodiments of the present application provide a test evaluation scene construction method and system based on multi-target game.

[0005] The embodiments of the present application provide a test evaluation scene construction method based on multi-target game, which comprises:

[0006] Receiving a test requirement of a vehicle under test, constructing a scene model based on the test requirement, wherein scene elements in the scene model include the vehicle under test, dynamic targets and static targets;

[0007] Dividing an obstacle area based on the scene elements, planning an initial path of the dynamic targets by avoiding the obstacle area, dividing the initial path into path segments according to time intervals, re-optimizing the obstacle area for each path segment, splicing the re-optimized path segments to generate a moving path of the dynamic targets;

[0008] Comparing whether there is a conflict in the moving paths of all dynamic targets at the same time, identifying a dynamic target subset with a conflict and a conflict point, and generating a non-conflict alternative path of the dynamic targets at the conflict point;

[0009] Establishing a cost function of the dynamic target subset, and establishing a game model by integrating the cost functions of all dynamic target subsets, adjusting the avoidance strategy of the dynamic target subset iteratively in combination with the initial avoidance strategy of the dynamic target subset until the game model converges, and outputting the best avoidance strategy;

[0010] Generating an optimal path of the dynamic target subset based on the best avoidance strategy.

[0011] In one embodiment, the cost function includes:

[0012] collision loss cost, operation cost and avoidance psychological cost of the dynamic target subset.

[0013] In one embodiment, the method further includes:

[0014] For the dynamic target subset i, the corresponding cost function is:

[0015]

[0016] wherein, is the total cost of the dynamic target subset i, is the path segment selected by the dynamic target subset i, is the path segment excluding other dynamic target subsets except i, is the collision loss cost, is the operation cost, is the avoidance psychological cost of the dynamic target subset j to the dynamic target subset i, and m is the number of dynamic target subsets.

[0017] In one embodiment, the method further includes:

[0018] When calculating the initial avoidance strategy, the first expected cost when choosing to avoid and the second expected cost when choosing not to avoid are calculated, and then the total expected cost is calculated;

[0019] The initial avoidance strategy is adjusted, a replicative dynamic equation of avoidance strategy change speed is determined, and when the avoidance strategy change speed tends to be stable, the optimal avoidance strategy is calculated.

[0020] In one embodiment, the method further includes:

[0021] Based on the dynamic target subset i and the dynamic target subset j, the first expected cost is calculated, and the calculation formula includes:

[0022]

[0023] wherein, is the first expected cost when the dynamic target subset i chooses to avoid, is the first expected cost when the dynamic target subset j chooses to avoid, p is the initial avoidance strategy of the dynamic target subset i, is the probability of choosing to avoid i, q is the initial avoidance strategy of the dynamic target subset j, and is the probability of choosing to avoid j, is the additional operation cost generated when the dynamic target subset i chooses to avoid, is the avoidance psychological cost when the dynamic target subset i chooses to avoid and the dynamic target subset j does not avoid, selecting an evasion for the dynamic target subset j, the evasion psychological cost when the dynamic target subset i does not evade;

[0024] calculating a second expected cost, the calculation formula comprising:

[0025]

[0026] wherein, selecting a second expected cost when the dynamic target subset i does not evade, selecting a second expected cost when the dynamic target subset j does not evade;

[0027] combining the first expected cost and the second expected cost, calculating a total expected cost, the calculation formula comprising:

[0028]

[0029] wherein, the total expected cost of the dynamic target subset i under the evasion strategy p, the total expected cost of the dynamic target subset j under the evasion strategy q;

[0030] the replication dynamic equation of the evasion strategy change speed, comprising:

[0031]

[0032]

[0033] wherein, dp / dt is the change speed of the dynamic target subset i, and dq / dt is the change speed of the dynamic target subset j;

[0034] when the evasion strategy change speed tends to be stable, i.e. F(p)=0 and F(q)=0, calculating an optimal evasion strategy, the calculation formula comprising:

[0035]

[0036] wherein, the optimal evasion strategy of the dynamic target subset i, the optimal evasion strategy of the dynamic target subset j.

[0037] In one embodiment, the method further comprises:

[0038] acquiring the pre-planned path of the vehicle to be tested, the initial point of the dynamic target and the position of the static target, and comprehensively determining the obstacle region.

[0039] In one embodiment, the method further comprises:

[0040] Based on the position change of the obstacle area in the corresponding time window, the starting point, the end point and the path shape of each path segment are re-optimized, and the optimized path segments are connected with the initial path through a Bezier curve.

[0041] Embodiments of the present application provide a test evaluation scene construction system based on multi-target game, the system comprises:

[0042] A scene module is configured to receive a test requirement of a vehicle under test, and construct a scene model based on the test requirement, wherein scene elements in the scene model include the vehicle under test, dynamic targets and static targets.

[0043] A path optimization module is configured to divide obstacle areas based on the scene elements, plan an initial path of the dynamic targets by avoiding the obstacle areas, divide the initial path into path segments according to time intervals, re-optimize the obstacle areas for each path segment, splice the re-optimized path segments, and generate a moving path of the dynamic targets.

[0044] A conflict module is configured to compare whether there is a conflict in the moving paths of all dynamic targets at the same time, identify a dynamic target subset and a conflict point where there is a conflict, and generate a non-conflict alternative path of the dynamic targets at the conflict point.

[0045] A game module is configured to establish a cost function of the dynamic target subset, integrate the cost functions of all dynamic target subsets, establish a game model, iteratively adjust an avoidance strategy of the dynamic target subset in combination with an initial avoidance strategy of the dynamic target subset, until the game model converges, and output an optimal avoidance strategy.

[0046] A generation module is configured to generate an optimal path of the dynamic target subset based on the optimal avoidance strategy.

[0047] Embodiments of the present application provide an electronic device, comprising a processor and a memory.

[0048] The processor is connected to the memory.

[0049] The memory is configured to store executable program codes.

[0050] The processor runs a program corresponding to the executable program codes by reading the executable program codes stored in the memory, so as to execute the method described in one or more embodiments.

[0051] Embodiments of the present application provide a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the above-mentioned test evaluation scene construction method based on multi-target game.

[0052] In view of the above, in one or more embodiments of the present specification, the test requirements of the vehicle under test are received, a scene model is constructed based on the test requirements, the obstacle area is divided based on the scene elements, the initial path of the dynamic target is planned to avoid the obstacle area, and the initial path is divided into path segments according to time intervals, each path segment is re-optimized for the obstacle area, the re-optimized path segments are spliced to generate the movement path of the dynamic target, it is compared whether the movement paths of all dynamic targets at the same time exist conflicts, a dynamic target subset existing conflicts and a conflict point are identified, and a non-conflict alternative path of the dynamic target at the conflict point is generated, a cost function of the dynamic target subset is established, and the cost functions of all dynamic target subsets are integrated to establish a game model, the initial avoidance strategy of the dynamic target subset is combined to iteratively adjust the avoidance strategy of the dynamic target subset until the game model converges, and the best avoidance strategy is output, and the optimal path of the dynamic target subset is generated based on the best avoidance strategy. Thus, through the intelligent cooperation of multiple target objects in the scene setting, more complex test conditions can be set, the complex real scene can be better reproduced, and the cost of the multi-target scene is also saved as much as possible. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0054] Figure 1 is a flowchart of a test evaluation scene construction method based on multi-target game provided by one embodiment of the present specification.

[0055] Figure 2 is a structural schematic diagram of a test evaluation scene construction system based on multi-target game provided by one embodiment of the present specification.

[0056] Figure 3 is a structural schematic diagram of an electronic device provided by one embodiment of the present specification. DETAILED DESCRIPTION

[0057] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that these implementations are discussed solely for the purpose of illustrating aspects of the subject matter described herein and are not a limitation of the scope, applicability, or examples set forth in the claims. Changes in the function and arrangement of elements discussed can be made without departing from the scope of the subject matter described herein. Various examples can omit, substitute, or add various procedures or components as appropriate. For instance, the methods described can be performed in an order different than that described, and / or various steps can be added, omitted, or combined. Also, features described with respect to some examples can be combined in other examples.

[0058] As used herein, the term "includes" and its variants are meant to be The term "based on" means "based, at least in part, on." The terms "one

[0059] As Figure 1 As shown in FIG. 1, the embodiment of the present application provides a test evaluation scene construction method based on multi-target game, which comprises the following steps:

[0060] In step S102, the test requirement of the vehicle under test is received, and a scene model is constructed based on the test requirement. The scene elements in the scene model include the vehicle under test, dynamic targets, and static targets.

[0061] Specifically, the test requirement of the vehicle under test is received, and each scene element in the scene model is determined according to the test requirement, including but not limited to the host vehicle (vehicle under test), the initial state (position, speed, heading) of the host vehicle, and the preset driving task (such as straight driving, left turn, lane change); dynamic targets, other dynamic targets (vehicles, pedestrians, etc.) except the host vehicle, and the dynamic parameters of each dynamic target are defined independently, including driving path, speed curve, direction, and behavior triggering condition, etc.; static targets, including road topology (such as crossroads, roundabouts), lane information, traffic signs and signal lights, and other static parameters. Thus, in the test scene model, the benchmark coordinate system and the expected behavior of the test are established, and the dynamic cooperation of multiple targets is realized.

[0062] Step S104, based on the scene element, divide the obstacle area, avoid the obstacle area to plan the initial path of the dynamic target, and divide the initial path into path segments according to time intervals, re-optimize the obstacle area for each path segment, splice the re-optimized path segments, and generate the moving path of the dynamic target.

[0063] Specifically, according to the scene elements in step S102, the initial state of the dynamic target and the static target is obtained, that is, the initial point of the dynamic target and the position of the static target, and then combined with the pre-planned path of the measured vehicle. The above position points are identified as all possible areas that hinder the movement of the dynamic target. These areas are marked as obstacle areas. The initial path of the dynamic target is planned by taking the obstacle area as a constraint condition through a path planning algorithm. The initial path is not the optimal path, but provides a globally feasible solution to avoid the obstacle area.

[0064] Further, the initial path is divided into multiple segments according to time intervals (such as every 0.1 second), and each segment corresponds to a path segment of the dynamic target in a specific time window. At the same time, according to the predicted speed of the dynamic target, the position change of the obstacle area in each time slice is predicted. For each time slice, according to the refined obstacle position (i.e. considering the obstacle state after time change), the starting point, end point and path shape of the segment are re-optimized. The optimization algorithm (such as gradient descent or local search) is used to ensure that the segment avoids obstacles. Then all the optimized segments are connected in sequence to form a continuous optimized path. When connecting the segments, a Bezier curve can be used for smoothing to reduce the sudden turn or jitter of the dynamic target in movement, making the movement more natural and more consistent with the real vehicle dynamics.

[0065] Step S106, compare whether there is a conflict between the moving paths of all dynamic targets at the same time, identify the dynamic target subset and the conflict point that exist conflict, and generate a non-conflict alternative path for the dynamic target at the conflict point.

[0066] Specifically, the smooth moving path of all dynamic targets is obtained, and based on the moving path and time information, it is detected whether there is a conflict such as intersection, overlap or insufficient safety distance between the moving paths. When a conflict is detected, the dynamic target subset and the conflict point that exist conflict are identified, and an alternative path is generated for the path segment near the conflict point. The adjustment method includes fine-tuning the segment starting point, adding a random midway point or changing the path shape. In the subsequent steps, the strategy set of the game is all the path segments (including the conflict path and the alternative path) that the dynamic target can select at the conflict point.

[0067] Step S108, a cost function of the dynamic target subset is established, and cost functions of all dynamic target subsets are integrated to establish a game model. The initial avoidance strategy of the dynamic target subset is combined, the avoidance strategy of the dynamic target subset is iteratively adjusted until the game model converges, and the optimal avoidance strategy is output.

[0068] Specifically, it is assumed that the number of dynamic targets simultaneously performing path planning at a certain moment is n, F represents a set of n dynamic targets performing path planning at a certain moment. The number of dynamic target subsets is m, represents a dynamic target subset designed for a certain specific conflict in F, the number of which is m, and m≤n. The candidate path set of each dynamic target in the dynamic target subset Fc It is assumed that the costs of each path are different; it is assumed that the decision-making processes of each dynamic target are independent. For a dynamic target i, the cost function includes the collision loss cost of the dynamic target subset, the operation cost, and the avoidance psychological cost. The corresponding function is:

[0069]

[0070] wherein, is the total cost of the dynamic target subset i, is the path segment selected by the dynamic target subset i, is the path segment of other dynamic target subsets excluding i, is the collision loss cost, is the operation cost, is the avoidance psychological cost of the dynamic target subset j to the dynamic target subset i, and m is the number of dynamic target subsets.

[0071] Among them, in the collision loss cost, when the dynamic target i collides with other dynamic targets, it can take the value of M (a large finite positive number), and vice versa. The collision detection is based on the intersection of path segments in space-time, and considers the potential conflict of the preferred path and the secondary conflict after compromise; the operation cost includes fuel / electricity consumption cost and time cost, which is related to path length, speed curve and travel time. For example, a longer path or a slower speed will increase ; the avoidance psychological cost is a quantitative value of the cumulative psychological influence of other dynamic targets on i. It may be based on the relative speed, distance or traffic rules between the targets.

[0072] The cost function is used as a game model for game solving. That is, the dynamic target selects the game of the candidate path based on the cost function, and iteratively trains the minimum expected cost in the selection strategy (avoidance (collision) strategy) of different paths. The iterative training process includes:

[0073] At the beginning of the game, each dynamic target is assigned an initial evasion strategy. The preference of dynamic target i for the evasion strategy is p (0≤p≤1), and the preference for the non-evasion strategy is 1-p; dynamic target j is similar, with preferences q and 1-q. Among them, p and q can be initialized to 0.5, simulating the "uncertainty" of the dynamic target at the beginning of the game, providing a starting point for subsequent evolution.

[0074] Based on the dynamic target subset i and the dynamic target subset j, the first expected cost of the evasion strategy is calculated, and the calculation formula includes:

[0075]

[0076] Among them, is the first expected cost of the evasion strategy selected by the dynamic target subset i, is the first expected cost of the evasion strategy selected by the dynamic target subset j, p is the initial evasion strategy of the dynamic target subset i, the probability of selecting evasion for i, q is the initial evasion strategy of the dynamic target subset j, the probability of selecting evasion for j, is the additional operating cost generated by the selection of evasion by the dynamic target subset i, is the evasion psychological cost when the dynamic target subset i selects evasion and the dynamic target subset j does not evade, is the evasion psychological cost when the dynamic target subset j selects evasion and the dynamic target subset i does not evade. In the above calculation formula, represents that in the case of j also evading (with a probability of q), the cost of i only has the basic operating cost and the evasion action cost, without the evasion psychological cost. represents that in the case of j not evading (with a probability of 1-q), the cost of i, in addition to the basic operating cost and the evasion action cost, also includes the evasion psychological cost generated by "forced yielding".

[0077] The second expected cost of the non-evasion strategy is calculated, and the calculation formula includes:

[0078]

[0079] Among them, is the second expected cost of the non-evasion strategy selected by the dynamic target subset i, is the second expected cost of the non-evasion strategy selected by the dynamic target subset j. In the above calculation formula, represents that in the case of j evading (with a probability of q), i can successfully run, and the cost only has the basic operating cost, without the evasion action cost and the evasion psychological cost. represents that in the case of j not evading (with a probability of 1-q), both parties will collide, and i will bear the collision loss.

[0080] The total expected cost is calculated by combining the first expected cost and the second expected cost, and the calculation formula includes:

[0081]

[0082] wherein, is the total expected cost of the dynamic target subset i under the evasion strategy p, is the total expected cost of the dynamic target subset j under the evasion strategy q, and the total expected cost represents the average performance of the dynamic target under the current game state.

[0083] The replication dynamic equation of the evasion strategy change speed includes:

[0084]

[0085]

[0086] wherein, dp / dt is the change speed of the dynamic target subset i, and dq / dt is the change speed of the dynamic target subset j; represents the difference between the "cost of choosing not to evade the strategy" and the "current average cost". If is greater than , it indicates that the "cost of choosing not to evade the strategy" is higher than the average level, and dp / dt tends to be negative, and the dynamic target i will reduce the probability p of choosing "evasion". Conversely, it will increase the probability p of choosing "evasion". In each round of game, according to the current p and q, the change speed of p and q is calculated through the above formula, and then the preference is updated.

[0087] The game is iterated until the strategy preference no longer changes, and the evasion strategy change speed tends to be stable, that is, F(p)=0 and F(q)=0, the system reaches the equilibrium point, and the best evasion strategy is calculated, and the calculation formula includes:

[0088]

[0089] wherein, is the best evasion strategy of the dynamic target subset i, is the best evasion strategy of the dynamic target subset j. Taking as an example, the numerator is the total opportunity cost of j choosing "evasion" (i.e. the collision loss that j may face when not evading, minus all the costs paid for evasion). The denominator is the cost difference between "evasion" and "non-evasion" of j. The ratio of accurately quantifies the probability at which the dynamic target i should choose the "evasion" strategy to be the best evasion strategy under the equilibrium state.

[0090] Step S110, generating an optimal path of the dynamic target subset based on the optimal avoidance strategy.

[0091] Specifically, according to the calculated optimal avoidance strategy, a path segment at the conflict point is selected for each dynamic target. Then the path segment is smoothly connected with the moving path of the dynamic target to determine the optimal path of the dynamic target itself. Further, the scene model is perfected, and the vehicle under test is waited to start testing.

[0092] The embodiment of the application provides a test evaluation scene construction method based on multi-target game, receives a test demand of a vehicle under test, constructs a scene model based on the test demand, divides an obstacle area based on scene elements, plans an initial path of a dynamic target by avoiding the obstacle area, divides the initial path into path segments according to time intervals, reoptimizes the obstacle area for each path segment, splices the reoptimized path segments, generates a moving path of the dynamic target, compares whether there is a conflict in the moving paths of all dynamic targets at the same moment, identifies a dynamic target subset existing conflict and a conflict point, and generates a non-conflict alternative path of the dynamic target at the conflict point, establishes a cost function of the dynamic target subset, establishes a game model by comprehensively considering the cost functions of all dynamic target subsets, iteratively adjusts an avoidance strategy of the dynamic target subset in combination with the initial avoidance strategy of the dynamic target subset, until the game model converges, and outputs an optimal avoidance strategy, and generates an optimal path of the dynamic target subset based on the optimal avoidance strategy. Therefore, a more complex test working condition is set through intelligent cooperation of multi-targets, a complex real scene can be better reproduced, and the cost of the multi-target scene is saved as much as possible.

[0093] Please refer to Figure 2 , Figure 2 is a structure schematic diagram of a test evaluation scene construction system based on multi-target game provided by the embodiment of the application. As Figure 2 indicated, the system comprises:

[0094] The scene module S202 is configured to receive a test demand of a vehicle under test, and construct a scene model based on the test demand, wherein scene elements in the scene model comprise the vehicle under test, dynamic targets and static targets.

[0095] The path optimization module S204 is configured to divide an obstacle area based on the scene elements, plan an initial path of the dynamic target by avoiding the obstacle area, divide the initial path into path segments according to time intervals, reoptimize the obstacle area for each path segment, splice the reoptimized path segments, and generate a moving path of the dynamic target.

[0096] The conflict module S206 is configured to compare whether there is a conflict in the moving paths of all dynamic targets at the same time, identify a dynamic target subset with a conflict and a conflict point, and generate a non-conflict alternative path of the dynamic target at the conflict point.

[0097] The game module S208 is configured to establish a cost function of the dynamic target subset, establish a game model by comprehensively considering the cost functions of all dynamic target subsets, combine an initial avoidance strategy of the dynamic target subset, iteratively adjust the avoidance strategy of the dynamic target subset until the game model converges, and output an optimal avoidance strategy.

[0098] The generation module S210 is configured to generate an optimal path of the dynamic target subset based on the optimal avoidance strategy.

[0099] Those skilled in the art can clearly understand that the technical solutions of the embodiments of the present application can be implemented by means of software and / or hardware. The "unit" and "module" in the specification refer to software and / or hardware that can independently complete or cooperate with other components to complete a specific function, and the hardware may, for example, be a field programmable gate array (FPGA), an integrated circuit (IC), and the like.

[0100] The various processing units and / or modules of the embodiments of the present application can be implemented by means of an analog circuit that implements the functions described in the embodiments of the present application, or can be implemented by means of software that implements the functions described in the embodiments of the present application.

[0101] Referring to Figure 3 , a structural schematic diagram of an electronic device related to the embodiments of the present application is shown, which can be used to implement the method in the embodiments shown in Figure 1 . As shown in Figure 3 , the electronic device 300 can include at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302.

[0102] The communication bus 302 is configured to realize the connection and communication between the components.

[0103] The user interface 303 can include a display screen (Display) and a camera (Camera), and the optional user interface 303 can further include a standard wired interface and a wireless interface.

[0104] The network interface 304 can optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0105] The processor 301 can include one or more processing cores. The processor 301 connects various parts within the entire electronic device 300 by various interfaces and lines, and performs various functions of the electronic device 300 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Alternatively, the processor 301 can be implemented in at least one of a hardware form of a digital signal processing (DSP), a field-programmable gate array (FPGA), a programmable logic array (PLA). The processor 301 can be integrated with a combination of one or several of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes an operating system, a user interface, and an application program; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; and the modem is used for processing wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 301, but can be realized by a separate chip.

[0106] The memory 305 can include a random access memory (RAM) and a read-only memory (ROM). Alternatively, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 can include a program storage area and a data storage area, wherein the program storage area can store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 305 can alternatively be at least one storage device located away from the aforementioned processor 301. As shown in the figure, the memory 305 as a computer storage medium can include an operating system, a network communication module, a user interface module, and program instructions. Figure 3 As shown in the figure, the memory 305 as a computer storage medium can include an operating system, a network communication module, a user interface module, and program instructions.

[0107] In Figure 3The electronic device 300 shown, the user interface 303 is mainly used for providing an interface for user input, obtaining user input data; and the processor 301 can be used to call the interactive application program based on image generation stored in the memory 305, and specifically perform the following operations: receiving a test requirement of a vehicle to be tested, constructing a scene model based on the test requirement; dividing an obstacle area based on a scene element, planning an initial path of a dynamic target in the obstacle area, and dividing the initial path into path segments according to a time interval, re-optimizing the obstacle area for each path segment, splicing the re-optimized path segments, and generating a moving path of the dynamic target; comparing whether the moving paths of all dynamic targets at the same time exist conflicts, identifying a dynamic target subset existing conflicts and a conflict point, and generating a non-conflict alternative path of the dynamic target at the conflict point; establishing a cost function of the dynamic target subset, and establishing a game model by comprehensively considering the cost functions of all dynamic target subsets, combining an initial avoidance strategy of the dynamic target subset, iteratively adjusting the avoidance strategy of the dynamic target subset, until the game model converges, and outputting an optimal avoidance strategy; and generating an optimal path of the dynamic target subset based on the optimal avoidance strategy.

[0108] The application also provides a computer readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps of the above method. The computer readable storage medium can include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0109] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all described as a series of action combinations, but those skilled in the art should know that the application is not limited to the action sequence described, because according to the application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily necessary for the application.

[0110] In the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0111] In several embodiments provided in the present application, it should be understood that the disclosed apparatus can be implemented in other manners. For example, the division of the apparatus embodiments described above is merely illustrative, and the division of the units can be changed according to actual needs. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0112] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0113] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can be physically present separately, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0114] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable memory. Based on this understanding, the technical solutions of the present application essentially or the part of the prior art that contributes to the technical solutions or all or part of the technical solutions can be embodied in the form of a software product, which is stored in a memory and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: a U disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0115] A person of ordinary skill in the art can understand that all or part of the steps of the various methods of the above embodiments can be completed by a program instructing relevant hardware, and the program can be stored in a computer readable memory, which can include a flash disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, etc.

[0116] The above described embodiments of the present description have been described. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order and still achieve desirable results. Additionally, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In certain implementations, multitasking and parallel processing can be advantageous.

Claims

1. A method for constructing a test evaluation scenario based on multi-objective game theory, the method comprising: Receive the test requirements of the vehicle under test, and construct a scenario model based on the test requirements. The scenario elements in the scenario model include the vehicle under test, dynamic targets, and static targets. Based on the scene elements, obstacle areas are divided, the initial path of the dynamic target is planned to avoid the obstacle areas, and the initial path is divided into path segments according to time intervals. The obstacle areas of each path segment are re-optimized, and the re-optimized path segments are spliced ​​together to generate the movement path of the dynamic target. Compare the movement paths of all dynamic targets at the same moment to see if there is a conflict, identify the subset of dynamic targets with conflict and the conflict points, and generate non-conflicting alternative paths for dynamic targets at the conflict points. Establish the cost function of the dynamic target subset, and combine the cost functions of all dynamic target subsets to establish a game model. Combine the initial avoidance strategy of the dynamic target subset, iteratively adjust the avoidance strategy of the dynamic target subset until the game model converges and outputs the optimal avoidance strategy. Based on the optimal avoidance strategy, the optimal path for the dynamic target subset is generated.

2. The method according to claim 1, characterized in that, The cost function includes: The collision loss cost, operational cost, and psychological cost of avoiding dynamic target subsets.

3. The method according to claim 2, characterized in that, The method further includes: For the dynamic objective subset i, the corresponding cost function is: ， in, The total cost of the dynamic target subset i. The path segment selected for the dynamic target subset i. To exclude path segments from other dynamic target subsets besides i, To cover collision damage costs, For operating costs, Let be the psychological cost of avoiding dynamic target subset j for dynamic target subset i, and m be the number of dynamic target subsets.

4. The method according to claim 3, characterized in that, The iterative adjustment of the avoidance strategy for the dynamic target subset, based on the initial avoidance strategy, includes: When calculating the initial avoidance strategy, the first expected cost of choosing avoidance and the second expected cost of choosing not to avoid are used to calculate the total expected cost. Adjust the initial avoidance strategy, determine the replication dynamic equation of the rate of change of the avoidance strategy, and calculate the optimal avoidance strategy when the rate of change of the avoidance strategy tends to stabilize.

5. The method according to claim 4, characterized in that, The method further includes: Based on dynamic target subsets i and j, the first expected cost is calculated using the following formula: ， in, The first expected cost when selecting an evasion option for a dynamic subset i. Let p be the initial expected cost of choosing avoidance for a dynamic target subset j, p be the initial avoidance policy for dynamic target subset i, q be the probability of choosing avoidance for i, and q be the initial avoidance policy for dynamic target subset j, q be the probability of choosing avoidance for j. To select the additional operating costs to avoid for the dynamic target subset i. Choose to avoid dynamic target subset i, and calculate the psychological cost of avoiding dynamic target subset j when not avoiding it. Choose to avoid dynamic target subset j, and the psychological cost of avoiding dynamic target subset i when it is not avoided; The second expected cost is calculated using the following formula: ， in, For the dynamic target subset i, select the second expected cost when not avoiding it. The second expected cost when choosing not to avoid a dynamic target subset j; Combining the first expected cost and the second expected cost, the total expected cost is calculated using the following formula: ， in, Let be the total expected cost of a dynamic target subset i under avoidance strategy p. Let be the total expected cost of the dynamic target subset j under the avoidance strategy q; The replication dynamic equation for the rate of change of the evasion strategy includes: ， ， Where dp / dt is the rate of change of the dynamic target subset i, and dq / dt is the rate of change of the dynamic target subset j; When the rate of change of the avoidance strategy tends to stabilize, i.e., F(p)=0 and F(q)=0, the optimal avoidance strategy is calculated, and the calculation formula includes: ， in, The optimal avoidance strategy for the dynamic target subset i. The optimal avoidance strategy for the dynamic target subset j.

6. The method according to claim 1, characterized in that, The method further includes: The pre-planned path of the vehicle under test, the initial point of the dynamic target, and the position of the static target are obtained, and the obstacle area is determined by combining these information.

7. The method according to claim 1, characterized in that, The process of re-optimizing the obstacle region for each path segment and then concatenating the re-optimized path segments includes: Based on the positional changes of obstacle areas within the corresponding time window, the starting point, ending point, and path shape of each path segment are re-optimized, and the optimized path segments are connected to the initial path using Bézier curves.

8. A test and evaluation scenario construction system based on multi-objective game theory, characterized in that, The system includes; The scenario module is used to receive the test requirements of the vehicle under test and construct a scenario model based on the test requirements. The scenario elements in the scenario model include the vehicle under test, dynamic targets, and static targets. The path optimization module is used to divide the obstacle area based on the scene elements, plan the initial path of the dynamic target by avoiding the obstacle area, divide the initial path into path segments according to time intervals, re-optimize the obstacle area for each path segment, and splice the re-optimized path segments to generate the movement path of the dynamic target. The conflict module is used to compare whether there is a conflict in the movement paths of all dynamic targets at the same time, identify the subset of dynamic targets with conflicts and the conflict points, and generate non-conflict alternative paths for dynamic targets at the conflict points. The game theory module is used to establish the cost function of the dynamic target subset, and to integrate the cost functions of all dynamic target subsets to establish a game model. Iteratively adjusts the avoidance strategy of the dynamic target subset in combination with the initial avoidance strategy of the dynamic target subset until the game model converges and outputs the optimal avoidance strategy. The generation module is used to generate the optimal path for the dynamic target subset based on the optimal avoidance strategy.

9. An electronic device, comprising a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code stored in the memory to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in any one of claims 1-7.

Images