Design support device and design support method

Abstract

Provided is a design support device capable of identifying a control parameter to be changed in response to a request in a free description format and facilitating setting.　This design support device comprises: an input unit that receives a request related to the behavior of a vehicle and described in natural language; a module selection unit that, on the basis of a first processing result obtained by natural language processing, selects a software module related to the behavior of the vehicle included in the request as a target module, with respect to the request and module description information which describes, in natural language, functions of software modules obtained by dividing vehicle control software of the vehicle for each function; a reference parameter extraction unit that extracts a reference parameter, which is a control parameter of the vehicle referenced by the target module; a parameter selection unit that, on the basis of a second processing result obtained by the natural language processing, selects a target parameter related to the behavior of the vehicle included in the request, with respect to the request and parameter description information which describes the function of the reference parameter in natural language; and an output unit that outputs the target parameter.

Description

Design Support Device and Design Support Method

[0001] The present invention relates to a design support device and a design support method.

[0002] In recent years, due to the diversification of driver preferences, the software-defined vehicle (SDV) transformation of automobiles has been accelerating. Until now, the software of automobiles has been developed in accordance with hardware such as an electronic control unit (ECU). However, with the progress of SDV transformation, there is no longer a need to develop software based on specific hardware, and functions will be realized mainly by software. Therefore, it becomes possible to provide software according to various driver preferences. Also, by individually adjusting the control parameters of the software, customization for specific applications and personalization for specific drivers become easier.

[0003] On the other hand, as the role of software becomes larger, the scale of software and the number of control parameters are increasing. Therefore, if each control parameter is set manually, an enormous amount of man-hours is required, and a mechanism that can easily set these is desired.

[0004] Conventionally, as an invention of this type, there is one described in Japanese Patent Application Laid-Open No. 2017-206239 (Patent Document 1). Patent Document 1 describes "a method performed by a vehicle including an ADAS system (advanced driver assistance system) and a communication unit, the receiving step in which the communication unit of the vehicle receives a wireless message via a wireless network, and the changing step in which a processor of the vehicle changes the control parameters of the ADAS system, the wireless message including optimization setting value data describing how to change the operation of the ADAS system based on the preferences of a first user who reserved the use of the vehicle regarding the operation of the ADAS system, and in the changing step, based on the optimization setting value data, the control parameters of the ADAS system are changed so that the operation of the ADAS system matches the preferences of the first user" (Claim 1).

[0005] Japanese Patent Publication No. 2017-206239

[0006] However, in order to map user settings to software control parameters, it is necessary to pre-define the items that users can set, which reduces the flexibility of customization and personalization. On the other hand, if the system is not pre-definitely limited in the items that users can set, and attempts are made to flexibly respond to various user requests defined by free-form natural language descriptions, it becomes difficult to identify the control parameters that need to be adjusted.

[0007] This invention has been made in view of the above circumstances, and its purpose is to provide a design support device that can identify control parameters to be changed in response to free-form user requests and facilitate their setting.

[0008] To achieve the above objective, the design support device of the present invention comprises: an input unit that receives requests relating to vehicle behavior described in natural language; a module selection unit that, based on a first processing result obtained by natural language processing, selects software modules related to vehicle behavior included in the requests as target modules, based on module description information and requests that describe the functions of software modules into which the vehicle's control software is divided by function, based on the requests; a reference parameter extraction unit that extracts reference parameters, which are vehicle control parameters referenced by the target modules; a parameter selection unit that, based on a second processing result obtained by natural language processing, selects target parameters related to vehicle behavior included in the requests, based on parameter description information and requests that describe the functions of the reference parameters in natural language; and an output unit that outputs the target parameters.

[0009] According to the present invention, it is possible to identify the control parameters to be changed in response to a free-form description request and to facilitate their setting. Problems, configurations, and effects other than those described above will be clarified by the following description of embodiments.

[0010] This is a diagram showing the schematic configuration of the design support device in the first embodiment. This is the hardware configuration of the design support server in the first embodiment. This is a functional block diagram of the design support operation in the first embodiment. This is a flowchart showing the design support operation in the first embodiment. This is an example of the data structure of the module storage unit in the first embodiment. This is an example of the data structure of the reference storage unit in the first embodiment. This is an example of the data structure of the parameter storage unit in the first embodiment. This is an example of the data structure of the setting condition storage unit in the first embodiment. This is an example of the data structure of the setting value storage unit in the first embodiment. This is an example of the data structure of the setting value storage unit in the first embodiment. This is the hardware configuration of the design support server in the second embodiment. This is a functional block diagram of the design support operation in the second embodiment. This is a flowchart showing the design support operation in the second embodiment. This is an example of the data structure of the setting history storage unit in the second embodiment. This is an example of the data structure of the scenario storage unit in the second embodiment.

[0011] The embodiments for carrying out the present invention will be described below with reference to the figures. In the embodiments described below, the present invention is explained using the example of its application to a design support device for the automatic driving control of an automobile as the vehicle, but the present invention can also be applied to design support devices for the control of hybrid vehicles, electric vehicles, transport vehicles that travel in warehouses, and construction vehicles that travel on construction sites.

[0012] Figure 1A shows a schematic configuration of the design support device 1 in the first embodiment. The dashed arrows in Figure 1 indicate the signal flow.

[0013] The design support device 1 includes a design support server 100 that adjusts control parameters affecting the behavior of the vehicle control software for a vehicle 3 that is autonomously driving or driving with driver assistance based on the execution results of the vehicle control software, and a design support server terminal 100A which is a user interface between the designer 2, who is a user of the design support device 1, and the design support server 100.

[0014] The design support server 100 consists of a CPU, GPU (Graphics Processing Unit), memory, hard disk drive (HDD), etc. It executes a design support program to adjust the control parameters of the vehicle control software of the vehicle 2, and transmits the adjusted control parameters to the vehicle 2 via wireless communication such as Wi-Fi, LTE (Long Term Evolution), 5G (Generation), or wired communication via a communication cable (not shown).

[0015] The design support server terminal 100A consists of a display device such as a screen and an input device such as a keyboard and mouse. It receives requests from the designer 2, who is the user of the design support device 1, in a free-text format via the keyboard and transmits them to the design support server 100. It also displays the control parameters adjusted by the design support server 100 on the screen or other display.

[0016] Vehicle 3 consists of measuring devices 31 such as cameras and LiDAR (Light Detection and Ranging) for measuring the driving environment, vehicle control software 32 for performing various vehicle controls, an ECU 33 for executing the vehicle control software 32, acceleration and deceleration mechanisms 34 such as motors and brakes for accelerating and decelerating vehicle 3 based on the execution results of the vehicle control software 32, and a steering mechanism 35 for steering vehicle 3 based on the execution results of the vehicle control software 32. Furthermore, the vehicle control software 32 consists of multiple software modules _1 to n (321 to 32n) divided according to function.

[0017] Figure 1B shows an example of the hardware configuration of the design support server 100, which consists of a main memory device 110 such as memory, an auxiliary storage device 120 such as an HDD, a processing unit 130 such as a CPU / GPU, and a Large Language Model (LLM) 140. The LLM 140 is not limited to a large language model; it can be any device that performs natural language processing, such as a general language model or a device that performs morphological analysis.

[0018] The main memory 110 stores programs that execute the functions of each of the following parts: the input unit 111, the module selection unit 112, the reference parameter extraction unit 113, the parameter selection unit 114, the output unit 115, the duplicate reference warning unit 116, the duplicate modification warning unit 117, and the non-satisfaction warning unit 118. These programs are executed by the arithmetic processing unit 130, causing each part to function. The main memory 110 also stores a design support program 110A. When the arithmetic processing unit 130 executes the design support program 110A, it uses the aforementioned parts (111 to 118) and the parts of the auxiliary memory 120 to perform the design support operations described later.

[0019] The auxiliary storage device 120 includes a module storage unit 121, a reference storage unit 122, a parameter storage unit 123, a setting condition storage unit 124, and a setting value storage unit 125. The contents of each part of the auxiliary storage device 120 and the functions and operations of each part of the main storage device 110 will be explained in the series of operation flows of the design support device described later.

[0020] Figure 2 is a functional block diagram of the design support operation of the design support device 1 in the first embodiment, illustrating the processing and signal flow between each part in the hardware configuration diagram shown in Figure 1B. Figure 3 is a flowchart of the design support operation (steps S301 to S311) of the design support device 1 in the first embodiment. The design support operation of the first embodiment will be explained below mainly using Figures 2 and 3.

[0021] <Step S301> The input unit 111 receives a request for free-form text in natural language from the design support server terminal 100A, sends it to the module selection unit 112 and the parameter selection unit 114, and proceeds to step S302.

[0022] <Step S302> First, the module description information used in this step will be explained. Figure 4 is an example of the data structure of the module storage unit 121, in which the vehicle control software 32 of the vehicle 3 is divided into multiple software modules, and the names 121A of these modules and their natural language descriptions (module description information) 121B are stored in pairs.

[0023] The module selection unit 112 compares the request 201 transmitted from the input unit 111 with the module description information 121B stored in the module storage unit 121, selects the software module related to the behavior of the vehicle 3 included in the request 201 as the target module, and transmits the name 121A of the target module to the reference parameter extraction unit 113 and the duplicate reference warning unit 116, and proceeds to step S303. For example, if the request 201 is "Please keep a distance when avoiding pedestrians," the module selection unit 112 compares this request 201 with each description in the module description information 121B in Figure 4, and selects the relevant "avoidance_module" as the target module. This narrows down the software modules that require adjustment in response to free-form user requests, making the design easier.

[0024] When the module selection unit 112 compares the request 201 with the module description information 121B to select a target module, a prompt containing the request 201 and the module description information 121B is input to the large-scale language model 140 executed on the design support server 100. The large-scale language model 140 can, for example, extract keywords from both or summarize them, allowing the module selection unit 112 to select the module description information 121B corresponding to the request 201. This allows for highly accurate narrowing down of software modules that require adjustment in response to free-form user requests, thereby simplifying the design process.

[0025] <Step S303> First, the reference information used in this step will be explained. Figure 5 shows an example of the data structure of the reference storage unit 122, in which the parameters 122B that the software modules of the vehicle control software 32 of the vehicle 3 refer to are stored as reference information indicated by "○" for each name 122A of the software modules.

[0026] The reference parameter extraction unit 113 refers to the module name 122A of the reference information stored in the reference storage unit 122 that matches the target module 121A input from the module selection unit 112, extracts the reference parameter 122B which is a control parameter referenced from the target module, and sends the extracted reference parameter 122B to the parameter selection unit 114 to proceed to step S304. For example, if "avoidance_module" is selected as the target module 121A from the module selection unit 112 and sent to the reference parameter extraction unit 113, the reference parameter extraction unit 113 will extract "min_dist_avo" as the reference parameter 122B. In this way, the control parameters that require adjustment in response to the free-form user request can be narrowed down, simplifying the design.

[0027] <Step S304> If multiple target modules are selected in step S302, the reference parameter extraction process in step S303 is repeatedly performed for each of them (NO in step S304), and once the processing for all target modules is completed (YES in step S304), the process proceeds to step S305.

[0028] <Step S305> First, the parameter description information, setting condition information, and setting value information used in this step will be explained. Figure 6 is an example of the data structure of the parameter storage unit 123, where the parameter name 123A and its natural language description of its function 123B (parameter description information) are stored in pairs. Figure 7 is an example of the data structure of the setting condition storage unit 124, where the parameter name 124A and its setting condition information (initial value, tolerance range (maximum / minimum value)) 124B are stored in pairs. Figure 8A is an example of the data structure of the setting value storage unit 125, where the parameter name 125A and its past (i.e., current) setting value 125B (setting value information) are stored in pairs.

[0029] The parameter selection unit 114 inputs the request 201 transmitted from the input unit 111, the reference parameter 122B extracted by the reference parameter extraction unit 113, the parameter description information 123B stored in the parameter storage unit 123 for that parameter, the setting condition information 124B stored in the setting condition storage unit 124, and the setting value information 125B stored in the setting value storage unit 125 to the large-scale language model 140. The large-scale language model 140 selects the target parameter from the reference parameter 122B that is related to the behavior of the vehicle 3 included in the request 201, and adjusts the setting value of the target parameter to satisfy the request 201. This allows for highly accurate narrowing down of control parameters that require adjustment for free-form user requests and adjusts the setting values, thereby simplifying the design. The target parameter and setting value (114A) are then transmitted to the setting value storage unit 125, the output unit 115, the non-satisfaction warning unit 118, the duplicate reference warning unit 116, and the duplicate change warning unit 117, and the process proceeds to step S306.

[0030] <Step S306> The setting value storage unit 125 stores (overwrites) the target parameter and setting value 114A transmitted from the parameter selection unit 114 and proceeds to step S307.

[0031] Here, an example of parameter extraction and adjustment in this embodiment will be described. As shown in Figure 8A, the setting value storage unit 125 contains the names 125A and setting values ​​125B of previously adjusted control parameters, and under these conditions, other control parameters are extracted and their setting values ​​are adjusted.

[0032] For example, if request 201 is "Keep your distance when avoiding pedestrians," the large-scale language model 140 selects the relevant "min_dist_avo" as the target parameter by inputting a prompt containing this request and the parameter description information explanation 123B into the large-scale language model 140 and comparing and referencing it. Furthermore, the large-scale language model 140 selects a prompt containing request 201, the setting conditions (initial value, allowable range (maximum / minimum value)) 124B from the setting condition storage unit 124, and the setting value 125B from the setting value storage unit 125 into the large-scale language model 140 and comparing and referencing it to satisfy the request "Keep your distance," by adjusting "min_dist_avo" from "1.0" to "1.5" as shown in Figure 8B.

[0033] This allows for narrowing down the control parameters that need adjustment in response to free-form user requests, and simplifies the design process by adjusting the set values. Furthermore, the setting value storage unit 125 stores the fact that "min_dist_avo" has been adjusted to "1.5," allowing for stepwise adjustment of control parameters by referring to this information when selecting and adjusting control parameters for subsequent requests.

[0034] <Step S307> If there are no suitable module parameters corresponding to request 201, or if the target parameter 114A transmitted from the parameter selection unit 114 is empty, the non-satisfaction warning unit 118 issues a non-satisfaction warning and displays it, for example, on the screen of the design support server terminal 100A, and proceeds to step S308. This informs the designer, who is user 2, that the control parameters have not been adjusted to satisfy request 201.

[0035] <Step S308> The duplicate reference warning unit 116 compares the target module 121A transmitted from the module selection unit 112, the target parameter 114A transmitted from the parameter selection unit 114, and the reference information 122A and B stored in the reference storage unit 122. If the target parameter 114A is referenced by a software module other than the target module 121A, the unit issues a warning, displays it on the screen of the design support server terminal 100A, and proceeds to step S309. In other words, since there may be parameters that are commonly used (referenced) by multiple software modules, the user 2, the designer, can be informed that adjusting the control parameters in response to a request may affect the behavior of other software modules.

[0036] <Step S309> The duplicate change warning unit 23 issues a warning if the target parameter 114A transmitted from the parameter selection unit 114 has been adjusted multiple times (two or more times), displays it on the screen of the design support server terminal 100A, and proceeds to step S310. This informs the designer, who is user 2, that the adjustment value of the control parameter for the previous request may have been overwritten by the adjustment of the control parameter for the later request, and that the parameter adjustment value for the previous request may no longer be appropriate.

[0037] <Step S310> The output unit 115 determines whether a termination command, which is an indication of the user 2's intention to finish parameter adjustment, has been received from the design support server terminal 100A. If it is determined that a termination command has been received (YES in step S310), the process proceeds to step S311. If it is determined that a termination command has not been received (NO in step S310), that is, if there are still unprocessed requests 201, the process returns to step S301 to receive the request 201 from user 2 and repeats the processing from step S302 onwards.

[0038] <Step S311> The output unit 115 transmits the target parameters and latest settings for all processed requests 201 to the vehicle 3, and terminates the design support operation.

[0039] As described above, according to this embodiment, by using natural language processing for design support, it is possible to identify control parameters that should be changed in response to free-form text requests and adjust their values ​​to adjust the vehicle's behavior.

[0040] Figure 9 shows an example of the hardware configuration of the design support server 100' in the second embodiment of the present invention, which is a modification of some of the configurations of the first embodiment described above. The same reference numerals are used for elements that are the same as in the first embodiment, and the differences will be described below.

[0041] The main memory 110' is equipped with an evaluation unit 119 in place of the duplicate reference warning unit 116, duplicate modification warning unit 117, and non-satisfaction warning unit 118 in the first embodiment (Figure 1B), and the auxiliary memory 120' is equipped with a setting history storage unit 126 and a scenario storage unit 127 in place of the setting value storage unit 125 in the first embodiment (Figure 1B).

[0042] Figure 10 is a functional block diagram of the design support operation of the design support device 1 in the second embodiment, illustrating the processing and signal flow between each part in the hardware configuration diagram shown in Figure 9. Figure 11 is a flowchart of the design support operation (steps S1101 to S1110) of the design support device 1 in the second embodiment. The design support operation of the second embodiment will be explained below mainly using Figures 10 and 11. <Step S1101> The input unit 111 transmits the free-text request 201 sent from the design support server terminal 100A to the module selection unit 112, the parameter selection unit 114, the setting history storage unit 126, and the evaluation unit 119, and proceeds to step S1102.

[0043] <Step S1102> The module selection unit 112 compares the request 201 transmitted from the input unit 111 with the module description information 121B stored in the module storage unit 121, selects the software module related to the behavior of the vehicle 3 included in the request 201 as the target module, transmits the name 121A of the target module to the reference parameter extraction unit 113, and proceeds to step S1103. Note that, as in the first embodiment, when the module selection unit 112 compares the request 201 with the module description information 121B to select the target module, a prompt including the request 201 and the module description information 121B is input to the large language model 140 executed by the design support server 100.

[0044] <Step S1103> The reference parameter extraction unit 113 refers to the module name 122A of the reference information stored in the reference storage unit 122 that matches the target module 121A input from the module selection unit 112, extracts the reference parameter 122B, which is the control parameter referred to from the target module, and transmits the extracted reference parameter 122B to the parameter selection unit 114 and proceeds to step S1104.

[0045] <Step S1104> In step S1102 described above, when a plurality of target modules are selected, the reference parameter extraction process in step S1103 is repeatedly performed for each of them (in step S1104, "NO"), and when the processing for all the target modules is completed (in step S1104, "YES"), the process proceeds to step S1105.

[0046] <Step S1105> First, the setting history information used in this step will be described. FIG. 12A is an example of the data structure of the setting history storage unit 126, and includes the past request 126A, the name 126B of the adjusted control parameter, the setting value 126C before adjustment, and the setting value 126D after adjustment.

[0047] Parameter selection unit 114 inputs the request 201 transmitted from input unit 111, the reference parameter 122B input from reference parameter extraction unit 113, the parameter description information 123B stored in parameter storage unit 123, the setting condition information 124B stored in setting condition storage unit 124, and the installation history information 126A - D stored in setting history storage unit 126 into large - language model 140. The large - language model 140 selects the target parameter, which is the reference parameter related to the behavior of vehicle 3 included in request 201 among the reference parameters 122B, and adjusts the set value of the target parameter so as to satisfy request 201.

[0048] For example, as shown in FIG. 12A, the previous request was "Please keep a distance when avoiding pedestrians", and the parameter "min_dist_avo" was adjusted from "1.0" to "1.5". If the current request is "Please keep a greater distance when avoiding pedestrians" as shown in FIG. 12B, by comparing the current request, the previous request, and the set values before and after adjustment, in order to satisfy the request of "Please keep a greater distance", for example, the parameter "min_dist_avo" is adjusted from "1.5" to "2.0", and this adjustment value is stored in setting history storage unit 126 as will be described later. Thereby, the control parameters can be gradually readjusted for the free - description user requests, and the design can be facilitated.

[0049] Then, the target parameter and the set value (114A) are transmitted to setting history storage unit 126, output unit 115, and evaluation unit 119, and the process proceeds to step S1106.

[0050] <Step S1106> Setting history storage unit 126 stores the request 201 transmitted from input unit 111 and the target parameter and the installation value (114A) transmitted from parameter selection unit 114, and the process proceeds to step S1107.

[0051] <Step S1107> First, the scenario description information used in this step will be described. FIG. 13 is an example of the data structure of scenario storage unit 127, in which the content description 127B in natural language of various simulation programs (simulators) held by evaluation unit 119 and the scenario number 127A for identifying it are stored in pairs.

[0052] The evaluation unit 119 inputs a prompt to the large-scale language model 140 that includes the request 201 transmitted from the input unit 111 and the scenario description information (127A, B) stored in the scenario storage unit 127. The large-scale language model 140 narrows down the simulation scenarios suitable for evaluating the adjusted control parameters and proceeds to step S1108. For example, if the request 201 is "Keep more distance when avoiding pedestrians," the system compares this request 201 with each of the scenario descriptions 127B to select "Pass pedestrians on a straight road," where scenario number 127A is "2." This narrows down the simulation scenarios suitable for evaluating free-form user requests and facilitates the evaluation of the control parameter adjustment results.

[0053] <Step S1108> The evaluation unit 119 performs a control simulation using the adjusted control parameters 114A under the narrowed-down simulation scenario, and displays the simulation results on the design support server terminal 100A, which include information that allows confirmation of changes in behavior in relation to the requirement 201, such as an autonomous driving simulation, charts, histograms, and distribution diagrams for each signal (speed, acceleration, distance to pedestrians, etc.), and proceeds to step S1109.

[0054] <Step S1109> The output unit 115 determines whether a termination command has been received from the design support server terminal 100A. If it is determined that a termination command has been received (YES in step S1109), the process proceeds to step S1110. If it is determined that a termination command has not been received (NO in step S1109), the process returns to step S1101 to receive the next request 201 from the user and repeats the processing from step S1102 onwards.

[0055] <Step S1110> The output unit 115 transmits the target parameters and latest settings for all processed requests 201 to the vehicle 3, and terminates the design support operation.

[0056] As described above, according to this embodiment, by using natural language processing for design support, it is possible to identify control parameters that should be changed in response to free-form text requests, adjust their values ​​to adjust the vehicle's behavior, and select an appropriate simulation scenario to verify the validity of those parameter values.

[0057] Furthermore, each of the above embodiments can be modified as appropriate to the extent possible, and components of one embodiment can be incorporated into another embodiment. For example, the LLM 140 may be located in an external cloud or the like, and does not have to be specifically trained for the design support device of this embodiment.

[0058] Furthermore, in each of the above embodiments, the output unit 115 selects the target parameter in response to the request, adjusts its setting value, and transmits it to the vehicle 3. However, it is also possible to transmit only the selected target parameter to the design support server terminal 100A, and have the user 2 adjust its setting value and then transmit it to the vehicle 3.

[0059] Furthermore, by providing the vehicle 3 with a request receiving unit that receives requests for free-form text and transmits them to the design support server 100 (100'), control parameters can be adjusted in response to the free-form text requests from the driver of the vehicle 3, thereby adjusting the vehicle's behavior to match the driver's preferences.

[0060] Furthermore, the duplicate reference warning unit 116, the duplicate modification warning unit 117, and the non-satisfaction warning unit 118 of Example 1 may be incorporated into the configuration of Example 2 as needed.

[0061] 1: Design support device 2: User 3: Vehicle 100: Design support server 100A: Design support server terminal 110: Main memory 111: Input unit 112: Module selection unit 113: Reference parameter extraction unit 114: Parameter selection unit 115: Output unit 116: Duplicate reference warning unit 117: Duplicate change warning unit 118: Non-satisfaction warning unit 110A: Design support program 120: Auxiliary memory unit 121: Module storage unit 122: Reference storage unit 123: Parameter storage unit 124: Setting condition storage unit 125: Setting value storage unit 130: Arithmetic processing unit 140: Large-scale language model

Claims

1. A design support device comprising: an input unit that receives a request relating to the behavior of a vehicle described in natural language; a module selection unit that selects a software module related to the behavior of the vehicle included in the request as a target module based on a first processing result obtained by natural language processing of module description information that describes the functions of software modules into which the vehicle control software of the vehicle is divided by function, and the request; a reference parameter extraction unit that extracts reference parameters which are control parameters of the vehicle referenced by the target module; a parameter selection unit that selects a target parameter related to the behavior of the vehicle included in the request based on a second processing result obtained by natural language processing of parameter description information that describes the functions of the reference parameter in natural language and the request; and an output unit that outputs the target parameter.

2. A design support device according to claim 1, wherein the parameter selection unit sets the setting value of the target parameter based on a second processing result which further includes the processing result of setting condition information including the initial value and tolerance range of the target parameter and the request.

3. A design support device according to claim 2, wherein the parameter selection unit obtains a setting value for one target parameter that has been set in the past from a storage device, and sets a setting value for another target parameter under the condition that the setting value has been set for the one target parameter.

4. A design support device according to claim 2, wherein the parameter selection unit, if a setting value for the same target parameter as a target parameter associated with a second request received by the input unit has been set in the past, obtains setting history information from a storage device, which includes the setting value for the target parameter set in the past and the first request corresponding to said setting value, and sets the setting value for the target parameter based on the setting history information for the target parameter and the second request, which further includes the result obtained by natural language processing.

5. A design support device according to claim 1, wherein the module selection unit acquires the first processing result output in response to inputting a prompt including the request and the module description information to a language model.

6. A design support device according to claim 1, wherein the parameter selection unit acquires the second processing result output in response to inputting a prompt including the request and the parameter description information to a language model.

7. A design support device according to claim 1, comprising: a non-satisfaction warning unit that issues a non-satisfaction warning when the target parameter does not exist; 8. A design support device according to claim 1, comprising a duplicate reference warning unit that issues a duplicate reference warning when the target parameter is referenced from a module other than the target module.

9. A design support device according to claim 3, comprising a duplicate change warning unit that issues a duplicate change warning when the same parameter is changed multiple times.

10. A design support device according to claim 2, comprising an evaluation unit for evaluating the behavior of the vehicle control software with respect to the set value of the target parameter.

11. A design support device according to claim 10, wherein the evaluation unit selects the evaluation scenario based on a third processing result obtained by natural language processing in response to scenario description information indicating the content of the evaluation scenario of the vehicle control software and the request.

12. A design support device according to claim 11, wherein the evaluation unit acquires a third processing result that is output in response to inputting a prompt including the request and the scenario description information to a language model.

13. A design support device comprising: an input unit that receives requests relating to the behavior of a vehicle described in natural language; a module storage unit that stores module description information in natural language that explains the functions of software modules into which the vehicle's control software is divided by function; a parameter storage unit that stores parameter description information in natural language that explains the functions of the vehicle's control parameters referenced by the software modules related to the vehicle's behavior included in the request; and an output unit that outputs the control parameters to which the parameter description information relating to the request is associated, from among the control parameters referenced by the software modules to which the module description information relating to the request is associated.

14. A design support method in a design support device, wherein the design support device comprises an input unit, a module selection unit, a reference parameter extraction unit, a parameter selection unit, and an output unit; the input unit receives a request relating to the behavior of a vehicle, described in natural language; the module selection unit selects a software module related to the behavior of the vehicle included in the request as a target module based on a first processing result obtained by natural language processing of the request and module description information that describes the functions of software modules into which the vehicle control software of the vehicle is divided by function, in natural language; the reference parameter extraction unit extracts reference parameters that are control parameters of the vehicle referenced by the target module; the parameter selection unit selects a target parameter related to the behavior of the vehicle included in the request based on a second processing result obtained by natural language processing of the request and parameter description information that describes the functions of the reference parameter in natural language; and the output unit outputs the target parameter.

Images