Security rule generation device, navigation system, security rule generation method, navigation method, and program

By segmenting complex scenarios into subsceived conditions and calculating prior conditions to generate safety rules, the problem that existing technology is difficult to cope with complex autonomous driving scenarios is solved, and safety and efficiency in complex scenarios are achieved.

CN119053976APending Publication Date: 2024-11-29UNIV SHARED USE AGENCY INFORMATION & SYST RES INST
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202380034619.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-04-19
Filing Date
2023-04-11
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

The existing responsibility-aware safety theory is difficult to deal with complex autonomous driving scenarios, such as multiple lane changes and emergency telephone locations in road shoulder stop scenarios, and it is impossible to clarify the prior conditions for safe stops.

Method used

Segment complex scenarios into multiple subscenes, and combine prior condition calculation and control strategies to generate security rules that can meet security and posterior conditions.

Benefits of technology

It has achieved the formulation of safety rules that can cope with the meeting of specified posterior conditions and safety conditions in complex scenarios, and improved the safety and driving efficiency of autonomous driving vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119053976A_ABST
    Figure CN119053976A_ABST
Patent Text Reader

Abstract

According to the method, the safety rules capable of coping with complex scenes are formulated. The security rule generation device includes: an input unit configured to receive an input of scene information indicating a predetermined scene by linking sub-scenes in which a control policy and a destination are associated; a priori condition calculation unit configured to calculate, for each sub-scene, a priori condition for executing a control policy while achieving an objective; and an output unit configured to output security rule information in which a priori condition and a control policy for each sub-scene are combined.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to a safety rule generating device, a navigation system, a safety rule generating method, a navigation method and a program. Background Art

[0002] There is a technology for formulating safety rules for autonomous driving. Safety rule formulation technology is a technology for formulating safety rules that can be mathematically proven to be safe. For example, Non-Patent Document 1 discloses a mathematical model called Responsibility-Sensitive Safety (RSS).

[0003] <Prior Art Literature>

[0004] <Non-patent literature>

[0005] Non-patent document 1: Shai Shalev-Shwartz, Shaked Shammah, and Amnon Shashua, "On a formal model of safe and scalable self-driving cars," CoRR, abs / 1708.06374, 2017. Summary of the invention

[0006] <Problems to be Solved by the Invention>

[0007] However, the existing responsibility-aware safety theory has the problem of being unable to cope with complex scenarios. For example, in Non-Patent Document 1, it can only cope with a simple scenario such as an autonomous driving vehicle avoiding a collision with other cars traveling in the same lane.

[0008] In view of the above-mentioned technical problems, an object of one embodiment of the present invention is to formulate security rules that can cope with complex scenarios.

[0009] <Methods used to solve the problem>

[0010] In order to solve the above-mentioned problems, a safety rule generation device of one embodiment of the present invention comprises: an input unit, which is configured to accept input of scene information representing a specified scene by linking sub-scenarios that establish corresponding control strategies and objectives; a prior condition calculation unit, which is configured to calculate, for each sub-scenario, the prior conditions for executing the control strategy while achieving the objective; and an output unit, which is configured to output safety rule information that combines the prior conditions and control strategies of each sub-scenario.

[0011] <Effects of the Invention>

[0012] According to one embodiment of the present invention, it is possible to formulate security rules that can cope with complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a conceptual diagram showing an example of a scene in the same lane and in the same direction.

[0014] Figure 2 This is a conceptual diagram showing an example of a roadside stop scene.

[0015] Figure 3 This is a conceptual diagram showing an example of a subscene.

[0016] Figure 4 This is a diagram showing an example of the workflow of the goal achievement RSS.

[0017] Figure 5 This is a conceptual diagram showing an example of the relationship between a subscene and a scene tree.

[0018] Figure 6 This is a conceptual diagram showing an example of the relationship between sub-scenario, sub-purpose, and safety condition.

[0019] Figure 7 This is a conceptual diagram showing an example of the relationship between sub-scenario and control policy.

[0020] Figure 8 This is a conceptual diagram showing an example of a calculation process of a priori conditions.

[0021] Fig. 9 This is a diagram showing an example of the program logic system of the goal achievement RSS.

[0022] Fig.10 This is a block diagram showing an example of the overall configuration of a security rule generation system.

[0023] Fig.11 This is a block diagram showing an example of the hardware configuration of a computer.

[0024] Fig.12 This is a diagram showing an example of the functional configuration of a security rule generation system.

[0025] Fig.13 This is a flowchart showing an example of a processing procedure of a security rule generating method.

[0026] Fig.14 is a diagram showing an example of a simplex architecture.

[0027] Fig.15 This is a block diagram showing an example of the functional configuration of a navigation system.

[0028] Fig.16 This is a flowchart showing an example of the processing procedure of the navigation method.

[0029] Fig.17 This is a diagram showing an example of experimental results.

[0030] Fig.18A This is a diagram showing an example of the distribution of the travel time using the simplex framework of the goal achievement RSS and the simplex framework of the collision avoidance RSS.

[0031] Fig.18B This is a diagram showing an example of the distribution of the travel time of the simplex architecture and the high-level controller using the goal-achievement RSS. DETAILED DESCRIPTION

[0032] Hereinafter, each embodiment of the present invention will be described with reference to the drawings. In addition, in this specification and the drawings, the same reference numerals are given to components having substantially the same functional structure, and repeated descriptions are omitted.

[0033] [summary]

[0034] <Autonomous driving safety rule-making technology>

[0035] Autonomous driving safety rule formulation technology is a technology that formulates safety rules that can mathematically prove that an autonomous driving vehicle is safe as long as it meets the safety rules. For example, when two cars are driving in the same lane and in the same direction, a formula is derived to find the safe distance between cars that the following autonomous driving vehicle can avoid colliding with the car in front.

[0036] The autonomous driving safety rules are envisioned to be used for the following purposes. First, to determine who is responsible for the accident. This is the idea that the party that violated the safety rules was at fault when the accident occurred. Second, to prove the safety of the autonomous driving vehicle. This is the idea that the safety of the autonomous driving vehicle can be proven by following specific safety rules.

[0037] Third, monitoring of the autonomous vehicle during operation. This is to control the autonomous vehicle to comply with the safety rules when it is detected that the autonomous vehicle may violate the safety rules. Fourth, the safety benchmark or standard for autonomous driving. This is a restriction that does not allow sales unless specific safety rules are met. Fifth, insurance rate calculation. This is to make the insurance premium lower / higher for autonomous vehicles that follow / do not follow specific safety rules.

[0038] Autonomous driving safety rules are a concept that forms the basis for society's need for autonomous driving. If society generally recognizes that autonomous driving vehicles that follow safety rules are safe and can be driven on public roads, it is believed that the popularization of autonomous driving vehicles will be promoted. In addition, autonomous driving safety rules serve as a benchmark for determining the scope of manufacturer responsibility. This is the idea that if an autonomous driving vehicle has an accident, the manufacturer does not need to bear responsibility as long as the safety rules are followed.

[0039] <Collision Avoiding RSS (CA-RSS)>

[0040] As an existing technology for formulating safety rules for autonomous driving, there is Responsibility-Based Safety (RSS) disclosed in Non-Patent Document 1. RSS is widely recognized as a safety rule for autonomous driving. For example, RSS is used in many academic studies, and research is underway to reflect it in international standards.

[0041] The existing RSS is constructed according to the following logical structure. That is, as long as the prior conditions are met, the safety conditions can be met by executing the control strategy (proper response). The so-called control strategy refers to the plan and countermeasures for controlling the autonomous driving vehicle. An example of a control strategy is driving operations such as turning the steering wheel, applying the brakes, or pressing the accelerator to accelerate.

[0042] The existing RSS, for example, assumes the same lane and same direction scenario. Figure 1 This is a conceptual diagram showing an example of a scene in the same lane and in the same direction. Figure 1 As shown, this scenario assumes two cars (car 后 and cars 前 ) are driving in the same lane and in the same direction. At least the following car, i.e., the car 后 The conventional RSS shows that as long as the inter-vehicle distance calculated by a predetermined formula can be maintained, a collision can be avoided by applying the acceleration brake.

[0043] However, in the existing RSS, only the safety condition is guaranteed to be satisfied. For example, in the above-mentioned same lane and same direction scenario, only the safety condition of avoiding collision is guaranteed. Hereinafter, the existing RSS is referred to as "collision avoidance RSS (or CA-RSS)".

[0044] <Goal-Aware RSS (GA-RSS)>

[0045] In the present invention, in addition to the safety conditions, safety rules are also formulated to ensure that the prescribed a posteriori conditions are met. Hereinafter, the RSS proposed in the present invention is referred to as "Goal Achievement RSS (or GA-RSS)".

[0046] The goal achievement RSS is constructed according to the following logical structure. That is, as long as the prior conditions are met, the safety conditions can be met and the prescribed a posteriori conditions can be achieved by executing the control strategy.

[0047] In the following embodiments, a road shoulder stop scenario is taken as an example for description. Figure 2 is a conceptual diagram showing an example of a roadside stop scene. Figure 2 As shown in the figure, in this scenario, multiple vehicles are driving on a road consisting of multiple lanes (lanes 1 to 3), and the subject autonomous driving vehicle (Subject Vehicle; SV) maintains a safe distance with other vehicles (Principal Other Vehicles; POVs) as a safety condition, and performs multiple lane changes to safely stop at the location of an emergency phone (SOS) set on the shoulder (lane 3). tgt as a posteriori condition.

[0048] In such a complex scenario, there are many methods that can be taken to achieve the goal. For example, when changing lanes from the first lane (lane 1) to the second lane (lane 2), there is a choice of merging before or after the other car (POV1) traveling in the second lane. At this time, in order to merge before the other car, there is a choice of merging at the current driving speed or accelerating to merge. In this way, the prior conditions for safely achieving the specified a posteriori conditions are not clear.

[0049] In order to achieve the specified a posteriori conditions and safety conditions in the road shoulder stop scenario, the target achievement RSS divides the scenario into multiple sub-scenarios. Figure 3 This is a conceptual diagram showing an example of dividing a road shoulder stop scene into a plurality of sub-scenes.

[0050] like Figure 3 As shown, the goal achievement RSS divides the scenario of setting a post-condition of safely stopping at the location of an emergency phone set on the shoulder of the road into sub-goals 1 to 4. Sub-goal 1 is to prepare for merging. Sub-goal 2 is to change lanes to the second lane. Sub-goal 3 is to change lanes to the shoulder of the road. Sub-goal 4 is to stop at the location of the emergency phone.

[0051] By sequentially achieving these sub-goals 1 to 4, the a posteriori condition of the scenario can be achieved. In addition, as long as the control strategy that achieves each sub-goal 1 to 4 can be executed while always satisfying the safety condition, the a posteriori condition can be achieved while satisfying the safety condition.

[0052] <Logical Workflow>

[0053] Figure 4 It is a logical workflow for generating security rules for the purpose of achieving RSS. Figure 4 The logical workflow shown is used to derive safety rules that guarantee the satisfaction of safety conditions and a posteriori conditions with realistic computational effort.

[0054] However, this logical workflow is not only applicable to the goal-achieving RSS, for example, it can also be applied to the collision avoidance RSS. This logical workflow derives safety rules that satisfy safety conditions and a posteriori conditions. Since the collision avoidance RSS only needs to satisfy the safety conditions, if the collision avoidance RSS is a complex scene that can be divided into multiple sub-scenes, it can be applied in the same way as the goal-achieving RSS.

[0055] The driving scenario S is input into the workflow of the goal achievement RSS. In addition, the workflow of the goal achievement RSS outputs the safety rule (α, A). Where α represents the set of control strategies in the scenario S. A represents the set of prior conditions in the scenario S.

[0056] In the first step of the workflow, a scenario S = (Var, Safe, Env, Goal) is obtained. Safe represents the safety conditions to be observed by the vehicle, Env represents the environmental conditions to be observed by other vehicles, and Goal represents the specified goal. Var is a finite set of variables covering all variables used in the derivation of safety rules.

[0057] In the second step of the workflow, the scene is segmented. Specifically, the Goals of N sub-goals are determined. (1) 、……、Goal (N) , split the scene S into N sub-scenes S (1) ,……,S (N) Among them, S (i) =(Var, Safe, Env, Goal (i) )(i=1,…,N).

[0058] In the third step of the workflow, the sub-scenes are updated. Specifically, each sub-scene S is determined (i) The situation under which the safety condition is defined (i) And the environmental conditions Env (i) Then, a sub-scenario S is generated (1),……,S (N) The scene tree of dependency relationship T = T1, T 11 、T 12 ,……,T 111 、T 121 ,…….

[0059] Figure 5 It shows the sub-scene S (i) With sub-scene T w A conceptual diagram of the corresponding relationship. Figure 5 As shown, sub-scene S (i) The subscript i in the scene indicates the order from the beginning of the scene S. w The subscript w is assigned in such a way that the number of digits increases from the end to the beginning of the scene S. In this way, through the subscene T included in the scene tree T w The subscript w is used to represent each sub-scene T w dependencies.

[0060] In the fourth to sixth steps of the workflow, the control strategy of each sub-scenario is determined. Specifically, for each sub-scenario T w =(Var, Safe w , Env w ,Goal w ), search for the safe condition Safe w ∧Env w At the same time, achieve the sub-goal w The control strategy of w,1 ,……,α w,Kw .

[0061] Figure 6 It shows sub-scene T w Goal of the sub-goal w Safe w And the environmental conditions Env w A conceptual diagram of an example of . Figure 6 As shown, the Goal of each sub-scenario w Safe w And the environmental conditions Env w According to sub-scenario T w to define the situation more specifically.

[0062] Figure 7 It shows sub-scene T w With control strategy α w,i The conceptual diagram of the corresponding relationship. Figure 7 As shown, for each sub-scenario T w , determine more than one control strategy α w,i Each sub-scene Tw The corresponding control strategy α w,i The number of different, there is a control strategy α w,i There are also multiple control strategies α w,i sub-scene.

[0063] In the seventh step of the workflow, the prior conditions of each sub-scenario are determined. Specifically, for each control strategy α w,k , search for assertions in the program logic system (A w,u ) w,u At this time, backward reasoning is performed from w with a small number of digits to w with a large number of digits, so that by executing the control strategy α w,k , thereby achieving the sub-goal w And the prior conditions of the subsequent sub-scenario are met. In addition, the program logic system in the goal-achieving RSS will be described later.

[0064] Figure 8 It shows the prior condition A w,u A conceptual diagram of the computational process of Figure 8 As shown, each sub-scene T w The prior condition A w,u According to sub-scenario T w The dependencies between them are calculated in sequence. At this time, the prior condition A w,u By meeting the safety condition Safe w ∧Env w While executing each control strategy α w,k To calculate the Goal to meet the sub-goal w .

[0065] In the eighth step of the workflow, the control strategy and prior conditions in the entire scene are calculated. Specifically, the control strategy α of the sub-scene is w,k And the prior condition A w,u Combine them respectively. Thus, we can get {A}α{Goal}: the control strategy α and the prior condition A where Safe∧Env is true.

[0066] In the ninth step of the workflow, the safety rule (α, A) is output. α represents the control strategy α in scenario S w,k A represents the prior condition A in scene S. w,u A collection of .

[0067] <Program logic system>

[0068] In the goal achievement RSS, in order to achieve Figure 4The program logic system dFHL (differential Floyd-Hoare Logic) with added safety conditions is introduced according to the workflow shown. The program logic system dFHL is an extension of the ordinary differential equation (ODE) used for continuous dynamics in the Hoare logic (see reference 1) used for program verification, and also adds safety conditions that should be true during program execution.

[0069] The details of the Hall logic are disclosed in the following reference 1. The details of the addition of the safety condition are disclosed in the following reference 2.

[0070] [Reference 1] CAR Hoare, "An axiomatic basis for computer programming," Communications of the ACM, vol. 12, pp. 576-580, 583, 1969.

[0071] [Reference 2] FSde Boer, U.Hannemann, and WPde Roever, "Hoare-stylecompositional proof systems for reactive shared variable concurency," in Foundations of Software Technology and Theoretical Computer Science, 17th Conference, Kharagpur, India, December 18-20, 1997.

[0072] Hoare logic ensures that if the prior condition A is true, the a posteriori condition B is true after the program α is executed. In addition, A, α, and B are called "Hoare triples". Hoare logic is expressed by the following formula.

[0073]

Formula 1

[0074] {A}α{B}

[0075] The program logic system dFHL adds the safety condition S to the Hoare triple. The program logic system dFHL ensures that the a priori condition A is true and the a posteriori condition B is true after the execution of program α, and the safety condition S is always true during the execution of program α. Hereinafter, A, α, B, and S are called "Hoare quadruples". dFHL is represented by the following formula.

[0076]

Formula 2

[0077] {A}α{B}:S

[0078] Fig. 9 It is the inference rule that constitutes the program logic system dFHL. In the goal achievement RSS, in order to execute Figure 4 Step 7 of the logical workflow shown uses Fig. 9 The program logic system dFHL is shown.

[0079] <Software Implementation>

[0080] Purpose Achievement RSS can calculate a priori conditions represented by tens of lines of logical expressions by implementing this workflow as software. In particular, Purpose Achievement RSS can achieve semi-formal software implementation using Mathematica (registered trademark), a well-known computer algebra system. Specifically, algebraic / logical notation operations (such as substitution, quadratic equation solution, inequality proof, etc.) that are not explicitly related to program dynamics can be formalized through Mathematica.

[0081] [Implementation Method]

[0082] An embodiment of the present invention is a safety rule generation system that generates safety rules corresponding to a scenario that achieves a predetermined a posteriori condition and a predetermined safety condition as input. In this embodiment, the above-mentioned road shoulder stop scenario is assumed as a scenario that achieves the predetermined a posteriori condition and safety condition. The safety rule generation system in this embodiment transforms the scenario that achieves the predetermined a posteriori condition into a scenario tree linked by multiple sub-scenarios, and generates safety rules that combine the control strategies and a priori conditions in each sub-scenario.

[0083] The safety rule generation system in this embodiment divides the complex scenario of achieving the specified a posteriori conditions and safety conditions into multiple simple sub-scenarios that achieve sub-goals. The so-called sub-goals are the posteriori conditions that should be achieved in the process of achieving the final a posteriori conditions and safety conditions. The control strategy in the sub-scenario represents the control (driving operation) to be performed in order to achieve these sub-goals. That is, the safety rules in this embodiment are safety rules that indicate that as long as the control strategy is repeatedly executed in the sub-scenario corresponding to the current situation, the specified a posteriori conditions and safety conditions can be finally achieved.

[0084] <Overall structure of the security rule generation system>

[0085] First, refer to Fig.10 The overall structure of the security rule generation system in this embodiment will be described. Fig.10This is a block diagram showing an example of the overall configuration of the security rule generation system in this embodiment.

[0086] like Fig.10 As shown, the security rule generation system 1 in this embodiment includes a security rule generation device 10 and a user terminal 30. The security rule generation device 10 and the user terminal 30 are connected to each other via a communication network N1 such as a LAN (Local Area Network) or the Internet so as to be data communicable.

[0087] The security rule generation device 10 is an information processing device such as a PC (Personal Computer), a workstation, or a server that generates security rules based on a request from a user terminal 30. The security rule generation device 10 receives scenario information from the user terminal 30. Scenario information is information indicating a scenario for which a security rule is to be generated. In addition, the security rule generation device 10 generates security rule information and sends it to the user terminal 30. The security rule information is information indicating a security rule for the received scenario information.

[0088] The user terminal 30 is an information processing terminal such as a PC, a tablet terminal, or a smart phone operated by a user. The user terminal 30 receives input of a scene as a target for generating security rules according to the user's operation, and sends scene information for transforming the scene into a scene tree to the security rule generating device 10. In addition, the user terminal 30 receives security rule information from the security rule generating device 10 and outputs it to the user.

[0089] in addition, Fig.10 The overall structure of the security rule generation system 1 shown is an example, and various system structure examples can be formed according to the use and purpose. For example, the security rule generation device 10 can also be implemented by multiple computers, or can be implemented as a cloud computing service. In addition, for example, the security rule generation system 1 can also be implemented by an independent information processing device that has the functions that the security rule generation device 10 and the user terminal 30 should each have.

[0090] <Hardware structure of the safety rule generation system>

[0091] Next, refer to Fig.11 The hardware configuration of the security rule generation system 1 in this embodiment will be described.

[0092] Computer Hardware Structure

[0093] The security rule generating device 10 and the user terminal 30 in this embodiment are implemented by, for example, a computer. Fig.11 This is a block diagram showing an example of the hardware configuration of a computer in this embodiment.

[0094] like Fig.11 As shown, the computer 500 in this embodiment has a CPU (Central Processing Unit) 501, a ROM (Read Only Memory) 502, a RAM (Random Access Memory) 503, a HDD (Hard Disk Drive) 504, an input device 505, a display device 506, a communication I / F (Interface) 507, and an external I / F 508. The CPU 501, the ROM 502, and the RAM 503 form a so-called computer. The hardware of the computer 500 is connected to each other via a bus 509. In addition, the input device 505 and the display device 506 may also be connected to the external I / F 508 and used.

[0095] The CPU 501 is a computing device that realizes the overall control and functions of the computer 500 by reading programs and data from a storage device such as the ROM 502 or the HDD 504 to the RAM 503 and executing the processing.

[0096] ROM502 is an example of a non-volatile semiconductor memory (storage device) that can retain programs and data even when the power is turned off. ROM502 functions as a main storage device that stores various programs and data required for CPU501 to execute various programs installed in HDD504. Specifically, ROM502 stores boot programs such as BIOS (Basic Input / Output System) and EFI (Extensible Firmware Interface) that are executed when the computer 500 is started, OS (Operating System) settings, network settings, and other data.

[0097] RAM 503 is an example of a volatile semiconductor memory (storage device) in which programs and data are deleted when power is turned off. RAM 503 is, for example, DRAM (Dynamic Random Access Memory), SRAM (Static Random Access Memory), etc. RAM 503 provides a work area that is expanded when CPU 501 executes various programs installed in HDD 504.

[0098] The HDD 504 is an example of a non-volatile storage device for storing programs and data. The programs and data stored in the HDD 504 include an OS, which is basic software for controlling the entire computer 500, and applications that provide various functions on the OS. In addition, the computer 500 may use a storage device (e.g., SSD: Solid State Drive) that uses a flash memory as a storage medium instead of the HDD 504.

[0099] The input device 505 is a touch panel, operation keys or buttons, a keyboard or a mouse for the user to input various signals, a microphone for inputting sound data such as sound, etc.

[0100] The display device 506 is composed of a display such as a liquid crystal or an organic EL (Electro-Luminescence) that displays a screen, a speaker that outputs audio data such as sound, and the like.

[0101] The communication I / F 507 is an interface for connecting to a communication network and for the computer 500 to perform data communication.

[0102] The external I / F 508 is an interface for connecting to an external device. The external device includes a driver device 510 and the like.

[0103] The drive device 510 is a component for setting the recording medium 511. Here, the so-called recording medium 511 includes a medium that records information optically, electronically or magnetically, such as a CD-ROM, a floppy disk, and an optical magneto-disk. In addition, the recording medium 511 may also include a semiconductor memory that records information electrically, such as a ROM, a flash memory, etc. Thus, the computer 500 can read and / or write the recording medium 511 via the external I / F 508.

[0104] In addition, various programs installed in HDD504 are installed by, for example, setting the distributed recording medium 511 in a drive device 510 connected to the external I / F508 and having the drive device 510 read out various programs recorded in the recording medium 511. Alternatively, various programs installed in HDD504 may be installed by downloading from another network different from the communication network via the communication I / F507.

[0105] <Functional structure of the safety rule generation system>

[0106] Next, refer to Fig.12 The functional structure of the security rule generation system in this embodiment is described. Fig.12 This is a block diagram showing an example of the functional configuration of the security rule generation system 1 in this embodiment.

[0107] <<Functional structure of user terminal 30>>

[0108] like Fig.12 As shown, the user terminal 30 in this embodiment includes a scene input unit 301, a scene segmentation unit 302, a scene information generation unit 303, a control strategy determination unit 304, a scene information transmission unit 305 and a safety rule display unit 306.

[0109] The scene input unit 301, the scene segmentation unit 302, the scene information generation unit 303, the control strategy determination unit 304, the scene information transmission unit 305 and the safety rule display unit 306 execute the following steps by the CPU 501: Fig.11 This is achieved by the processing of the program expanded from HDD 504 to RAM 503 as shown.

[0110] The scenario input unit 301 receives input of a scenario according to a user's operation. The scenario is information that associates a safety condition with a post-hoc condition. The post-hoc condition is information that indicates a post-hoc condition of the entire scenario.

[0111] The scene segmentation unit 302 segments the scene received by the scene input unit 301 into sub-scenes according to the user's operation. The sub-scene is information that associates a safety condition with a posteriori condition. The posteriori condition is information that indicates a sub-purpose of a sub-scene.

[0112] The scene information generation unit 303 determines the status of each sub-scene divided by the scene division unit 302 according to the user's operation. In addition, the scene information generation unit 303 defines the safety conditions of each sub-scene according to the user's operation. The scene information generation unit 303 generates a scene tree (hereinafter also referred to as "scene information") in which the sub-scenes are linked in a tree shape according to the dependency relationship.

[0113] The control strategy determination unit 304 determines the control strategy of each sub-scenario included in the scenario information generated by the scenario information generation unit 303. The control strategy is a control strategy that can satisfy the a posteriori condition by executing while satisfying the safety condition. In addition, the control strategy determination unit 304 sets the determined control strategy in the scenario information.

[0114] The scenario information transmitting unit 305 transmits the scenario information for which the control policy is set by the control policy determining unit 304 to the security rule generating device 10 .

[0115] The security rule display unit 306 displays the security rule information received from the security rule generating device 10 on the display device 506 or the like.

[0116] 《Functional structure of security rule generation device》

[0117] like Fig.12As shown, the security rule generating device 10 in this embodiment includes a scenario information receiving unit 101 , a priori condition calculating unit 102 , a security rule generating unit 103 , and a security rule output unit 104 .

[0118] The scene information receiving unit 101, the prior condition calculating unit 102, the security rule generating unit 103 and the security rule output unit 104 execute the following operations by the CPU 501: Fig.11 This is achieved by the processing of the program expanded from HDD 504 to RAM 503 as shown.

[0119] The scene information accepting unit 101 accepts input of scene information received from the user terminal 30 .

[0120] The prior condition calculation unit 102 calculates the prior condition of each sub-scenario included in the scenario information accepted by the scenario information acceptance unit 101. At this time, the prior condition calculation unit 102 uses a program verifier based on the above-mentioned program logic system dFHL to calculate the prior condition. The program verifier verifies whether the a posteriori condition is true after the control strategy is executed through the prior condition being true, and whether the safety condition is maintained during the execution of the control strategy.

[0121] In addition, the prior condition calculation unit 102 traces back on the chain of sub-scenario through progressive back propagation, and calculates the prior conditions of each sub-scenario in turn. The prior condition calculation unit 102 calculates the prior conditions of the sub-scenario so that the posterior conditions of the sub-scenario are satisfied and the prior conditions of subsequent sub-scenario are satisfied by executing the control strategy while satisfying the safety condition.

[0122] The safety rule generation unit 103 combines the control strategies of each sub-scenario included in the scenario information. In addition, the safety rule generation unit 103 combines the prior conditions calculated by the prior condition calculation unit 102. Furthermore, the safety rule generation unit 103 generates safety rule information composed of the combined control strategies and prior conditions.

[0123] The security rule output unit 104 transmits the security rule information generated by the security rule generation unit 103 to the user terminal 30 .

[0124] <Processing of the security rule generation system>

[0125] Next, refer to Fig.13 The processing procedure of the security rule generation method executed by the security rule generation system 1 in this embodiment will be described. Fig.13 This is a flowchart showing an example of the processing procedure of the security rule generation method in this embodiment.

[0126] In step S301 , the scene input unit 301 of the user terminal 30 receives input of a scene according to a user operation. Then, the scene input unit 301 sends the received scene to the scene segmentation unit 302 .

[0127] In step S302, the scene segmentation unit 302 of the user terminal 30 receives the scene from the scene input unit 301. Then, the scene segmentation unit 302 segments the scene into sub-scenes according to the user's operation. Then, the scene segmentation unit 302 sends the multiple sub-scenes generated by segmentation to the scene information generation unit 303.

[0128] In step S303, the scene information generating unit 303 of the user terminal 30 receives a plurality of sub-scenes from the scene dividing unit 302. Next, the scene information generating unit 303 determines the status and safety condition of each sub-scene according to the user's operation.

[0129] Next, the scene information generation unit 303 generates scene information in which the sub-scenes are linked in a tree form according to the dependency relationship according to the user's operation. Next, the scene information generation unit 303 sends the generated scene information to the control strategy determination unit 304 .

[0130] In step S304, the control policy determination unit 304 included in the user terminal 30 receives the scene information from the scene information generation unit 303. Next, the control policy determination unit 304 determines the control policy of each sub-scene included in the scene information according to the user's operation.

[0131] Next, the control strategy determination unit 304 sets the determined control strategy in the scene information. Next, the control strategy determination unit 304 transmits the scene information in which the control strategy is set to the scene information transmission unit 305.

[0132] In step S305, the scenario information transmitting unit 305 included in the user terminal 30 receives the scenario information from the control policy determining unit 304. Next, the scenario information transmitting unit 305 transmits the received scenario information to the security rule generating device 10.

[0133] In step S101, the scenario information accepting unit 101 included in the security rule generating device 10 receives scenario information from the user terminal 30. Next, the scenario information accepting unit 101 transmits the received scenario information to the priori condition calculating unit 102.

[0134] In step S102, the priori condition calculation unit 102 included in the security rule generation device 10 receives the scenario information from the scenario information reception unit 101. Next, the priori condition calculation unit 102 calculates the priori condition of each sub-scenario included in the scenario information.

[0135] Next, the priori condition calculation unit 102 sets the calculated priori condition in the scenario information. Next, the priori condition calculation unit 102 sends the scenario information in which the priori condition is set to the security rule generation unit 103.

[0136] In step S103, the security rule generation unit 103 of the security rule generation device 10 receives the scenario information from the priori condition calculation unit 102. Next, the security rule generation unit 103 combines the control strategy and the priori condition of each sub-scenario.

[0137] Next, the security rule generation unit 103 generates security rule information composed of the combined control strategy and the priori condition, and then sends the generated security rule information to the security rule output unit 104 .

[0138] In step S104, the security rule output unit 104 included in the security rule generation device 10 receives the security rule information from the security rule generation unit 103. Next, the security rule output unit 104 transmits the received security rule information to the user terminal 30.

[0139] In step S306, the security rule display unit 306 included in the user terminal 30 receives the security rule information from the security rule generating device 10. Next, the security rule display unit 306 displays the received security rule information on the display device 506 or the like.

[0140] <Effects of implementation>

[0141] The safety rule generation system in this embodiment generates safety rules that can cope with complex scenarios that achieve the prescribed a posteriori conditions and safety conditions by transforming the scenario that achieves the prescribed a posteriori conditions and safety conditions into a scenario tree composed of multiple sub-scenario links, and calculating the a priori conditions for the control strategies determined in each sub-scenario. Since the sub-scenario is a simple scenario for achieving the prescribed sub-purpose, the a priori conditions for the control strategies can be easily calculated. Therefore, according to the safety rule generation system in this embodiment, safety rules that can cope with complex scenarios that achieve the prescribed a posteriori conditions and safety conditions can be formulated.

[0142] In particular, the security rule generation system in this embodiment imports a program logic system to which security conditions are added in order to calculate the prior conditions of each sub-scenario. In addition, the security rule generation system in this embodiment can implement security conditions represented by dozens of lines of logic expressions in software. Therefore, according to the security rule generation system in this embodiment, security rules that can cope with complex scenarios can be calculated with realistic calculation amounts.

[0143] [Application Example 1]

[0144] The security rules derived by the purpose-achieved RSS can be applied to the simplex architecture. The details of the simplex architecture are disclosed in the following references 3 and 4.

[0145] [Reference 3] TLCrenshaw, E.Gunter, CLRobinson, L.Sha, and PRKumar, "The simplex reference model:Limiting fault-propagation due to unreliablecomponents in cyber-physical system architectures," in 28th IEEEInternationalReal-Time Systems Symposium (RTSS2007), 2007, pp.400-412.

[0146] [Reference 4] D.Seto, B.Krogh, L.Sha, and A.Chutinan, "The simplexarchitecture for safe online control system upgrades," in Proceedings of the 1998American Control Conference.ACC (IEEE Cat.No.98CH36207), vol.6, 1998, pp.3504-3508.

[0147] Fig.14 is a conceptual diagram showing an example of a simplex architecture. Fig.14 As shown, the simplex architecture is composed of an advanced controller (Advanced Controller) 401, a baseline controller (Baseline Controller) 402, a decision module (Decision Module) 403 and a controlled object (Plant) 404.

[0148] The advanced controller 401 performs complex control that pursues not only safety but also various performance indicators, such as comfort, speed, and fuel efficiency, etc. The baseline controller 402 performs relatively simple control that emphasizes safety.

[0149] The decision module 403 switches between the advanced controller 401 and the baseline controller 402. The decision module 403 usually uses the advanced controller 401 for better performance. On the other hand, when a serious problem is detected in terms of safety, the decision module 403 switches to the baseline controller 402.

[0150] The advanced controller 401 performs complex control, so it is difficult to predict and verify the behavior. On the other hand, the baseline controller 402 performs simple control, so it is easy to predict and verify. Therefore, the simplex architecture can ensure safety as the entire system by appropriately switching control between the advanced controller 401 and the baseline controller 402.

[0151] Conventional collision avoidance RSS has been applied to the simplex architecture. In this case, the high-level controller 401 is a general autonomous driving control module. Since the autonomous driving control module is a machine learning model that learns from accumulated driving data, its judgment process is a black box.

[0152] The decision module 403 monitors the prior conditions of the collision avoidance RSS. The decision module 403 switches control to the baseline controller 402 when it detects that the prior conditions are about to be unsatisfied.

[0153] The baseline controller 402 performs control according to the control strategy of the collision avoidance RSS. The baseline controller 402 performs control in pursuit of safety without considering performance indicators such as comfort. Therefore, according to the simplex architecture to which the collision avoidance RSS is applied, an autonomous driving vehicle that generally performs high-performance autonomous driving and can avoid collisions in safety crises can be realized.

[0154] The safety rules based on the goal-attainment RSS ensure compatibility with the safety rules based on the collision-avoidance RSS. Therefore, similar to the collision-avoidance RSS, the goal-attainment RSS can be applied to the simplex architecture.

[0155] Safety rules based on RSS for purpose guarantee that safety conditions and a posteriori conditions are met. Safety rules based on RSS for collision avoidance guarantee that safety conditions are met. Therefore, safety rules based on RSS for purpose can include safety rules based on RSS for collision avoidance and meet specified a posteriori conditions.

[0156] Therefore, by applying the simplex architecture of the purpose-achieved RSS, it is possible to realize an autonomous driving vehicle that can normally achieve high-performance autonomous driving and achieve specified a posteriori conditions while avoiding conflicts in the event of a safety crisis.

[0157] [Application Example 2]

[0158] The simplex architecture to which the purpose-achieving RSS is applied can be used, for example, for implementation in a navigation system for an autonomous driving vehicle. For example, Reference 5 below discloses a navigation system that implements the conventional collision avoidance RSS.

[0159] [Reference 5] International Publication No. 2018 / 115963

[0160] The navigation system in this application example is an on-board system that controls the behavior or action of the autonomous driving vehicle using the safety rules generated by the safety rule generation system 1 in the embodiment. In order to achieve the specified navigation goal, the navigation system determines the path to the destination, plans the behavior in the autonomous driving, plans the actions in each behavior, and performs vehicle body control to achieve the planned behavior or action.

[0161] The navigation system in this application example is installed in an autonomous vehicle, for example. For example, the navigation system is installed in a dashboard or a center console of the autonomous vehicle. The navigation system can also be implemented by cloud computing that cooperates with an information processing device installed in a remote location via a mobile phone network or the like.

[0162] <Functional structure of the navigation system>

[0163] First, refer to Fig.15 The functional structure of the navigation system in this application example is described. Fig.15 : is a block diagram showing an example of the functional configuration of the navigation system 6 in this application example.

[0164] like Fig.15 As shown, the navigation system 6 in this application example includes a target input unit 61, a path planning unit 62, an environment acquisition unit 63, a behavior planning unit 64, an action planning unit 65, an action control unit 66, a path information storage unit 601, a scene storage unit 602, and a safety rule storage unit 603. The behavior planning unit 64 includes a scene acquisition unit 641 and a scene verification unit 642. The action planning unit 65 includes an action determination unit 651 and an action verification unit 652.

[0165] Route information is pre-stored in the route information storage unit 601. The route information includes map information, road information, congestion information, etc. The route information can also be updated at any time using a communication unit connected to a mobile communication network or the like.

[0166] A plurality of scenes are pre-stored in the scene storage unit 602. A navigation target is set in the scene. The navigation target is equivalent to the safety condition and the a posteriori condition of the scene. Therefore, by making the autonomous driving vehicle move according to the scene to achieve the navigation target, the safety condition and the a posteriori condition of the scene can be achieved.

[0167] The security rule storage unit 603 stores security rules corresponding to each scene stored in the scene storage unit 602 in advance. The security rules are generated in advance by the security rule generation system 1 in the embodiment for each scene stored in the scene storage unit 602. The security rule storage unit 603 stores, for each scene stored in the scene storage unit 602, security rules for calculating the prior conditions for executing the control strategy while achieving the sub-purpose of the sub-scene for each sub-scene divided from the scene.

[0168] The destination input unit 61 receives input of a desired navigation destination. The navigation destination may be input by the user or may be automatically determined based on the scenario.

[0169] The route planning unit 62 determines a route for achieving the navigation target based on the route information stored in the route information storage unit 601 .

[0170] The environment acquisition unit 63 acquires environment information representing the environment of the autonomous driving vehicle based on images captured by a vehicle-mounted camera or the like of the surroundings of the autonomous driving vehicle. The vehicle-mounted cameras are arranged at positions and numbers that can capture images so that there are no blind spots around the vehicle body. The vehicle-mounted cameras can be arranged at any positions such as rearview mirrors, bumpers, side mirrors, etc.

[0171] The behavior planning unit 64 plans the behavior of the autonomous driving vehicle based on the environmental information acquired by the environment acquisition unit 63. The behavior planning unit 64 uses the scenario acquisition unit 641 and the scenario verification unit 642 to determine a scenario that can achieve the navigation goal input to the goal input unit 61.

[0172] The scenario acquisition unit 641 acquires a scenario for achieving the navigation target from the scenario storage unit 602. The scenario for achieving the navigation target is a scenario in which the a posteriori condition includes the navigation target and the navigation target can be achieved by achieving the a posteriori condition.

[0173] The scenario verification unit 642 verifies whether each scenario can achieve the navigation goal based on the safety rules corresponding to the scenario acquired by the scenario acquisition unit 641 and the environmental information acquired by the environment acquisition unit 63 .

[0174] If the scenario verification unit 642 determines that the scenario cannot achieve the navigation goal, the scenario is rejected. If there are multiple scenarios that can achieve the navigation goal, the scenario verification unit 642 determines the scenario to be executed according to a predetermined rule.

[0175] The action planning unit 65 plans the action of the autonomous driving vehicle based on the environmental information acquired by the environment acquisition unit 63. The action planning unit 65 uses the action determination unit 651 and the action verification unit 652 to determine the navigation action for achieving the navigation target input to the target input unit 61.

[0176] The action determination unit 651 determines a navigation action for achieving the navigation goal input to the goal input unit 61 based on the scenario determined by the behavior planning unit 64 and the environment information acquired by the environment acquisition unit 63 .

[0177] The action verification unit 652 verifies whether the control strategy corresponding to the navigation action determined by the action determination unit 651 can achieve the navigation target input to the target input unit 61 based on the safety rules corresponding to the scene determined by the behavior planning unit 64 and the environmental information obtained by the environment acquisition unit 63.

[0178] When the operation verification unit 652 determines that the navigation operation can achieve the navigation target, it determines to execute the navigation operation. When the operation verification unit 652 determines that the navigation operation cannot achieve the navigation target, it rejects the navigation operation.

[0179] The action planning unit 65 can be implemented in the simplex architecture shown in Application Example 1. In this case, the action determination unit 651 is implemented by the high-level controller 401 or the baseline controller 402. The action verification unit 652 is implemented by the determination module 403. The controlled object 404 corresponds to the environment acquisition unit 63.

[0180] The motion control unit 66 controls the body of the automatically driven vehicle according to the navigation motion determined by the motion planning unit 65 .

[0181] <Navigation system processing>

[0182] Next, refer to Fig.16 The processing procedure of the navigation method executed by the navigation system 6 in this application example is described below. Fig.16 This is a flowchart showing an example of the processing procedure of the navigation method in this application example.

[0183] In step S61, the destination input unit 61 included in the navigation system 6 receives input of a navigation destination. The navigation destination may be input by the user or may be automatically determined based on the scenario.

[0184] When the navigation destination is input by the user, the user may input the navigation destination through console operation or through voice input using voice recognition or the like.

[0185] In step S62, the route planning unit 62 of the navigation system 6 receives the navigation target from the target input unit 61. Next, the route planning unit 62 determines a route to achieve the navigation target based on the route information stored in the route information storage unit 601. Then, the route planning unit 62 sends the determined route to the behavior planning unit 64.

[0186] In step S63, the environment acquisition unit 63 of the navigation system 6 acquires environment information indicating the environment of the autonomous driving vehicle. The environment information may include any information that may affect the safety conditions and a posteriori conditions of the autonomous driving vehicle, such as the status of other vehicles traveling around the autonomous driving vehicle, the status of the road being traveled, and the status of traffic lights or traffic signs set on the road.

[0187] In step S64, the behavior planning unit 64 included in the navigation system 6 receives the route from the route planning unit 62. Next, the behavior planning unit 64 receives the navigation target from the target input unit 61. Next, the behavior planning unit 64 inputs the received navigation target to the scenario acquisition unit 641.

[0188] The scenario acquisition unit 641 receives the navigation target from the behavior planning unit 64. Then, the scenario acquisition unit 641 acquires a scenario for achieving the navigation target from the scenario storage unit 602. The scenario acquisition unit 641 may acquire multiple scenarios. Then, the scenario acquisition unit 641 sends the acquired scenario to the scenario verification unit 642.

[0189] The scenario verification unit 642 receives the scenario from the scenario acquisition unit 641. Next, the scenario verification unit 642 acquires the environment information from the environment acquisition unit 63. Next, the scenario verification unit 642 acquires the security rule corresponding to the scenario received from the scenario acquisition unit 641 from the security rule storage unit 603.

[0190] The scenario verification unit 642 verifies whether each scenario can achieve the navigation goal based on the obtained safety rules and environmental information. If the scenario is determined to be unable to achieve the navigation goal, the scenario verification unit 642 rejects the scenario. If there are multiple scenarios that can achieve the navigation goal, the scenario verification unit 642 determines the scenario to be executed according to predetermined rules. The scenario verification unit 642 may also prompt the user with multiple scenarios that can achieve the navigation goal and determine the scenario to be executed based on the user's selection. The scenario verification unit 642 sends the determined scenario to the action planning unit 65.

[0191] In step S65, the motion planning unit 65 included in the navigation system 6 receives the scenario from the behavior planning unit 64. Next, the motion planning unit 65 inputs the received scenario to the motion determining unit 651.

[0192] The action decision unit 651 receives the scenario from the action planning unit 65. Next, the action decision unit 651 acquires the environment information from the environment acquisition unit 63. Next, the action decision unit 651 determines the navigation action for achieving the navigation target based on the environment information acquired from the environment acquisition unit 63. Then, the action decision unit 651 sends the navigation action to the action verification unit 652.

[0193] The action verification unit 652 receives the navigation action from the action determination unit 651. Next, the action verification unit 652 obtains the environment information from the environment acquisition unit 63. Then, the action verification unit 652 obtains the safety rule corresponding to the scenario determined by the behavior planning unit 64 from the safety rule storage unit 603.

[0194] The action verification unit 652 verifies whether the control policy corresponding to the navigation action can achieve the navigation target based on the safety rules and environmental information. If the action verification unit 652 determines that the navigation action can achieve the navigation target, it sends the navigation action to the action control unit 66.

[0195] On the other hand, if the action verification unit 652 determines that the navigation action cannot achieve the navigation target, the action verification unit 652 rejects the navigation action. If the navigation action is rejected by the action verification unit 652, the action determination unit 651 determines another navigation action for achieving the navigation target.

[0196] In step S66, the motion control unit 66 included in the navigation system 6 receives the navigation motion from the motion planning unit 65. Next, the motion control unit 66 controls the body of the automatically driven vehicle based on the received navigation motion.

[0197] As described above, in the navigation system 6 in this application example, the safety rule based on the purpose-achieved RSS is used for two purposes. The first purpose is to determine the conditions of the scenario executed by the behavior planning unit 64. For example, the first purpose can be used to select whether to execute the first roadside stop scenario of stopping at the position of the first emergency phone close to the current position or to execute the second roadside stop scenario of stopping at the position of the second emergency phone far from the current position when two emergency phones are set in the direction of travel.

[0198] For example, if one of the first shoulder stop scenario and the second shoulder stop scenario does not meet the safety condition and the a posteriori condition, the scenario that meets the safety condition and the a posteriori condition can be selected. For example, if both the first shoulder stop scenario and the second shoulder stop scenario meet the safety condition and the a posteriori condition, the first shoulder stop scenario of stopping at a position close to the current position can be selected.

[0199] The second purpose of the safety rule is to determine the conditions of the action executed by the action planning unit 65. For example, the second purpose can be used to select whether to execute the navigation action of changing lanes before other cars running parallel in the adjacent lane or to execute the navigation action of changing lanes after other cars running parallel in the adjacent lane during the execution of the second roadside stop scenario of stopping at the second emergency phone position.

[0200] For example, if the navigation action of changing lanes before other cars cannot satisfy the safety condition and the a posteriori condition, the action of changing lanes after other cars can be selected. In addition, for example, if the safety condition and the a posteriori condition are satisfied whether the lane is changed before other cars or after other cars, the navigation action with the shortest driving time in the scene can be selected.

[0201] [Experimental results]

[0202] Reference Fig.17 , Fig.18A and Fig.18B The following is an experimental result of a comparative test to evaluate the performance of the target achievement RSS. In the comparative experiment, the implementation of only the advanced controller (hereinafter referred to as "AC") and the implementation of the simplex architecture using the collision avoidance RSS (hereinafter referred to as "AC+RSS") were compared. CA ”) and the implementation of a simplex architecture using the purpose-achieving RSS (hereinafter referred to as “AC+RSS GA ”) to simulate the shoulder stop scenario. In the evaluation experiment, 2350 shoulder stop scenarios with different parameters were simulated using the above three implementations.

[0203] Fig.17 is a diagram showing a table in which the experimental results are cumulatively summarized. Fig.17 In the scenario, goal is the percentage of the goal achieved. collision is the percentage of collisions that occurred. RSS violation is the situation where a safe distance between vehicles is not maintained. time is the driving time until the end of the scenario. jerk is the total amount of acceleration during driving. BC time is the average time controlled by the baseline controller.

[0204] like Fig.17 As shown, in AC+RSS CA In AC, although collisions were avoided, the a posteriori conditions could not be met. In AC, although collisions were avoided and the a posteriori conditions were met, many situations occurred in which the safety conditions of RSS could not be met. In addition, in AC, it was observed that the time until the a posteriori conditions were met became longer. On the other hand, in AC+RSS GA In the process, the posterior condition can be achieved while avoiding collisions, and all situations can be completed within a certain time.

[0205] Fig.18A This is a diagram showing an example of the distribution of the travel time using the simplex framework of the goal achievement RSS and the simplex framework of the collision avoidance RSS. Fig.18A AC+RSS GAThe travel time is set as the vertical axis, and AC+RSS CA The travel time is set as the horizontal axis, and the simulation results are plotted in a curve graph. Fig.18B This is a diagram showing an example of the distribution of the travel time of the simplex architecture and the high-level controller using the goal-achievement RSS. Fig.18B AC+RSS GA The graph plots the simulation results with the travel time of as the vertical axis and the travel time of AC as the horizontal axis. The size of each circle indicates the number of simulations corresponding to the combination of travel times.

[0206] exist Fig.18A and Fig.18B The sample plotted at the bottom right of the graph is AC+RSS GA The situation where the a posteriori condition is met in a relatively short period of time. Fig.18A and Fig.18B As shown, in the AC and AC+RSS CA In any comparison of AC+RSS GA A trend showing favorable results was observed.

[0207] As described above, according to the results of this experiment, it was found that if the purpose of use fulfills the safety rule of RSS, the predetermined a posteriori condition can be fulfilled safely and in a short time.

[0208] [Replenish]

[0209] Each function of the above-described embodiments can be implemented by one or more processing circuits. Here, the "processing circuit" in this specification includes a processor programmed to execute each function by software, such as a processor implemented by an electronic circuit, an ASIC (Application Specific Integrated Circuit), a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array), an existing circuit module, and other devices designed to execute the above functions.

[0210] As mentioned above, although embodiment of this invention was demonstrated in detail, this invention is not limited to these embodiment, Various deformation|transformation and change are possible within the range of the summary of this invention described in a claim.

[0211] This application claims priority from Japanese Patent Application No. 2022-69068 filed in the Japan Patent Office on April 19, 2022, the entire contents of which are incorporated herein by reference.

[0212] Explanation of symbols

[0213] 1: Security rule generation system

[0214] 10: Security rule generation device

[0215] 101: Scene Information Receiving Department

[0216] 102: Prior Condition Calculation Department

[0217] 103: Security rule generation department

[0218] 104: Security rule output unit

[0219] 30: User terminal

[0220] 301: Scene input unit

[0221] 302: Scene segmentation

[0222] 303: Scene information generation unit

[0223] 304: Control strategy decision section

[0224] 305: Scene information sending unit

[0225] 306: Safety rules display unit

[0226] 6: Navigation system

[0227] 601: Path information storage unit

[0228] 602: Scene storage unit

[0229] 603: Security rules storage department

[0230] 61: Target input unit

[0231] 62: Path Planning Department

[0232] 63: Environmental Acquisition Department

[0233] 64: Behavior Planning Department

[0234] 641: Scene Acquisition Department

[0235] 642: Scenario Verification Department

[0236] 65: Action Planning Department

[0237] 651: Action Decision Department

[0238] 652: Action Verification Department

[0239] 66: Motion control unit.

Claims

1. A security rule generating device, comprising: An input unit configured to receive input of scenario information representing a predetermined scenario by linking sub-scenarios that associate control strategies with purposes; a priori condition calculation unit configured to calculate, for each of the sub-scenario, a priori condition for executing the control strategy while achieving the objective; and The output unit is configured to output safety rule information that combines the prior conditions and the control strategy of each sub-scenario.

2. The security rule generating device according to claim 1, wherein: The stated purpose includes the specified safety conditions as well as the a posteriori conditions, The a priori condition calculation unit is configured to calculate the a priori condition that can satisfy the a posteriori condition by executing the control strategy while satisfying the safety condition.

3. The security rule generating device according to claim 2, wherein: The priori condition calculation unit is configured to calculate the priori condition of the sub-scenario so that the priori condition of the subsequent sub-scenario is satisfied.

4. The security rule generating device according to claim 3, wherein: The priori condition calculation unit is configured to sequentially calculate the priori condition by tracing the links of the sub-scenario.

5. The security rule generating device according to claim 2, wherein: The priori condition calculation unit is configured to calculate the priori condition using a program verifier that verifies that the safety condition is maintained during execution of the control strategy.

6. A navigation system is a navigation system for an autonomous driving vehicle, comprising: A scenario storage unit storing scenarios for causing the autonomous driving vehicle to operate; A safety rule storage unit storing safety rules, wherein the safety rules calculate, for each sub-scenario divided from the scene, a priori conditions for executing a control strategy while achieving a purpose of the sub-scenario; an environment acquisition unit that acquires environment information representing the environment of the autonomous driving vehicle; an action determination unit, which determines a navigation action for achieving a navigation goal based on the scene and the environmental information; as well as An action verification unit verifies whether the control strategy corresponding to the navigation action can achieve the navigation target based on the safety rule and the environmental information.

7. A navigation system is a navigation system for an autonomous driving vehicle, comprising: A scenario storage unit storing a plurality of scenarios for causing the autonomous driving vehicle to operate; A safety rule storage unit storing safety rules, wherein the safety rules calculate, for each sub-scenario divided from the scene, a priori conditions for executing a control strategy while achieving a purpose of the sub-scenario; an environment acquisition unit that acquires environment information representing the environment of the autonomous driving vehicle; a scenario acquisition unit that acquires the scenario for achieving the navigation goal input by the user; as well as A scenario verification unit verifies whether the scenario acquired by the scenario acquisition unit can achieve the navigation goal based on the safety rule and the environmental information.

8. A method for generating security rules, wherein: The computer performs the following process: An input process for accepting input of scenario information representing a specified scenario by linking sub-scenarios that establish correspondence between control strategies and objectives; A priori condition calculation process, for each of the sub-scenario, calculating a priori conditions for executing the control strategy while achieving the purpose; as well as The output process outputs the safety rule information formed by combining the prior conditions and the control strategy of each sub-scenario.

9. A navigation method is a navigation method for an autonomous driving vehicle, wherein: The scene storage unit stores scenes for causing the autonomous driving vehicle to move. The safety rule storage unit stores safety rules, which calculate, for each sub-scenario divided from the scene, a priori conditions for executing a control strategy while achieving a purpose of the sub-scenario. The computer performs the following process: An environment acquisition process, acquiring environmental information representing the environment of the autonomous driving vehicle; An action decision process, based on the scene and environment information, determines the navigation action for achieving the navigation goal; as well as The action verification process verifies whether the control strategy corresponding to the navigation action can achieve the navigation goal based on the safety rules and the environmental information.

10. A navigation method is a navigation method for an autonomous driving vehicle, wherein: A plurality of scenarios for causing the autonomous driving vehicle to operate are stored in the scenario storage unit. The safety rule storage unit stores safety rules, which calculate, for each sub-scenario divided from the scene, a priori conditions for executing a control strategy while achieving a purpose of the sub-scenario. The computer performs the following process: An environment acquisition process, acquiring environmental information representing the environment of the autonomous driving vehicle; A scene acquisition process, acquiring the scene used to achieve the navigation goal input by the user; as well as The scene verification process verifies whether the scene obtained by the scene acquisition process can achieve the navigation goal based on the safety rules and the environmental information.

11. A program for causing a computer to execute the following process: An input process for accepting input of scenario information representing a specified scenario by linking sub-scenarios that establish correspondence between control strategies and objectives; A priori condition calculation process, for each of the sub-scenario, calculating a priori conditions for executing the control strategy while achieving the purpose; as well as The output process outputs the safety rule information formed by combining the prior conditions and the control strategy of each sub-scenario.

12. A program for an autonomous vehicle, wherein: The scene storing unit stores scenes for causing the autonomous driving vehicle to move. The safety rule storage unit stores safety rules, which calculate, for each sub-scenario divided from the scene, a priori conditions for executing a control strategy while achieving a purpose of the sub-scenario. The program causes the computer to execute the following process: An environment acquisition process, acquiring environmental information representing the environment of the autonomous driving vehicle; An action decision process, based on the scene and environment information, determines the navigation action for achieving the navigation goal; as well as The action verification process verifies whether the control strategy corresponding to the navigation action can achieve the navigation goal based on the safety rules and the environmental information.

13. A program for an autonomous vehicle, wherein: A plurality of scenarios for causing the autonomous driving vehicle to operate are stored in the scenario storage unit. The safety rule storage unit stores safety rules, which calculate, for each sub-scenario divided from the scene, a priori conditions for executing a control strategy while achieving a purpose of the sub-scenario. The program causes the computer to execute the following process: An environment acquisition process, acquiring environmental information representing the environment of the autonomous driving vehicle; A scene acquisition process, acquiring the scene used to achieve the navigation goal input by the user; as well as The scene verification process verifies whether the scene obtained by the scene acquisition process can achieve the navigation goal based on the safety rules and the environmental information.

Citation Information

Patent Citations

  • Refill container

    JP2022069068A