Judgment method and device for autonomous decision-making of unmanned ship

Abstract

The invention provides an unmanned ship autonomous decision judgment method and device, and the method comprises the steps: determining interference data corresponding to an interference behavior according to the interference behavior, and constructing a test sample library corresponding to the interference data through a multi-dimensional attack algorithm; performing sample difference degree calculation on the data in the test sample library to obtain a calculation result; the autonomous decision-making ability of the unmanned ship is judged according to a calculation result, starting with an intelligent grading characterization method of task completion ability and an autonomous algorithm safety characterization method, an autonomous unmanned system cognitive ability test and evaluation process and an autonomous algorithm characterization index system are designed, the autonomous algorithm safety characterization method is studied, and the autonomous unmanned system cognitive ability test and evaluation process is established. An autonomous algorithm polymorphic behavior exploration and robustness representation evaluation method and an unmanned system autonomous intelligent algorithm failure behavior exploration test and evaluation method in an uncertain environment are provided, and a relatively complete autonomous unmanned system cognitive ability and decision behavior representation theoretical system framework including processes, indexes and methods is formed.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned boat cognition and decision-making, and in particular relates to a judgment method and device for autonomous decision-making of an unmanned boat. Background Art

[0002] The current application system form is transitioning from "human control" to "intelligent autonomy", and the role of unmanned systems is becoming more and more prominent. Due to the unmanned, complex tasks and high confrontation environment, unmanned systems must have high autonomy. Scientifically evaluating the reliability of unmanned systems and proposing evaluation methods suitable for the technical characteristics and usage characteristics of unmanned systems can provide a reasonable basis for the division of labor between man and machine, which is an urgent need. At present, the comprehensive evaluation of the reliability of intelligent autonomous systems at home and abroad is in its infancy. Looking at the reliability evaluation models of foreign intelligent autonomous systems, the more typical ones are Sheridan's Levels of Automation (LOA), Autonomous Control Level (ACL), Autonomy Levels for Unmanned Systems (ALFUS), the four-level model of human-machine authority and the autonomous system reference framework, and the ODAS (Observation, Decision, Action, Security) autonomous capability evaluation model. Scholars represented by Huang et al. in the United States have conducted a relatively systematic study on the reliability level framework of unmanned systems; in addition, the United States' research results in the standards of intelligent autonomous platforms, the development roadmap of autonomous control levels of drones, and task planning of different autonomy levels are at the forefront of the world. In recent years, scholars from Europe, Japan and other countries or regions have conducted in-depth research on the reliability level of unmanned systems or robot systems. With the rise of the research boom of intelligent autonomous control in unmanned systems abroad, the reliability evaluation method of intelligent autonomous systems has also attracted the attention of Chinese scholars, and articles on reliability level, autonomous control and autonomous capabilities, and autonomy-related issues have been published continuously. In the second Sino-Russian UAV academic seminar in 2005, experts in the domestic academic community paid attention to the reliability level of autonomous control of UAVs. Since then, the first China Navigation, Guidance and Control Academic Conference in 2007, the China Automation Conference and the Summit Conference on the Integration of Industrialization and Information Technology in 2009, the Fourth China Navigation, Guidance and Control Academic Conference in 2010, and the 2010 World Congress on Intelligent Control and Automation have all published research work on autonomous control and autonomy reliability.In addition, domestic scholars have organized and held a number of special seminars related to the reliability of intelligent control and autonomous control, such as the "Mobile Robot / Unmanned System Autonomous Behavior" high-level forum hosted by the Shenyang Institute of Automation of the Chinese Academy of Sciences in August 2009, the "Autonomous System Intelligent Control" seminar hosted by the Intelligent Automation Professional Committee of the Chinese Automation Society in August 2010, and the "UAV Advanced Control Technology" seminar held at the Fourth China Navigation, Guidance and Control Academic Conference in October 2010. These seminars have extremely richly displayed and exchanged the latest research results in the field of autonomous research in my country. Existing research shows that effective evaluation of the reliability of unmanned systems or formulation of relevant standards has a leading role in technology.

[0003] Looking at the current reliability evaluation methods for intelligent autonomous systems, there are mainly the following: grade method, double-axis method, three-axis method, table lookup method and formula method.

[0004] The grading method usually divides the reliability of the system into different levels according to the performance of the system. These levels are clearly quantified coordinate values ​​such as 0 to 10 or 1 to 10. The more typical ones are Sheridan's automatic device level LOA and NASA's autonomy level for its aircraft system. Therefore, how to provide a judgment method for autonomous decision-making of unmanned boats has become a technical problem that needs to be solved urgently in this field. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for determining autonomous decision-making of an unmanned boat.

[0006] According to a first aspect of the present invention, there is provided a method for determining autonomous decision-making of an unmanned boat, comprising:

[0007] Determining interference data corresponding to the interference behavior according to the interference behavior, wherein the interference data at least includes polymorphic sample space data of natural extreme environments and active malicious behavior data;

[0008] Using a multi-dimensional attack algorithm, construct a test sample library corresponding to the interference data;

[0009] Performing sample difference calculation on the data in the test sample library to obtain a calculation result;

[0010] The autonomous decision-making capability of the unmanned boat is judged based on the calculation results.

[0011] Optionally, different hardware carried on the unmanned boat is used to install one of a navigation system, a perception system, a propulsion system, a communication system or a public service system, and judging the autonomous decision-making capability of the unmanned boat according to the calculation result includes:

[0012] Based on the calculation results, the results of the execution of any of the navigation system, perception system, propulsion system, communication system or public service system on the unmanned boat are judged.

[0013] Optionally, the method further comprises:

[0014] Use the route planning system to obtain obstacle information of the sea area that the voyage will pass through from the electronic nautical chart, including coastlines, islands, reefs and sunken ships;

[0015] A star algorithm, distance optimization Dijkstra algorithm, genetic algorithm and artificial potential field method are used to find a collision-free path from the starting point to the end point.

[0016] Optionally, the method further comprises:

[0017] During the navigation of the unmanned boat, the course and speed are adjusted according to the data detected by the sensors, and local dangers and obstacles are avoided, so that the unmanned boat can still reach the target smoothly.

[0018] Optionally, the adopting a multi-dimensional attack algorithm to construct a test sample library corresponding to the interference data includes:

[0019] Rotational test profile construction: based on the relative position change between the intelligent system platform and the object caused by the bumping, the rotational test profile data is constructed;

[0020] Construct scaled test profile data based on the distance between objects;

[0021] Based on the perspective relationship between objects, construct perspective test profile data;

[0022] Based on the composition defects of the imaging system and the disturbance of the optical medium, the distortion test profile data is constructed.

[0023] According to a second aspect of the present invention, there is provided a judgment device for autonomous decision-making of an unmanned boat, comprising:

[0024] A determination module, configured to determine interference data corresponding to the interference behavior according to the interference behavior, wherein the interference data at least includes polymorphic sample space data of natural extreme environments and active malicious behavior data;

[0025] A construction module, used to construct a test sample library corresponding to the interference data using a multi-dimensional attack algorithm;

[0026] A calculation module, used to calculate the sample difference of the data in the test sample library to obtain a calculation result;

[0027] The judgment module is used to judge the autonomous decision-making ability of the unmanned boat according to the calculation result.

[0028] Optionally, different hardware carried on the unmanned boat is used to install one of a navigation system, a perception system, a propulsion system, a communication system or a public service system, and the judgment module is used to:

[0029] Based on the calculation results, the results of the execution of any of the navigation system, perception system, propulsion system, communication system or public service system on the unmanned boat are judged.

[0030] Optionally, the judging module is used to:

[0031] Use the route planning system to obtain obstacle information of the sea area that the voyage will pass through from the electronic nautical chart, including coastlines, islands, reefs and sunken ships;

[0032] A star algorithm, distance optimization Dijkstra algorithm, genetic algorithm and artificial potential field method are used to find a collision-free path from the starting point to the end point.

[0033] Optionally, the judging module is used to:

[0034] During the navigation of the unmanned boat, the course and speed are adjusted according to the data detected by the sensors, and local dangers and obstacles are avoided, so that the unmanned boat can still reach the target smoothly.

[0035] Optionally, the building block is used to:

[0036] Rotational test profile construction: based on the relative position change between the intelligent system platform and the object caused by the bumping, the rotational test profile data is constructed;

[0037] Construct scaled test profile data based on the distance between objects;

[0038] Based on the perspective relationship between objects, construct perspective test profile data;

[0039] Based on the composition defects of the imaging system and the disturbance of the optical medium, the distortion test profile data is constructed.

[0040] In a third aspect, the present application shows an electronic device, which includes: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the method described in any of the above aspects.

[0041] In a fourth aspect, the present application illustrates a non-temporary computer-readable storage medium, which, when instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to execute a method as described in any of the above aspects.

[0042] In a fifth aspect, the present application illustrates a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to execute the method described in any of the above aspects.

[0043] The beneficial effects brought by the present invention are as follows:

[0044] It can be seen from the above scheme that an embodiment of the present invention provides a method and device for judging autonomous decision-making of an unmanned boat, including: determining interference data corresponding to the interference behavior according to the interference behavior, the interference data at least including polymorphic sample space data of natural extreme environments and active malicious behavior data; using a multidimensional attack algorithm to construct a test sample library corresponding to the interference data; performing sample difference calculation on the data in the test sample library to obtain a calculation result; judging the autonomous decision-making ability of the unmanned boat according to the calculation result, starting from the intelligent grading characterization method of task completion ability and the autonomous algorithm security characterization method, designing an autonomous unmanned system cognitive ability test and evaluation process and an autonomous algorithm characterization index system, studying the autonomous algorithm security characterization method, proposing an autonomous algorithm polymorphic behavior exploration and robustness characterization evaluation method under uncertain environment and an unmanned system autonomous intelligent algorithm failure behavior exploration test and evaluation method, forming a relatively complete autonomous unmanned system cognitive ability and decision-making behavior characterization theoretical system framework including process, indicators, and methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A schematic diagram of a flow chart of a method for determining autonomous decision-making of an unmanned boat provided according to an embodiment;

[0046] Figure 2 A schematic diagram of the composition of an unmanned boat system provided according to an embodiment;

[0047] Figure 3 A schematic diagram of the overall design of an autonomous cognitive ability and a method for characterizing its decision-making behavior provided according to an embodiment;

[0048] Figure 4 This is a structural block diagram of a judgment device for autonomous decision-making of an unmanned boat in the present application.

[0049] Figure 5 It is a block diagram of an electronic device of the present application.

[0050] Figure 6 It is a block diagram of a computer-readable storage medium of the present application. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Reference Figure 1 , shows a flowchart of the steps of a method for determining autonomous decision-making of an unmanned boat of the present application, which can be applied to electronic equipment, wherein the method can specifically include the following steps:

[0053] S101. Determine interference data corresponding to the interference behavior according to the interference behavior, where the interference data at least includes polymorphic sample space data of natural extreme environments and active malicious behavior data;

[0054] S102, using a multi-dimensional attack algorithm to construct a test sample library corresponding to the interference data;

[0055] S103, performing sample difference calculation on the data in the test sample library to obtain a calculation result;

[0056] S104. Determine the autonomous decision-making capability of the unmanned boat based on the calculation results.

[0057] The overall design of the autonomous cognitive ability of the surface WR system and the characterization method of its decision-making behavior is shown in the figure below. By combining the three typical cases of maritime transportation, XLJJ, and maritime rescue with the intelligent system testing requirements, three typical tasks of target recognition, route planning, and autonomous obstacle avoidance are selected as autonomous ability test targets.

[0058] like Figure 3 As shown in the figure, unlike the traditional traversal test method, in order to improve the test efficiency and efficiently discover the failure behavior, the failure case analysis found two major causes that affect the failure of the autonomous algorithm: natural environment interference and human malicious attacks. The autonomous algorithm performs well in an ideal experimental environment, but it often fails in actual use. After a large number of case analyses and literature research, the consensus formed is that the WR system will inevitably cause interference when interacting with the environment during operation, such as bumps, light, climate, etc. Another major type of failure comes from the targeted attack algorithm proposed in the field of counterattack. This field actively launches attacks on autonomous systems through methods such as data poisoning and model reversal.

[0059] In combination with these two types of failure mechanisms, the embodiment of the present application proposes to construct a test sample library using a polymorphic sample space that simulates natural extreme environments and a multidimensional attack algorithm that simulates active malicious behaviors. The task difficulty is evaluated through sample differences, and the task capability is evaluated through task completion status. The task difficulty and task capability are comprehensively considered to objectively and accurately characterize the cognitive ability and decision-making behavior of the autonomous system.

[0060] Starting from the intelligent hierarchical characterization method of task completion capability and the autonomous algorithm safety characterization method, we design the autonomous unmanned system cognitive ability test and evaluation process and the autonomous algorithm characterization indicator system, study the autonomous algorithm safety characterization method, and propose the autonomous algorithm polymorphic behavior exploration and robustness characterization evaluation method under uncertain environment and the unmanned system autonomous intelligent algorithm failure behavior exploration test and evaluation method, to form a relatively complete theoretical system framework for the characterization of autonomous unmanned system cognitive ability and decision-making behavior including process, indicators and methods.

[0061] Another embodiment of the present application further supplements the method for determining autonomous decision-making of the unmanned boat provided in the above embodiment.

[0062] Optionally, different hardware carried on the unmanned boat is used to install one of a navigation system, a perception system, a propulsion system, a communication system or a public service system, and the autonomous decision-making capability of the unmanned boat is judged according to the calculation result, including:

[0063] Based on the calculation results, the results of the execution of any of the navigation system, perception system, propulsion system, communication system or public service system on the unmanned boat are judged.

[0064] As an unmanned surface system, WRT is mainly used to perform dangerous tasks that are not suitable for manned ships. Once equipped with advanced control systems, sensor systems, communication systems and weapon systems, it can perform a variety of war and non-war military tasks, such as reconnaissance, search, detection and mine clearance; search and rescue, navigation and hydrographic surveys; anti-submarine warfare, anti-special operations, patrols, anti-piracy, anti-terrorism attacks, etc.

[0065] like Figure 2As shown in the figure, the hardware on the unmanned boat is classified into five systems according to its functionality: navigation system, perception system, propulsion system, communication system, and public service system. The navigation system is a combination of navigation equipment on the boat, consisting of navigation radar, heading equipment, Beidou integrated machine, differential GPS, meteorological instrument, etc.; the perception system is a system for the unmanned boat to monitor the environment, identify and track targets, consisting of laser radar, monitoring camera, optoelectronic tracker, etc.; the propulsion system is a combination of power, transmission, and fuel devices of the unmanned boat, consisting of engine device, fuel device, transmission device, propeller / water pump, etc.; the communication system is a combination of hardware for communication functions between unmanned boats and between boats and ground, consisting of receiving channel, local oscillator source, transmitting channel, L-end processing unit, and antenna servo device; the public service system is a boat-borne computing, storage, and interaction center, consisting of unified computing center, unified data storage, unified high-speed network, unified software interface, unified hardware interface, unified hardware and power management, unified packaging, and time system.

[0066] The goal of maritime defense is to ensure the safety of the target point and prevent it from being destroyed. Maritime defense is a defense action to ensure the security of the country and the destruction of the target point by external personnel. In order to ensure the safety of the target point, a shore-based monitoring system and ship patrol operations are generally set up at the target point. The shore-based monitoring system monitors the environmental changes in a large range around the target point in real time. Once an abnormality is found, it will immediately track and observe the maritime target, contact other ships through the communication system, warn and prevent the intention to enter the port. The remote warning relying solely on the shore-based monitoring system is sometimes not enough to achieve the purpose of expulsion. The unmanned boat patrol mission can approach the suspicious target and check the target purpose at the same time, and then track, drive away, and strike it. Maritime military strength is also a concrete manifestation of national defense strength. Security defense is the first step in maritime operations. Only by ensuring the safety and unobstructedness of target points, such as ports and bays, can subsequent force output and logistical support be provided. Maintaining maritime security in peacetime is a manifestation of maintaining national security, avoiding the deliberate destruction of maritime facilities, preventing the leakage of maritime information, and safeguarding maritime interests. In wartime, the sea is a channel for the output of troops. Warships, submarines and other warships as well as supply ships need to enter and exit here. The integrity of maritime facilities and the safety and smoothness of waterways will affect the war situation. Therefore, no matter at any time, maritime defense must be done at all times.

[0067] The OODA loop, also known as the Boyd loop, is based on the idea that armed conflict can be seen as a competition between two hostile parties to see who can complete the "observation-adjustment-decision-action" cycle faster and better. The OODA loop is named after the combination of the first letters of the four words: observation, orientation, decision, and action. Based on the idea of ​​OODA, typical tasks are divided into four stages: patrol-observation, identification-positioning, assignment-decision-making, and tracking-collision-action. In the patrol stage, the task ship circles around a protection point, involving the propulsion system, communication system, and public service system on the boat; in the identification stage, the enemy ship appears in the mission area, and the patrolling task ship identifies the enemy ship, involving the perception system and public service system on the boat; in the assignment stage, our algorithm assigns some ships to encircle the enemy ship, involving the communication system and public service system on the boat; in the tracking-collision stage, our task ship tracks and collides with the designated enemy ship, involving the navigation system, propulsion system, and public service system on the boat.

[0068] The WR system's automatic detection and identification of the marine environment and surrounding ships is of great significance to reducing the frequency of ship accidents and enhancing my country's maritime defense capabilities. Due to the harsh and unpredictable marine environment, coupled with the small platform of the unmanned boat itself, it is extremely unstable in the vast ocean, making the research on the visual target recognition system of the unmanned boat a challenging and arduous task. Testing the target recognition function is of great significance to improving the visual intelligence of the WR system.

[0069] Optionally, the method further comprises:

[0070] Use the route planning system to obtain obstacle information of the sea area that the voyage will pass through from the electronic nautical chart, including coastlines, islands, reefs and sunken ships;

[0071] A star algorithm, distance optimization Dijkstra algorithm, genetic algorithm and artificial potential field method are used to find a collision-free path from the starting point to the end point.

[0072] The WRT route planning system obtains obstacle information of the sea area that the voyage will pass through from the electronic nautical chart, including coastlines, islands, reefs and shipwrecks. Then it uses methods such as A-star algorithm, distance optimization Dijkstra algorithm, genetic algorithm and artificial potential field method to find a collision-free path from the starting point to the end point. The route planning function dominates the navigation behavior, and testing it can effectively verify the intelligent autonomy capability of the WR system.

[0073] Optionally, the method further comprises:

[0074] During the navigation of the unmanned boat, the course and speed are adjusted according to the data detected by the sensors, and local dangers and obstacles are avoided, so that the unmanned boat can still reach the target smoothly.

[0075] During the navigation process of WRT, it is necessary to respond to some unpredictable obstacles, adjust the course and speed according to the data detected by the sensors, and avoid local dangers so that the unmanned boat can still reach the target smoothly. Therefore, local danger avoidance must be fast, real-time and efficient. The autonomous obstacle avoidance function is the fundamental guarantee for the success rate of WRT missions. Testing it can effectively understand the underlying safety capabilities of WRT.

[0076] Optionally, a multi-dimensional attack algorithm is used to construct a test sample library corresponding to the interference data, including:

[0077] Rotational test profile construction: based on the relative position change between the intelligent system platform and the object caused by the bumping, the rotational test profile data is constructed;

[0078] Construct scaled test profile data based on the distance between objects;

[0079] Based on the perspective relationship between objects, construct perspective test profile data;

[0080] Based on the composition defects of the imaging system and the disturbance of the optical medium, the distortion test profile data is constructed.

[0081] Target identification test profile, including:

[0082] 1) Rotation test profile construction

[0083] The bumps of the intelligent system platform cause the relative position between it and the object to change, which causes the image rotation and has a certain impact on the running results. It is planned to generate test data from 12 sampling points with a sampling point every 300.

[0084]

[0085] 2) Scaling test profile construction

[0086] The distance between the object and the object poses a great challenge to the working condition of the intelligent system platform in actual deployment. Exploring the limit distance at which the model can work normally is a good way to understand the failure boundary of the model. It is planned to take 9 sampling points, 9 / 10, 8 / 10, ..., 1 / 10, from the object that fills the image to generate test data.

[0087]

[0088] 3) Viewing angle test profile construction

[0089] The perspective relationship between objects poses a great challenge to the working condition of the intelligent system platform in actual deployment. Exploring the extreme perspective at which the model can work normally is a good way to understand the failure boundary of the model. It is proposed to take 10 sampling points from the observation perspective to generate test data.

[0090]

[0091] 4) Distortion test profile construction

[0092] Due to the defects in the imaging system and the disturbance of the optical medium, distortion occurs in almost all data under real working conditions, and the characteristic changes it brings will have a certain impact on the results. It is proposed to include six items including radial distortion and tangential distortion, and each item has 10 sampling points to generate data for a total of 60 sets of test data.

[0093] (1) Positive barrel distortion

[0094] x′=x(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0095] y′=y(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0096] r 2 =x 2 +y 2

[0097] k 1 >0,k 2 >0

[0098] (2) Negative barrel distortion

[0099] x′=x(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0100] y′=y(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0101] r 2 =x 2 +y 2

[0102] k 1 >0,k2 <0

[0103] (3) Positive pincushion distortion

[0104] x′=x(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0105] y′=y(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0106] r 2 =x 2 +y 2

[0107] k 1 <0,k 2 >0

[0108] (4) Negative pincushion distortion

[0109] x′=x(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0110] y′=y(1+k 1 r 2 +k 2 r 4 +k 3 r 6 )

[0111] r 2 =x 2 +y 2

[0112] k 1 <0,k 2 <0

[0113] (5) Tangential distortion

[0114] x′=x+[2p 1 xy+p 2 (r 2 +2x 2 )]

[0115] y′=y+[2p 2 xy+p1 (r 2 +2y 2 )]

[0116] r 2 =x 2 +y 2

[0117] p 1 >0,p 2 >0

[0118] (6) Negative tangential distortion

[0119] x′=x+[2p 1 xy+p 2 (r 2 +2x 2 )]

[0120] y′=y+[2p 2 xy+p 1 (r 2 +2y 2 )]

[0121] r 2 =x 2 +y 2

[0122] p 1 <0,p 2 <0

[0123] 1. Untargeted white box single-step L0 attack

[0124] The white-box single-step attack based on the saliency map method uses the model weight to generate point noise so that the positive sample can be identified as any other result by the intelligent algorithm. The attack principle is as follows:

[0125] f(x true )=y true

[0126]

[0127] subject to f(x′ true )=y anyother， min||x′ true -x true || 0

[0128] 2. Untargeted White Box Iterative L0 Attack

[0129] The white-box iterative attack based on the saliency map method uses the model weight to generate point noise so that the positive sample can be identified as any other result by the intelligent algorithm. The attack principle is as follows:

[0130] f(x true )=y true

[0131] while num <step:

[0132]

[0133] subject to f(x′ true )=y anyother， min||x′ true -x true || 0

[0134] 3. Untargeted White-box Single-step L∞ Attack

[0135] A white-box single-step attack based on a fast gradient method is used to generate global noise with the help of model weights so that positive samples can be identified as any other results by the intelligent algorithm. The attack principle is as follows:

[0136] f(x true )=y true

[0137]

[0138] subject to f(x′ true )=y anyother， min||x′ true -x true || 0

[0139] 4. Untargeted White-Box Iterative L∞ Attack

[0140] The white-box iterative attack based on the fast gradient method generates global noise with the help of the model weights so that the positive samples are recognized as any other results by the intelligent algorithm. The attack principle is as follows:

[0141] f(x true )=y true

[0142] while i <step:

[0143]

[0144] subject to f(x′ true )=y anyother， min||x′ true -x true || 0

[0145] 5. Targeted White Box Single-Step L0 Attack

[0146] The white-box single-step attack based on the saliency map method uses the model weight to generate point noise so that the negative samples are recognized as positive examples by the intelligent algorithm. The attack principle is as follows:

[0147] f(x false )=y false

[0148]

[0149] subject to f(x′ false )=y true ,min||x′ false -x false || 0

[0150] 6. Targeted White Box Iterative L0 Attack

[0151] The white-box iterative attack based on the saliency map method uses the model weight to generate point noise so that the positive sample is identified as a negative example by the intelligent algorithm. The attack principle is as follows:

[0152] f(x true )=y true

[0153] while num <step:

[0154]

[0155] subject to f(x′ true )=y false ,min||x′ true -x true || 0

[0156] 7. Targeted White-Box Single-Step L∞ Attack

[0157] The white-box single-step attack based on the gradient fast method generates global noise with the help of the model weight so that the negative samples are recognized as positive examples by the intelligent algorithm. The attack principle is as follows:

[0158] f(x false )=y false

[0159]

[0160] subject to f(x′ false )=y true ,min||x′ false -x false || 0

[0161] 8. Targeted White-Box Iterative L∞ Attack

[0162] The white-box iterative attack based on the fast gradient method generates global noise with the help of the model weights so that the positive samples are identified as negative examples by the intelligent algorithm. The attack principle is as follows:

[0163] f(x true )=y true

[0164] while i <step:

[0165]

[0166] subject to f(x′ true )=y false ,min||x′ true -x true || 0

[0167] 9. Untargeted black box single-step L0 attack

[0168] The substitute model is obtained through gradient transfer training, and adversarial attacks are carried out through the substitute model. A white-box single-step attack based on the saliency map method is used to generate point noise with the help of the model weight so that the positive sample is recognized as any other result by the intelligent algorithm. The attack principle is as follows:

[0169] f(x true )=y true

[0170]

[0171] subject to f(x′ true )=y anyother ,min||x′ true -x true || 0

[0172] 10. Untargeted black box iterative L0 attack

[0173] The substitute model is obtained through gradient transfer training, and adversarial attacks are carried out through the substitute model. The white-box iterative attack based on the saliency map method is used to generate point noise with the help of the model weight so that the positive sample is recognized as any other result by the intelligent algorithm. The attack principle is as follows:

[0174] f(x true )=y true

[0175] while num <step:

[0176]

[0177] subject to f(x′ true )=y anyother ,min||x′ true -x true || 0

[0178] 11. Untargeted Black Box Single-step L∞ Attack

[0179] Through gradient transfer training, a substitute model is obtained, and adversarial attacks are carried out through the substitute model. A white-box single-step attack based on a gradient fast method is used to generate global noise with the help of model weights so that positive samples are recognized as any other results by the intelligent algorithm. The attack principle is as follows:

[0180] f(x true )=y true

[0181]

[0182] subject to f(x′ true )=y anyother ,min||x′ true -x true || 0

[0183] 12. Untargeted Black Box Iterative L∞ Attack

[0184] Through gradient transfer training, a substitute model is obtained, and adversarial attacks are carried out through the substitute model. A white-box iterative attack based on a gradient fast method is used to generate global noise with the help of model weights so that positive samples are recognized as any other results by the intelligent algorithm. The attack principle is as follows:

[0185] f(x true )=y true

[0186] while i <step:

[0187]

[0188] subject to f(x′ true )=y anyother ,min||x′ true -x true || 0

[0189] 13. Targeted Black Box Single-Step L0 Attack

[0190] The substitute model is obtained through gradient transfer training, and adversarial attacks are carried out through the substitute model. The white-box single-step attack based on the saliency map method is used to generate point noise with the help of the model weight so that the negative sample is recognized as a positive sample by the intelligent algorithm. The attack principle is as follows:

[0191] f(x false )=y false

[0192]

[0193] subject to f(x′ false )=y true ,min||x′ false -x false || 0

[0194] 14. Targeted Black Box Iterative L0 Attack

[0195] The substitute model is obtained through gradient transfer training, and adversarial attacks are carried out through the substitute model. The white-box iterative attack based on the saliency map method is used to generate point noise with the help of the model weight so that the positive sample is recognized as a negative example by the intelligent algorithm. The attack principle is as follows:

[0196] f(x true )=y true

[0197] while num <step:

[0198]

[0199] subject to f(x′ true )=y false ,min||x′ true -x true || 0

[0200] 15. Targeted Black Box Single-step L∞ Attack

[0201] Through gradient transfer training, a substitute model is obtained, and adversarial attacks are carried out through the substitute model. A white-box single-step attack based on a gradient fast method is used to generate global noise with the help of model weights so that negative samples are recognized as positive examples by the intelligent algorithm. The attack principle is as follows:

[0202] f(x false )=y false

[0203]

[0204] subject to f(x′ false )=ytrue ,min||x′ false -x false || 0

[0205] 16. Targeted Black Box Iterative L∞ Attack

[0206] Through gradient transfer training, a substitute model is obtained, and adversarial attacks are carried out through the substitute model. A white-box iterative attack based on a gradient fast method is used to generate global noise with the help of model weights so that positive samples are identified as negative examples by the intelligent algorithm. The attack principle is as follows:

[0207] f(x true )=y true

[0208] while i <step:

[0209]

[0210] subject to f(x′ true )=y false ,min||x′ true -x true || 0

[0211] An embodiment of the present invention provides a method for judging autonomous decision-making of an unmanned boat, comprising: determining interference data corresponding to the interference behavior according to the interference behavior, the interference data at least including polymorphic sample space data of natural extreme environments and active malicious behavior data; adopting a multidimensional attack algorithm to construct a test sample library corresponding to the interference data; performing sample difference calculation on the data in the test sample library to obtain a calculation result; judging the autonomous decision-making ability of the unmanned boat according to the calculation result, starting from an intelligent grading characterization method of task completion ability and a characterization method of autonomous algorithm security, designing an autonomous unmanned system cognitive ability test and evaluation process and an autonomous algorithm characterization index system, studying the autonomous algorithm security characterization method, proposing an autonomous algorithm polymorphic behavior exploration and robustness characterization evaluation method under uncertain environment and an unmanned system autonomous intelligent algorithm failure behavior exploration test and evaluation method, forming a relatively complete autonomous unmanned system cognitive ability and decision-making behavior characterization theoretical system framework including processes, indicators and methods.

[0212] It should be noted that each implementable method in this embodiment may be implemented separately, or may be implemented in combination in any combination without conflict, and this application is not limited thereto.

[0213] Another embodiment of the present application provides a judgment device for autonomous decision-making of an unmanned boat, which is used to execute the judgment method for autonomous decision-making of an unmanned boat provided in the above embodiment.

[0214] like Figure 4 , which is a schematic diagram of the structure of the judgment device for autonomous decision-making of an unmanned boat provided in an embodiment of the present application. The judgment device for autonomous decision-making of an unmanned boat includes a determination module 401, a construction module 402, a calculation module 403 and a judgment module 404, wherein:

[0215] The determination module 401 is used to determine interference data corresponding to the interference behavior according to the interference behavior, and the interference data at least includes polymorphic sample space data of natural extreme environment and active malicious behavior data;

[0216] The construction module 402 is used to construct a test sample library corresponding to the interference data using a multi-dimensional attack algorithm;

[0217] The calculation module 403 is used to calculate the sample difference of the data in the test sample library to obtain the calculation result;

[0218] The judgment module 404 is used to judge the autonomous decision-making capability of the unmanned boat according to the calculation results.

[0219] Regarding the device in this embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0220] Another embodiment of the present application further supplements the description of the judgment device for autonomous decision-making of the unmanned boat provided in the above embodiment.

[0221] Optionally, different hardware carried on the unmanned boat is used to install one of a navigation system, a perception system, a propulsion system, a communication system or a public service system, and the judgment module is used to:

[0222] Based on the calculation results, the results of the execution of any of the navigation system, perception system, propulsion system, communication system or public service system on the unmanned boat are judged.

[0223] Optionally, the judging module is used to:

[0224] Use the route planning system to obtain obstacle information of the sea area that the voyage will pass through from the electronic nautical chart, including coastlines, islands, reefs and sunken ships;

[0225] A star algorithm, distance optimization Dijkstra algorithm, genetic algorithm and artificial potential field method are used to find a collision-free path from the starting point to the end point.

[0226] Optionally, the judging module is used to:

[0227] During the navigation of the unmanned boat, the course and speed are adjusted according to the data detected by the sensors, and local dangers and obstacles are avoided, so that the unmanned boat can still reach the target smoothly.

[0228] Optionally, build modules for:

[0229] Rotational test profile construction: based on the relative position change between the intelligent system platform and the object caused by the bumping, the rotational test profile data is constructed;

[0230] Construct scaled test profile data based on the distance between objects;

[0231] Based on the perspective relationship between objects, construct perspective test profile data;

[0232] Based on the composition defects of the imaging system and the disturbance of the optical medium, the distortion test profile data is constructed.

[0233] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0234] An embodiment of the present invention provides a judgment device for autonomous decision-making of an unmanned boat, comprising: determining interference data corresponding to the interference behavior according to the interference behavior, the interference data at least including polymorphic sample space data of natural extreme environments and active malicious behavior data; using a multidimensional attack algorithm to construct a test sample library corresponding to the interference data; performing sample difference calculation on the data in the test sample library to obtain a calculation result; judging the autonomous decision-making ability of the unmanned boat according to the calculation result, starting from an intelligent grading characterization method of task completion ability and a characterization method of autonomous algorithm security, designing an autonomous unmanned system cognitive ability test and evaluation process and an autonomous algorithm characterization index system, studying the autonomous algorithm security characterization method, proposing an autonomous algorithm polymorphic behavior exploration and robustness characterization evaluation method under uncertain environment and an unmanned system autonomous intelligent algorithm failure behavior exploration test and evaluation method, forming a relatively complete autonomous unmanned system cognitive ability and decision-making behavior characterization theoretical system framework including processes, indicators, and methods.

[0235] Optionally, an embodiment of the present application further provides an electronic device, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the various processes of the above-mentioned method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0236] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0237] Figure 5 800 is a block diagram of an electronic device 800 shown in the present application. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0238] Reference Figure 5 , the electronic device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0239] The processing component 802 generally controls the overall operation of the electronic device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0240] The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phone book data, messages, images, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0241] The power supply component 806 provides power to the various components of the electronic device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 800.

[0242] The multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and the rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0243] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the electronic device 800 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

[0244] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: home button, volume button, start button, and lock button.

[0245] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the electronic device 800. For example, the sensor assembly 814 can detect the open / closed state of the device 800, the relative positioning of components, such as the display and keypad of the electronic device 800, and the sensor assembly 814 can also detect the position change of the electronic device 800 or a component of the electronic device 800, the presence or absence of contact between the user and the electronic device 800, the orientation or acceleration / deceleration of the electronic device 800, and the temperature change of the electronic device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0246] The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 can access a wireless network based on a communication standard, such as WiFi, a carrier network (such as 2G, 3G, 4G or 5G), or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast operation information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0247] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above methods.

[0248] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by a processor 820 of an electronic device 800 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0249] Figure 619 is a block diagram of a computer-readable storage medium 1900 shown in the present application. For example, the computer-readable storage medium 1900 may be provided as a server.

[0250] Reference Figure 6 , the computer-readable storage medium 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932 for storing instructions, such as an application, that can be executed by the processing component 1922. The application stored in the memory 1932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above method.

[0251] The computer readable storage medium 1900 may also include a power supply component 1926 configured to perform power management of the computer readable storage medium 1900, a wired or wireless network interface 1950 configured to connect the computer readable storage medium 1900 to a network, and an input / output (I / O) interface 1958. The computer readable storage medium 1900 may operate based on an operating system stored in the memory 1932, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™ or the like.

[0252] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0253] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0254] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0255] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0256] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0257] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0258] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0259] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0260] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0261] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0262] The above are preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for determining autonomous decision-making of an unmanned boat, characterized in that: include: Determining interference data corresponding to the interference behavior according to the interference behavior, wherein the interference data at least includes polymorphic sample space data of natural extreme environments and active malicious behavior data; Using a multi-dimensional attack algorithm, construct a test sample library corresponding to the interference data; Performing sample difference calculation on the data in the test sample library to obtain a calculation result; The autonomous decision-making capability of the unmanned boat is judged based on the calculation results.

2. The method for determining autonomous decision-making of an unmanned boat according to claim 1, characterized in that: Different hardware carried on the unmanned boat, the different hardware being used to install one of a navigation system, a perception system, a propulsion system, a communication system or a public service system, and the autonomous decision-making capability of the unmanned boat being judged according to the calculation result, comprising: Based on the calculation results, the results of the execution of any of the navigation system, perception system, propulsion system, communication system or public service system on the unmanned boat are judged.

3. The method for determining autonomous decision-making of an unmanned boat according to claim 2, characterized in that: The method further comprises: Use the route planning system to obtain obstacle information of the sea area that the voyage will pass through from the electronic nautical chart, including coastlines, islands, reefs and sunken ships; A star algorithm, distance optimization Dijkstra algorithm, genetic algorithm and artificial potential field method are used to find a collision-free path from the starting point to the end point.

4. The method for determining autonomous decision-making of an unmanned boat according to claim 3, characterized in that: The method further comprises: During the navigation of the unmanned boat, the course and speed are adjusted according to the data detected by the sensors, and local dangers and obstacles are avoided, so that the unmanned boat can still reach the target smoothly.

5. The method for determining autonomous decision-making of an unmanned boat according to claim 1, characterized in that: The multi-dimensional attack algorithm is used to construct a test sample library corresponding to the interference data, including: Rotational test profile construction: based on the relative position change between the intelligent system platform and the object caused by the bumping, the rotational test profile data is constructed; Construct scaled test profile data based on the distance between objects; Based on the perspective relationship between objects, construct perspective test profile data; Based on the composition defects of the imaging system and the disturbance of the optical medium, the distortion test profile data is constructed.

6. A judgment device for autonomous decision-making of an unmanned boat, characterized in that: include: A determination module, configured to determine interference data corresponding to the interference behavior according to the interference behavior, wherein the interference data at least includes polymorphic sample space data of natural extreme environments and active malicious behavior data; A construction module, used to construct a test sample library corresponding to the interference data using a multi-dimensional attack algorithm; A calculation module, used to calculate the sample difference of the data in the test sample library to obtain a calculation result; The judgment module is used to judge the autonomous decision-making ability of the unmanned boat according to the calculation result.

7. The judgment device for autonomous decision-making of an unmanned boat according to claim 6, characterized in that: The different hardware carried on the unmanned boat is used to install one of a navigation system, a perception system, a propulsion system, a communication system or a public service system, and the judgment module is used to: Based on the calculation results, the results of the execution of any of the navigation system, perception system, propulsion system, communication system or public service system on the unmanned boat are judged.

8. The method for determining autonomous decision-making of an unmanned boat according to claim 7, characterized in that: The judging module is used for: Use the route planning system to obtain obstacle information of the sea area that the voyage will pass through from the electronic nautical chart, including coastlines, islands, reefs and sunken ships; A star algorithm, distance optimization Dijkstra algorithm, genetic algorithm and artificial potential field method are used to find a collision-free path from the starting point to the end point.

9. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the method according to any one of claims 1 to 5 when executed by the processor.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images