A METHOD FOR AN ALGORITHM TO AVOID KAMIKAZE UNMANNED SEA VEHICLES

Abstract

The invention relates to a method for evading attacking kamikaze unmanned underwater vehicles (UAVs). This method enables the platforms targeted by the KIDA (Knowledge-Based Surveillance System) to perform optimal avoidance maneuvers in dynamic and uncertain surface environments, thanks to a reinforcement learning-based route generation and autonomous control AI model developed using sensor data. The method analyzes the movements of the threatening KIDA in real-time and optimizes avoidance decisions autonomously by fusion of data from environmental sensors (LIDAR, mm-wave radar, etc.) and inertial measurement units.

Description

1 TARIFF KAMIKAZE UNMANNED SEA VEHICLE AVOIDANCE A METHOD FOR THE ALGORITHM Technical Area The invention provides 5 ways to evade attacking kamikaze unmanned underwater vehicles (UAVs). It is related to a method. This method enables KIDA to operate in dynamic and uncertain surface environments. platforms developed using sensor data and based on reinforcement learning Optimal avoidance thanks to route generation and autonomous control AI model. This enables the maneuvers to be performed. The method uses environmental sensors (LIDAR, 10 fusion of data from mm-wave radar (etc.) and inertial measurement units Using this, it analyzes the movements of threatening KIDA in real time and It autonomously optimizes avoidance decisions. Previous Technique Prior to the method described in this invention, there were 15 aggressive kamikaze unmanned underwater vehicles. (KIDA) avoidance operation is usually performed manually with remote intervention by operators. This was being carried out as follows: Platforms on the sea, in response to the attack. attempting to evade by performing operator-controlled maneuvers was located. Current systems use data from radar and other sensors to identify threats. In addition to providing detection, this data will automate avoidance maneuvers. It was unable to combine the data manually. As a result, the data was compromised during the attack. assessment and determination of defense strategy through operator intervention, It was reducing operational efficiency and was inadequate in preventing threats. 25 2 It remained. The limited autonomy of traditional methods, especially for multiple users, In threat scenarios, avoiding attacks became even more difficult. In the document CN111881580A, which is included in the known state of the art, autonomous There is a multi-sensor fusion model in route generation for marine vessels and 5 This refers to a method that utilizes reinforcement learning. The method described in the document allows an autonomous marine vehicle to navigate through obstacles. It offers a solution for avoiding and planning actions. However, this The solution remains limited to the individual vehicle level and a more restricted sensor. It is supported by a fusion approach. Also, there is no central critic or 10 It does not include a task improvement mechanism. Autonomous according to document CN116225008A in the known state of the art. There is a multi-sensor fusion model in route generation for marine vessels, and A method using reinforcement learning is being discussed. 15 Current applications range from aggressive kamikaze unmanned underwater vehicles (UAVs) It shows various shortcomings and deficiencies in avoidance. Firstly, Because these systems are generally operator-controlled, they reduce human reaction time. It is dependent and slow to deal with real-time threats. 20 remains. Operators' threat detection and avoidance maneuvers The execution time may be insufficient depending on the speed and direction of the attack. This significantly reduces the defensive effectiveness of the platforms. In addition, although current applications detect threats using sensor data, 25 This data will enable effective automated avoidance maneuvers. It cannot be processed in this way. Data from radar and other environmental sensors, Because it requires manual evaluation, operational efficiency decreases and much When faced with so many threats, avoiding attacks is nearly impossible. 3 This is becoming the case. Current systems require a fast and reliable transmission of sensor data. By not fusing in this way, it is unable to respond quickly to threats. In addition, against the attack speeds and sudden changes of direction of KIDAs The lack of a dynamically adaptable avoidance algorithm is also a current issue. 5 This is a significant limitation in the systems. Current methods are subject to environmental changes. Threats posed by KIDs because they cannot react in real time. It remains vulnerable to electronic countermeasures (RF jammers). These systems become even more ineffective when they encounter (or jammers) and It cannot provide adequate protection against attacks. 10 Purposes of the Invention The purpose of this invention is to protect against attacking kamikaze unmanned underwater vehicles (UAVs). It is the implementation of a method that enables avoidance. The invention provides an effective defense against aggressive kamikaze unmanned underwater vehicles (UAVs). Important factors that enhance the safety of offshore platforms by providing evasiveness. It offers benefits. The developed method allows for the dynamic and uncertain nature of KIDAs. By performing real-time evasive maneuvers against the target's maneuvers It ensures that platforms are effectively protected from attacks. This allows for fast and efficient operation. This makes it possible to respond instantly to sudden threat changes. The method, 20 It operates completely autonomously, eliminating the need for operator intervention. Thus, defense processes are carried out more quickly and effectively, and human Errors caused by various factors are minimized. Efficient combination of data from environmental and inertial sensors. Thanks to this, the system operates with high accuracy in uncertain and dynamic environments. Sensor 25 Fusion reduces sensor-related errors, making avoidance decisions more reliable. This enables electronic countermeasures such as RF mixers and jammers. They are able to perform evasive maneuvers without being affected by the measures, and thus 4 increasing the operational effectiveness and reliability of the defense system platforms It enables them to successfully carry out their duties. The developed avoidance algorithm operates in real-time in dynamic and uncertain environments. by changing course to accommodate moving obstacles and evolving threat scenarios. It adapts quickly. The use of a reinforcement learning-based algorithm ensures continuous 5 by rapidly adapting to changing environmental conditions, even in large state spaces. It improves the accuracy of evasive maneuvers and increases operational efficiency. These benefits ensure the safety of commercial, civilian, and military platforms at sea. by increasing their protection against aggressive KIDAs and enabling them to continue their missions This ensures that both human safety and operational success are critical factors. It offers an advantage. Detailed Description of the Invention The avoidance method implemented to achieve the purpose of this invention is shown in the attached figures. It has been shown. These figures; 15 Figure 1: The kamikaze unmanned underwater vehicle to be used in the method described in the invention. It is a schematic view of its components. Figure 2: Schematic representation of the avoidance scenario in the method described in the invention. Figure 3: Schematic of the sensor fusion algorithm in the method described in the invention. It is a representation. 20 The parts shown in the figures are individually numbered, and these numbers... The corresponding answers are given below. 101. Autonomous marine vehicle 102. Dynamically generated route 103. Fixed radar 104. Fixed obstacle 105. Kida 106. Kida sensor systems 5 107. Target platform actuators 108. Companion computer 109. Positioning and communication module 110 mm radar 111. Lidar 10 112. Inertial sensor unit 113. Sensor fusion block 114. Fusion algorithm 115. Map and position 116. Actor 15 117. New situation 118th Award 119. Environment 6 120. General environment 122. Critic The method described in the invention, in its most basic form, - mm radar (110), lidar (111) via KIDA sensor systems (106), Using data from environmental sensors such as fixed radar (103) 5 Real-time detection of moving or stationary threat elements, - Sensor data in the sensor fusion block (113) Fusion Algorithm (114) by combining using and local map and position (115) information by creating a basis for the escape algorithm, - Learning-based route planning and tracking model by actors (116) 10 by adapting to environmental conditions, new situation (117) and reward (118) determining the most suitable routes through mechanisms, - Escape from threats and maintain a fixed position in accordance with the obtained route plan. Implementation of maneuvering strategies to get away from obstacles (104), - Optimized by the companion computer (108) unit of the decision mechanisms 15 being done, - To the escape route determined via the target platform actuators (107) According to this, the realization of the maveras and the system are both free from fixed obstacles. (104) and also to safely avoid mobile threats, - The accuracy of system decisions with regard to the environment (119) and the general environment (120) is 20 continuous evaluation and improvement, - Criticism of system performance and the success of decision-making mechanisms (121) measurement and evaluation by It includes the steps. This invention is 25 from aggressive kamikaze unmanned marine vehicles (KIDA (105)) It offers an autonomous defense system developed to ensure evasion. 7 The system encountered the threatening KIDA (105) when the Autonomous Marine Vehicle (101) encountered the KIDA (105). in these situations, it receives from environmental sensors and inertial measurement units (112) a safe route with dynamically generated route (102) based on data This enables it to perform an evasive maneuver. In this process, fixed obstacles (104) and fixed radar (103) coverage areas are areas to be avoided as 5 fixed radar (103) and fixed obstacle (104) elements in an environment where they are defined It offers the ability to maneuver without approaching. The system is designed to protect against situations where KIDA (105) poses a threat. It is an improved method and in this process KIDA (105) on KIDA Sensor systems (106) detect threatening situations, target 10 The platform's mobility is provided by the target platform actuators (107). In managing this process, the platform's decision-making and control mechanisms are operated. The companion computer (108) unit comes into play, performs the necessary calculations and They determine avoidance strategies. The avoidance algorithm uses 15 tools that provide the platform's location and contact information. It is supported by the positioning and communication module (109) and global It processes location and communication data in real time. Threat elements During the detection and avoidance process, environmental data are obtained from mm radar (110), lidar (111) and IMU – Data is collected using the Inertial Sensor Unit (112). Combining this data, This is done via the sensor fusion block (113) and the accuracy of the data is 20 Avoidance decisions are optimized by increasing [the number of avoidance attempts]. Fusion Algorithm (114) (Bernoulli in the process of combining sensor data) Local map and platform position information are obtained using RFS (Real-Functional Framework) Environment Modeling. This information is determined. This information is expressed as a Map and Position (115) element and It is the primary source of information for avoidance maneuvers. The 25 most important components of the avoidance algorithm. One of its components is the actor (116), reinforcement learning-based route planning. It adapts to environmental conditions by working with a tracking model. 8 The new situation (117) is that the platform’s environmental protection is in line with the policies determined by the actor. It is the result of the actions he took against the actor. The route determined by the actor (116) and The maneuvers are updated to reflect the physical obstacles encountered, and the reward (118) mechanism, optimizing the model according to the performance of the actions This provides the environment that represents the physical activities and assessments of this process. 5 (119) integrated with the general environment (120) covering scenario and mission actions. It works. However, the system performance and the success of the decision-making mechanisms are evaluated by the critic. (121) is evaluated and measured with the element. This method enables autonomous action against the dynamic threats of aggressive KIDAs (105). by offering avoidance, obstacle avoidance and safe route planning capabilities 10 It enhances the security of autonomous marine vehicles and protects them from attacks.

Claims

9 REQUESTS 1. The invention is based on attacking kamikaze unmanned marine vehicles (KIDA (105)) It is a method that enables avoidance, - mm radar (110), lidar (111) via KIDA sensor systems (106), Using data from environmental sensors such as fixed radar (103) 5 Real-time detection of moving or stationary threat elements, - Sensor data in sensor fusion block (113) fusion algorithm (114) by combining using and local map and position (115) information by creating a basis for the escape algorithm, - Learning-based route planning and tracking model by actors (116) 10 by adapting to environmental conditions, new situation (117) and reward (118) determining the most suitable routes through mechanisms, - Escape from threats and maintain a fixed position in accordance with the obtained route plan. Implementation of maneuvering strategies to get away from obstacles (104), - Optimized by the companion computer (108) unit of the decision mechanisms 15 being done, - To the escape route determined via the target platform actuators (107) According to this, the realization of the maveras and the system are both free from fixed obstacles. (104) and also to safely avoid mobile threats, - The accuracy of system decisions with regard to the environment (119) and the general environment (120) is 20 continuous evaluation and improvement, - Criticism of system performance and the success of decision-making mechanisms (121) measurement and evaluation by It is characterized by including its steps.

Images