ARTIFICIAL INTELLIGENCE-BASED AIR TRAFFIC MANAGEMENT SYSTEM

Abstract

The invention relates to an air traffic management system with a control architecture based on multi-factor reinforcement learning algorithms, supported by low-latency artificial intelligence models operating at the edge, which monitors numerous parameters of aircraft in real-time, such as position, speed, direction, mission type, urgency level, battery status, and environmental conditions, and determines the safest, most efficient, and appropriate air corridor for each aircraft.

Description

1 TARIFF ARTIFICIAL INTELLIGENCE-BASED AIR TRAFFIC MANAGEMENT SYSTEM Technical Area 5 The invention is supported by low-latency artificial intelligence models that operate at the edge, enabling multi-faceted... It has a control architecture based on agent-reinforcement learning algorithms, aircraft location, speed, direction, mission type, urgency level, battery status, and It monitors many parameters such as environmental conditions in real time and provides at least 10 for each vehicle. with an air traffic management system that determines a safe, efficient and appropriate air corridor It is related. State of the Art Today, unmanned aerial vehicles (UAVs) are used for civil, commercial, and corporate purposes. This is becoming increasingly common, leading to new management and coordination in the airspace. This increases the need for new approaches. Existing air traffic management systems are essentially manned aircraft operate on fixed routes and around specific airports. It is designed to regulate flights, autonomous and dense unmanned aerial traffic 20 It does not have the ability to manage in real time. Currently used... In these systems, centralized control structures, predefined static routes, and fixed... Altitude planning and manual operator interventions are relied upon; this leads to autonomous air transport. This contradicts the dynamic and variable nature of their tools. With current technology, the use of unmanned aerial vehicles is rapidly increasing, especially in urban areas. flights are increasing, and consequently, both civilian and corporate flights are using the airspace. The coordination of operations is becoming increasingly complex. Emergency cargo transportation, autonomous air taxis, security and surveillance missions, disaster areas Applications such as search and rescue operations and urban drone deliveries, aerial 30 This necessitates the multi-dimensional and dynamic management of the field. Current air traffic control systems rely on fixed routes and manual coordination processes. Because it relies on it, autonomous systems can operate in real time in dense and variable air traffic. This prevents the efficient management of the airspace. Therefore, the airspace is very 35 2 treated as a layered environment, autonomous systems avoid collisions with each other. It can move, create dynamic routes according to task priorities, and change a new generation traffic management system that can adapt to weather conditions and environmental impacts Infrastructure is needed. In current technological applications, the routes of autonomous aerial vehicles are determined either before flight or... Manually determined or basic obstacle detection systems for a limited number of vehicles. It is guided through these systems. These systems are mostly GPS-based navigation, simple It is based on route planning and basic collision avoidance algorithms. Advanced artificial intelligence. applications, especially reinforcement learning and multi-factor 10 At the multi-agent systems level, there are currently limited laboratory applications. It has not been able to go beyond that. Furthermore, in current systems, the interaction between autonomous vehicles... Route replanning based on traffic density, airspace management based on mission priority. Issues such as usage are not adequately addressed. However, the simultaneous operation of numerous drones in urban airspace... The risk of collisions that may arise in this situation, frequency and bandwidth conflicts, route Existing measures are in place to address scenarios such as bottlenecks and delays in critical tasks. Technologies are unable to offer an effective solution. Existing algorithms are mostly based on individual tools. It operates from this perspective, meaning each aircraft operates solely based on its own environmental data. This situation leads to poor coordination between vehicles and multiple vehicles moving around. This causes system integrity to be compromised in drone missions. Furthermore, in the air... sudden changes in conditions, unexpected obstacles, or unplanned events flexibility for dynamic scenarios, such as integrating tasks into the system. It cannot be displayed. 25 Another significant shortcoming is that existing systems are largely dependent on centralized architectures. The reason is that central control systems handle high volumes of data under heavy air traffic. It gets bogged down under flow and decision load; this is especially true for low latency and high In tasks requiring rapid reaction time (e.g., emergency drones, air ambulance 30 (systems) cause the system to become inadequate. In addition, central The vulnerability of structures to individual failures poses a risk to system integrity. It constitutes. 3 The fundamental technical problem that arises as a result of all these limitations is autonomous air transport. their vehicles in real-time, in intense, variable and mission-prioritized scenarios, It is the inability to navigate through the airspace safely, without interference, and flexibly. Especially when a large number of autonomous vehicles need to use the airspace simultaneously. In these situations, task overlaps, the risk of collision increases, and critical services are delayed. 5 And the system's insufficient adaptive capacity causes serious operational disruptions. This technical problem is not only operational but also security-related. strategic dimensions such as energy efficiency and airspace management in public areas It is also effective. Application number US10332405B2 concerns low-altitude air traffic control for unmanned aerial vehicles. management, collision avoidance, route planning, air corridor management and communication It relates to a system encompassing its infrastructure. However, the application mentions distributed artificial intelligence, the task. Prioritization and edge computing-supported dynamic air corridor optimization are not included. In conclusion, due to the negative aspects described above and the current solutions, the subject matter... Due to its shortcomings, an improvement is needed in the relevant technical field. It has been observed. Purpose of the Invention 20 The main purpose of the invention is to enable unmanned aerial vehicles and autonomous drone systems to be used in urban and rural areas. to be able to operate safely, regularly and without interference in rural airspace to provide. The system developed as part of the invention is low-latency artificial intelligence that operates at the edge. supported by models, based on multi-factor reinforcement learning algorithms It offers a control architecture. The system monitors the position, speed, direction, and mission type of aircraft. It instantly monitors many parameters such as urgency level, battery status, and environmental conditions. We monitor and determine the safest, most efficient and most suitable air corridor for each vehicle. This process determines the direction of the device. This process allows edge devices to operate without the need for a central control unit. and this is accomplished through distributed artificial intelligence agents. Thus, the system, It offers a scalable, latency-free, and localized solution. 4 The system described in the invention enables the safe performance of time-sensitive tasks such as emergency medical transport. while enabling this to be carried out in this way, autonomous air taxi systems provide vertical urban mobility. It enables transportation to provide services in a regular and controlled manner. Furthermore, effectively performing critical tasks such as search and rescue and aid delivery in disaster situations Planning and implementation of public safety and surveillance activities, optimization 5 and urban cargo and logistics transportation via specific air routes It also provides an important infrastructure in terms of implementation. The system described in this invention is also expected to become widespread in urban air conditioning in the future. directly related to the technical requirements of mobility and advanced air mobility concepts. civil aviation authorities, city administrations, and transportation planners are responding. It is designed for use by both public and private drone fleet operators. This is developed in a way that will provide high added value for both institutions and the private sector. The structure is holistic, supporting the safe and efficient use of autonomous aerial vehicles. It offers a solution. 15 This invention enables multiple autonomous aerial vehicles to operate simultaneously in dense and dynamic airspace. enabling it to fly simultaneously, without conflicts and synchronized according to mission priorities, Edge-based, multi-agent reinforcement learning, based on a distributed AI architecture. It offers an innovative air traffic management system supported by algorithms. 20 The invention solves the problems of low flexibility and high latency of classical central control systems. eliminating the inadequacy of existing individual drone navigation systems. Highly sensitive and adaptive solution proposals in the multi-interactive scenarios it faces. It is developing. The developed system is based on autonomous systems built on edge computing infrastructure. It creates decision-making agents. Each agent uses the local data of the relevant aircraft. while carrying out its work, it also takes into account information from other agents in its vicinity. It defines the operational strategy. This structure is decentralized, distributed, and cooperative. It creates a learning process and directly increases the scalability of the system. Decision 30 the acquisition process, multi-agent reinforcement learning MARL) algorithms are used to support each agent's own policy function. It is learning. The main goal here is for each aircraft to successfully perform its own mission. while ensuring that it does not interfere with other vehicles sharing the same airspace, It is about completing its route without colliding and without violating priority rules. 35 The system's decision-making model is a defined Markov Decision Process (Markov) for each agent. It is based on the Decision Process (MDP). Each agent 𝑖 is as follows: It has a defined decision model: ℳ = (𝑆, 𝐴, 𝑃, 𝑅, 𝛾) Here:  𝑆 is the state space representing the state of agent 𝑖 (for example: (position, speed, direction, battery status, position of surrounding vehicles).  A represents the set of actions that can be taken in the current situation (e.g., direction 10). (change, decrease speed, increase).  𝑃 is the transition probability function, and the expression 𝑃 (𝑠 , 𝑠, 𝑎) represents the transition from state 𝑠 of agent 𝑖. It indicates the possibility of transitioning from the action 𝑎 to the state 𝑠.  The reward that agent 𝑖 receives when performing action 𝑎 in state 𝑠 is R (s, 𝑎). Its value is 15.  𝛾 ∈ [0,1] discount that allows future rewards to be reduced to their present value It is the coefficient. Each agent optimizes its policy function, 𝜋 (𝑎 ∣ 𝑠), to maximize its total reward. It attempts to maximize. The goal is to maximize the total reward function as follows: The optimal policy is to find 𝜋∗: 20 𝜋∗ = arg max 𝔼[ 𝛾 𝑅 (𝑠 , 𝑎 )] The reward function used in the system is multi-dimensional and includes not only arrival success but also... safe distance in time, task priority, energy efficiency and collision avoidance, etc. It also takes into account the elements. The reward function is generally outlined as follows: 25 definable: 𝑅 (𝑠, 𝑎) = 𝛼 ⋅ 𝑅task + 𝛼 ⋅ 𝑅distance + 𝛼 ⋅ 𝑅c̒collision_prevention + 𝛼 ⋅ 𝑅energy Here, each 𝛼 coefficient is determined by the system administrator and serves different purposes. They can be weighted according to their type. For example, 30 for an emergency medical transport mission. Higher emphasis is placed on priority route selection, which is more important than collision-free movement. can be appointed. However, the system performs collision prediction and air corridor management in 3D space. This is done in real-time with the coordinates (x, y, z) of each vehicle. 6 They are being monitored. Agents are aware of each other's positions through edge devices. By sharing data at low latency, it can predict potential future collision points. They are doing this and re-optimizing their routes based on this information. Estimated The probability of collision is 𝑃coll s on, the current velocity vectors of the vehicles are 𝑣 and 𝑣, their positions are 𝑝 and 𝑝⃗, and The safe distance limit is calculated as follows, taking into account the distance d: 5 𝑃collision = { 1, if ∥ 𝑝 − 𝑝⃗ ∥< 𝑑safe in predicted t 0, otherwise This information is reflected as a negative value in the reward function by each agent, and It encourages the system to generate a safe route. The system's learning process is ongoing online. For this purpose, each agent over time, it develops better policy functions based on the air traffic dynamics in its surrounding environment. It can be improved, and as a result, the system becomes safer under similar environmental conditions. It is becoming capable of generating efficient decisions. Among the algorithms used is Proximal Modern reinforcement learning methods such as Policy Optimization (PPO) and Soft Actor-Critic (SAC) 15 These techniques include algorithms with continuous action fields and high dimensions. They are preferred because they offer high sample efficiency for state spaces. In addition, the system connects to the central server thanks to artificial intelligence agents running at the edge. It has the capacity to make decisions in a distributed manner without needing to be externally controlled. This structure is only 20 It not only reduces latency, but also improves the scalability and security of the system. This can also increase the impact of network connection interruptions or overloading of central servers. loads allow tasks to continue without affecting the overall performance of the system. provides. In conclusion, this invention addresses the shortcomings of existing air traffic management systems. It offers an innovative solution in these areas; distributed, learning, real-time and It creates an overlap-free autonomous airspace management infrastructure. The system only not only in terms of safety and mission performance; but also energy efficiency, mission It is also important in terms of aspects such as prioritization, dynamic adaptation and system scalability. 30 It provides advantages. The structural and characteristic features and all the advantages of the invention are given in the figures below. This becomes clearer thanks to the detailed explanation written with references to these figures. It will be understood. 35 7 Explanation of the Figures Figure 1 is a representative view of the system that is the subject of the invention. The drawings do not necessarily need to be scaled and are useful for understanding the invention. Unnecessary details may have been omitted. Explanation of Part References 1. Autonomous aerial vehicle 2. Air corridor network 3. Artificial intelligence agent 4. Collision prediction module 5. Task prioritization module 15 6. Edge computing points 7. Communication layer 8. Common task management interface Detailed Description of the Invention 20 In this detailed explanation, the preferred configurations of the invention are not merely for better understanding the subject. in order to facilitate understanding and without imposing any limiting effects It is explained. The invention is an air traffic management system that includes GPS, lidar, radar, and camera. autonomous vehicles that detect their surroundings using sensors and collect position, speed, and orientation parameters. The aircraft (1) receives and processes the data obtained, and the autonomous aircraft (1) is available AI agent (3) that determines the status of the collected environmental and performance data autonomous air transport 30 by dividing the airspace into three-dimensional layers as reflected on it. The flight route of the vehicle (1) is a corridor according to traffic density and safety requirements. air corridor network (2) that enables placement in the channel, autonomous aerial vehicles (1) and by analyzing position, velocity and orientation data from artificial intelligence agents (3) AI predicts collisions at the right time, generates route suggestions based on the predictions. The collision prediction module (4), which transmits the mission type of each autonomous aerial vehicle (1) to its agent (3), 35 8 Task prioritization module that assesses urgency level and operational priority. (5) by performing low latency data communication between autonomous aerial vehicles (1) It enables the sharing of information such as location, speed, energy level, route, and collision risk. It includes a communication layer (7). An autonomous aircraft (1) is an aircraft that performs autonomous flight. GPS, camera, lidar, Environmental sensing, location determination, and route tracking with radar sensors and built-in Edge AI. The air corridor network (2) provides safe distance and conflict-free flight. It is the structure that creates dynamic 3D virtual airways. Autonomous aerial vehicles (1) are defined It enables safe movement within the operational area. 10 Artificial intelligence agents (3) are autonomous decision units that operate on autonomous aerial vehicles (1). It provides coordination between autonomous aerial vehicles (1) and tasks through reinforcement learning. It provides decision-making support in these processes. Collision prediction module (4) estimates the risk of collision based on position and speed data. It is the system component that calculates. It identifies risky situations and adds new ones when necessary. It generates route suggestions. Task prioritization module (5) autonomous air 20 according to task type and urgency level It is the module that gives priority to the vehicles (1). Emergency task autonomous aerial vehicles (1) This allows them to obtain priority access and expedited passage within the corridor. Edge computing points (6) are computing units that provide local processing power. Collision prediction, Real-time calculations for route optimization and traffic analysis are at a low level (25). It happens with a delay. Communication layer (7), low latency data between autonomous aerial vehicles (1) It is a communication structure that enables the sharing of location, route, and hazard information in real time. It supports the coordinated movement of artificial intelligence agents by facilitating the transfer of information. 30 The common task management interface (8) allows the operator to monitor air traffic and assign tasks. It is the interface that enables the system to be implemented and intervened in when necessary. data from the task prioritization module (5) and the collision prediction module (4) Shows it to the operator. 35 9 The invention enables safe, autonomous operation of unmanned aerial vehicles in the airspace. to enable it to operate in a non-conflicting, prioritized and synchronized manner. It is an integrated artificial intelligence-based air traffic management system developed for this purpose. The basic working principle of the system is the environmental data obtained by autonomous aerial vehicles (1). processing of situational data, interpretation of this data by artificial intelligence agents (3), Planning flight paths safely on the air corridor network (2) and these It relies on updating the plan in real time. At the start of the operation, GPS, lidar, and other equipment were found on the autonomous aerial vehicles (1). Radar and camera sensors detect the environment, determining critical flight parameters such as position, speed, and orientation. collects the parameters. The data obtained are placed on autonomous aerial vehicles (1). The current state of the vehicle is processed by artificial intelligence agents (3) working as is determined. At this stage, artificial intelligence agents (3) identify both the objects around them and the other 15 It makes instant decisions by evaluating data received from autonomous aerial vehicles (1). The collected environmental and performance data represent an air corridor network for the flight area. (2) is projected onto the air corridor network (2), a virtual grid model of the airspace. It enables the division into three-dimensional layers. Thus, each autonomous aerial vehicle (1) 20 The flight route is selected according to a suitable corridor channel based on traffic density and safety requirements. This structure plays a fundamental role in ensuring the system's non-conflict flight capability. It fulfills. The prevention of collisions in the system is carried out by the collision prediction module (4). 25 Collision prediction module (4), from autonomous aerial vehicles (1) and artificial intelligence agents (3) Future collision prediction by analyzing incoming position, velocity and orientation data It does. Safe between the future possible locations of the two autonomous aerial vehicles (1). When it is predicted that the distance will be breached, the collision prediction module (4) will activate the relevant autonomous air It determines that the route of the vehicle (1) needs to be updated. This information is given to the AI ​​agents 30 (3) is communicated to create a new route. Task prioritization is done by the task prioritization module (5). The task prioritization module (5) is implemented. Each autonomous aircraft (1) has a task. It assesses the type, urgency level, and operational priority. Airspace congestion. In the event of traffic congestion occurring or forming in certain areas, task 35 prioritization module (5) provides a more suitable autonomous aerial vehicle (1) with a critical mission. This allows them to be directed to the corridor channel or to gain priority right of way. These processes are executed in real time by endpoints (6) is provided. End information points (6) are located in areas with high air traffic. local processing units are located and the collision prediction module (4) with artificial intelligence by performing the calculations needed by the agents (3) locally It reduces latency. Thus, the system operates quickly and efficiently without relying on a centralized processing infrastructure. It operates in a scalable manner. Coordination between autonomous aerial vehicles (1) is via the communication layer (7) Communication layer (7) is provided between autonomous aerial vehicles (1) with low latency. By communicating data, information such as location, speed, energy level, route, and collision risk can be obtained. This allows the data to be shared. The shared data is processed by artificial intelligence agents (3). By evaluating the situation, all aircraft are enabled to act in a coordinated manner. 15 System monitoring and management via common task management interface (8) is performed. The common task management interface (8) provides instantaneous information for all autonomous aerial vehicles (1). their status, types of missions, established air corridors, anticipated collisions It presents the risks and task prioritization results to the operator. If necessary, the operator can 20 Route change, task cancellation manually via the common task management interface (8) or can make task assignments. When all elements work together, the system is secure, conflict-free, and task-oriented. and creates an optimized air traffic flow in real time. Each element has its own 25 while performing its function, continuous data exchange takes place between the elements and Thanks to AI-based coordination, the system functions as a holistic and autonomous structure. It works. The working principle of the invention is 30 Autonomous aerial vehicle (1), air corridor network (2) and artificial intelligence agent (3) are activated The initial scan of the operation area is carried out and the initial corridor structure is established. Autonomous The aircraft (1) transmits position and speed data to the collision prediction module (4). 35 11 The system calculates the collision probabilities within the air corridor network (2) and identifies the risky areas. detects. The artificial intelligence agent (3) cooperates with the task prioritization module (5) Emergency mission autonomous aerial vehicles (1) priority route and accelerated within the corridor It grants the right of way. Edge information points (6), collision prediction module (4) and necessary for route optimization It performs real-time calculations with low latency. Communication layer (7), by simultaneously transmitting location, route and hazard information between autonomous aerial vehicles (1) It enables the artificial intelligence agent (3) to act in a coordinated manner. Common task management interface (8), task prioritization module (5) and conflict prediction It displays the data from module (4) to the operator. If necessary, the operator can manually route or can change tasks. System-wide air corridor network (2), artificial intelligence agent (3) and collision prediction module (4) operates with a continuous feedback loop to adapt to changing traffic. It automatically re-optimizes route and corridor structures accordingly. 15 Autonomous aerial vehicles (1), air corridor network (2), artificial which form the main structure of the invention intelligence agents (3), collision prediction module (4), task prioritization module (5), end information points (6), communication layer (7) and common task management interface The interactions between (8) are explained below from a mathematical and technical point of view. The system's basic decision-making mechanism is the artificial intelligence agents of each autonomous aircraft (1) 20 (3) is controlled by distributed reinforcement learning (Reinforcement Learning- It is based on the RL infrastructure. A separate agent is defined for each autonomous aerial vehicle (1), Agents' state, actions, and rewards, Air Corridor Network (2) is modeled to be defined on. This structure is a Markov Decision Process Within the framework of (MDP), it is expressed as follows: 25  S = state space  A = action space  P = transition probabilities  R = reward function  γ = discount coefficient 30 For each agent, the MDP is defined as follows: ℳ = (𝑆, 𝐴, 𝑃, 𝑅, 𝛾) The drone's position in 3D space is represented as follows: 12 𝑝 (𝑡) = (𝑥 (𝑡), 𝑦 (𝑡), 𝑧 (𝑡)) Velocity vector: 𝑣 (𝑡) = (?̇? (𝑡), ?̇? (𝑡), ?̇? (𝑡)) The agent's action consists of parameters that change the direction and magnitude of its velocity: 𝑎 (𝑡) = (Δ𝜃, Δ𝜙, Δ𝑣) Here, θ and φ represent direction angles. The implementation of each action results in autonomous air 10 determines the new location of the vehicle (1): 𝑝 (𝑡 + 1) = 𝑝 (𝑡) + 𝑣 (𝑡) ⋅ Δ𝑡 Since all autonomous aerial vehicles (1) use the same airspace, the difference between agents is 15 Interaction is critically important. Therefore, the reward function is only relevant for the autonomous aerial vehicle. (1) not its own success, safe distance, energy with other autonomous aerial vehicles (1) multidimensional, including factors such as consumption, mission priority, and collision risk. It has been defined as: 𝑅 = 𝛼 𝑅 ̈ + 𝛼 𝑅 + 𝛼 𝑅\ \ + 𝛼 𝑅 The reward based on collision risk is defined by the following function: 𝑅\ \ = −𝛽 ⋅ 𝑓(∥ 𝑝 − 𝑝 ∥) When the distance between autonomous aerial vehicles (1) falls below the safe limit, it becomes negative. It is a function that produces punishment: f(d) = { 1 d d < d 0 d ≥ d This expression is continuously calculated by the collision prediction module (4) and artificial 30 The intelligence agents (3) are directly reflected in the policy functions. Task The priority is integrated into the model by the task prioritization module (5). Each drone A task priority coefficient is assigned to it. Ω ∈ [0, 1] 35 The reward for the mission is defined as follows. 13 𝑅 ̈ = 𝜆 ⋅ Ω ⋅ 𝑔 𝑑 Here, the function g becomes positive as the distance of the autonomous aerial vehicle (1) to the target decreases. It contributes. The air corridor network (2) is created using a three-dimensional grid structure. The airspace is divided into N×M×K cells. 5 𝐶 = 𝑐 , ∣, 0 < 𝑥 < 𝑁 0 < 𝑦 < 𝑀 0 < 𝑧 < 𝐾 Each cell has a transit cost. This cost depends on air traffic density, wind, Obstacles and collision risks are taken into account when calculating as follows. 𝑐𝑜𝑠𝑡 𝑐 = 𝜂 𝜌 + 𝜂 𝜔 + 𝜂 𝑃 This cost function applies to both the route planning algorithm and the artificial intelligence agents. (3) affects the reward structure. The system also operates on an edge computing basis. Edge computing points (6), each autonomous By reducing the data processing load of the aircraft (1), the following operations can be performed simultaneously 15 It accomplishes.  Local collision prediction.  Air corridor update.  Data synchronization between autonomous aerial vehicles (1) (Communication layer (7)) These calculations eliminate dependence on a central server, reducing system latency to 20. It minimizes it. The communication infrastructure of the system is the communication layer. (7) is carried out. In the communication model, autonomous aerial vehicles (1) The relationships between them are treated as a time-variable graphical structure. 𝐺(𝑡) = 𝑉, 𝐸(𝑡) 25 Here, V represents aircraft and E(t) represents communication links. The following condition must be met for the connection to be active. ∥ 𝑝 (𝑡) − 𝑝 (𝑡) ∥< 𝑟 14 This graphical structure is used in the joint policy learning of artificial intelligence agents (3). Finally, the common task management interface (8) collects data from all modules presenting it to the operator and integrating manual interventions into the system when necessary. The common task management interface (8) provides the overall status of the system, risk their regions, mission priority levels and real-time air corridor configurations 5 It visualizes.

Claims

REQUESTS 1. It is an air traffic management system whose features include:  It monitors its surroundings with its GPS, lidar, radar and camera sensors. Autonomous aerial systems collect position, speed, and orientation parameters by sensing and monitoring. vehicle (1),  by receiving and processing the obtained data, the current status of the autonomous aerial vehicle (1) Determining artificial intelligence agent (3),  air on which collected environmental and performance data are reflected by dividing the field into three-dimensional layers, the flight of the autonomous aerial vehicle (1) 10 its route to a corridor channel according to traffic density and safety requirements. air corridor network that enables placement (2),  Location from autonomous aerial vehicles (1) and artificial intelligence agents (3), It makes future collision predictions by analyzing speed and orientation data. Collision 15 which generates route suggestions based on the prediction result and transmits them to the AI ​​agent (3) prediction module (4),  each autonomous aerial vehicle (1) mission type, urgency level and operation Task prioritization module (5) which evaluates its priority.  Low latency data communication between autonomous aerial vehicles (1) by performing this operation, information on location, speed, energy level, route, and collision risk is collected. It includes a communication layer (7) that enables sharing.

2. Air traffic management system in accordance with Claim 1, and its feature is; autonomous aircraft (1) current status, types of tasks, established air corridors, predicted Joint task 25 presents the operator with the risks of conflicts and the results of task prioritization. It includes the management interface (8).

3. An air traffic management system compliant with Claim 1, characterized by its suitability for areas with high air traffic density. AI agents (3) with collision prediction module (4) located in regions 30 real route optimization and traffic analysis needed by Edge computing points that reduce processing latency by performing timely calculations. (6) is included. 16 4. Air traffic management system compliant with Claim 1, characterized by its reinforcement learning capabilities. AI agent (3) that provides coordination between autonomous aerial vehicles (1) It includes.

Images