UAV subsystem for enhancing flight efficiency using artificial intelligence-based remote sensing of air movements

Abstract

The invention relates to a UAV subsystem developed for the remote sensing of rising air currents, which are used by birds and attempted to be utilized by unpowered manned aircraft (such as gliders and paragliders). With this invention, the flight time of UAV systems can be significantly increased by utilizing the energy of atmospheric movements.

Description

[0001] UAV SUBSYSTEM FOR ENHANCING FLIGHT EFFICIENCY USING ARTIFICIAL INTELLIGENCE-BASED REMOTE SENSING OF AIR MOVEMENTS

[0002] Technical Field

[0003] The invention relates to a UAV subsystem developed for the remote sensing of rising air currents, which are used by birds and unpowered manned aircraft (such as gliders and paragliders). With this invention, the flight duration of UAV systems can be extended by utilizing the energy of atmospheric movements.

[0004] State of the Art

[0005] The most fundamental approach to extending the flight time of an aircraft is to increase the capacity of the onboard fuel and / or battery. However, this method increases the aircraft's weight and dimensions, resulting in higher costs, logistical challenges, and operational difficulties. As an alternative approach, enhancing the energy efficiency of the aircraft's subsystems has become a goal. In this context, advancements in battery technology have aimed to achieve higher energy density relative to weight. However, despite technological progress, the rate of development in battery technology has slowed recently, and significant breakthroughs have not been achieved.

[0006] As a result, the existing systems have turned to alternative energy acquisition methods for aircraft. The integration of solar panels, which have become sufficiently lightweight, has provided additional electrical energy. Nevertheless, the most preferred method of energy acquisition has been the use of atmospheric movements caused by solar radiation. Simply put, thermal air currents formed by the heating of the Earth's surface by the sun — causing the surrounding air to heat, become less dense, and rise — along with updrafts formed by the interaction of terrain and wind, have been utilized as energy sources. These phenomena, efficiently used by birds in nature, are also used by pilots of manned gliders during extended gliding flights. In the specific context of unmanned aerial vehicle (UAV) systems, various studies have been conducted on autonomous soaring algorithms. In all current studies encountered in the literature, the atmospheric conditions in which the aircraft is located have been evaluated using onboard sensors; the focus has been on the real-time use of air movements observable in the immediate vicinity.

[0007] The applications of UAVs are expanding every day, and they are being used for increasingly complex missions. To successfully complete these missions, UAV systems are being equipped with a wide variety of sensors and payloads, thereby increasing their mission capabilities. However, this integration also increases the weight and energy consumption of the aircraft, reducing flight time. Meanwhile, the execution of complex missions requires longer flight durations. For this reason, aircraft are being scaled up to carry more fuel or batteries, and efforts are being made to optimize the energy needs of the payloads. From an aerodynamic perspective, increasing aircraft efficiency with current technology is quite limited, and the number of studies in this area has significantly decreased.

[0008] Using land cover, topographic features, and real-time meteorological conditions analytically to detect rising air currents in advance or remotely has been considered a highly complex subject. Therefore, previous studies and tests have primarily relied on data obtained from sensors measuring the effects directly on the aircraft; methods have been developed to capture only nearby or directly encountered air movements. In other words, all current studies involve the real-time evaluation of air movements encountered along the UAVs flight path and propose control solutions accordingly.

[0009] When the studies in literature are examined in the context of the current system, one of the earliest examples in literature is the study titled "Guidance and Control of an Autonomous Soaring UAV," published on 10.11.2007. In this study, conducted as part of a NASA project, a fixed-wing glider-type UAV was used, and autopilot software was integrated with control and guidance capabilities within rising air currents. The study tested an algorithm designed for climbing within detected updrafts.

[0010] A more recent publication titled "An Autonomous Soaring for Small Drones Using the Extended Kalman Filter Thermal Updraft Center Prediction" focuses on the analysis of environmental effects and presents a model for predicting the center of rising thermal air currents for small UAVs using the Extended Kalman Filter. The study "Improving Fixed Wing UAV Endurance by Cooperative Autonomous Soaring" developed a system and concept that allows two unmanned gliders to autonomously and cooperatively ascend. This system aims to improve the detection of updraft locations and overall system performance.

[0011] Another publication titled "Autonomous Soaring Flight for Unmanned Aerial Vehicles" examined dynamic mechanisms for climbing flight. Algorithms that automatically sense known wind environments and generate wind maps for planning and control were developed to extend flight time without using any propulsion. Lastly, the publication "Unscented Kalman Filtering for Real-Time Atmospheric Thermal Tracking" presented an open- and closed-loop model based on the Unscented Kalman Filter, capable of accurately estimating the position, strength, and size of atmospheric thermal currents for realistic and real-time flight routes.

[0012] In the known state of the art, there is also a patent application numbered US2023106432A1. US2023106432A1 relates to an unmanned system maneuver controller (USMC). The system includes an inertial navigation system, a communication device, and a processor. US2023106432A1 describes an Al-assisted control system that manages the flight, maneuver, or dive of an unmanned system in real-time by considering the movement of the USMC in 3D space. The processor analyzes both the motion data of the USMC and the flight, maneuver, and dive data from the unmanned system to generate control instructions.

[0013] Literature reviews show that solutions focus on enabling the aircraft to gain altitude by using air movements within known or encountered updrafts. Most of the studies examined are theoretical and verified by simulations, with only a few providing outputs based on actual flight tests. In piloted unpowered aircraft, updrafts can be detected based on predictions informed by pilot experience.

[0014] As seen, the fundamental need of current systems is to develop sustainable, effective, and autonomous solutions to increase UAV flight durations. Although previous studies have developed algorithms for the real-time detection of updrafts and gliding within them to gain altitude, these systems can only evaluate atmospheric data in the immediate vicinity of the aircraft. This limits the UAV’s ability to plan routes proactively and utilize potential energy sources at a distance. For advanced energy acquisition, the remote sensing of updrafts requires multidimensional, high-level systems that integrate environmental data, topographic information, and sensor inputs. The limited predictive capacity of current systems is a major obstacle preventing UAVs from achieving longer and more efficient flights.

[0015] In conclusion, due to the above-mentioned challenges and the inadequacies of current solutions in evaluating updrafts only locally, there is a need for innovative subsystem solutions that can remotely sense atmospheric movements, focus on energy acquisition during unpowered flight, operate with autonomous decision-making mechanisms, and significantly extend the flight time of UAV systems.

[0016] Brief Description and Objectives of the Invention

[0017] The invention relates to a UAV subsystem developed for the remote sensing of rising air currents, which are used by birds and attempted to be utilized by unpowered manned aircraft (such as gliders and paragliders). The proposed invention, inspired by the principles used by birds in nature and the experiences of pilots, enables UAVs to predict rising air currents and use them for flight efficiency, similar to how they are used in piloted gliders. It includes subsystems equipped with special algorithms and sensor components capable of remotely detecting such air currents. Thus, UAVs transform from systems that merely react to environmental conditions into intelligent aerial vehicles capable of operating with predictive energy strategies, carrying out longer and more efficient missions. With this invention, the flight time of UAV systems can be significantly increased by utilizing the energy of atmospheric movements.

[0018] The developed system uses deep artificial neural network models to remotely detect dynamic air currents with high energy density and autonomously recommends modifications to the existing flight path to the UAV’s flight controller. In this way, flight efficiency is significantly enhanced.

[0019] With this invention, rather than perceiving only the current environment, predictions of air movements in not-yet-reached regions around the route are made remotely, enabling reorientation and flight planning. Figures

[0020] Figure 1 : View of an example glider UAV system integrated with the UAV subsystem proposed in the invention.

[0021] Figure 2: Block diagram of the UAV subsystem and training system structure of the invention.

[0022] References

[0023] To better describe the UAV subsystem developed in this invention, the parts and components in the figures are numbered. Below are the corresponding definitions:

[0024] 101: Aircraft processing unit

[0025] 102: Aircraft camera system

[0026] 200: Unmanned aerial vehicle

[0027] 201: UAV flight control unit

[0028] 202: Sensor and actuator

[0029] 301: Memory unit

[0030] 302: Input-output module-B

[0031] 303: Neural network-B

[0032] 304: Supervised learning module-B

[0033] 305: Embedded visual training dataset

[0034] 401: Geographical processing unit

[0035] 501: Input-output module- A

[0036] 502: Neural network- A

[0037] 503: Supervised learning module- A

[0038] 504: Geographical training dataset 505: Geographical data set

[0039] 506: Geographical region data

[0040] Detailed Description of the Invention

[0041] The invention relates to a UAV subsystem developed for the remote sensing of rising air currents, which are used by birds and attempted to be utilized by unpowered manned aircraft (such as gliders and paragliders).

[0042] With this invention, a system has been developed that aims to extend the flight time and range of fixed-wing UAVs and unpowered gliding manned aircraft by using high-energy upward air currents generated by the sun heating the Earth's surface. This system can process multiple data sources through an artificial intelligence-based algorithm. The causes, characteristics, and magnitudes of these air movements can vary depending on the location, surrounding topography, and instantaneous weather conditions. In the developed invention, these air movements are sensed remotely through deep learning methods before being encountered, enabling autonomous route planning for the aircraft that maximizes energy efficiency. As a result, the flight time of UAV systems can be significantly increased by utilizing the energy of atmospheric movements.

[0043] The UAV subsystem developed with this invention is used offline and prepared for a mission by using a geographical processing unit (401) that generates a geographical dataset (505), which includes the atmospheric trend of the geographical structure covering the planned flight region. The geographical processing unit (401) maps the atmospheric trend of the geographical region where the UAV subsystem will be used. This data, created by the geographical processing unit (401), is stored in the memory unit (301) within the aircraft processing unit (101) and is used during flight depending on the current location and route.

[0044] The geographical processing unit (401), which generates the offline geographical dataset (505), includes a neural network moduleA (502) trained by the supervised learning module-A (503) using a geographical training dataset (504), which consists of flight data obtained from unpowered manned aircraft and UAV test flights conducted in the same geographical region, satellite imagery, and topographic data. The geographical training dataset (504) contains flight records, satellite images, and topographic information from the region where unpowered manned aircraft and UAV test flights were conducted. This trained system receives geographical region data (506), including satellite images and topographic information of the region where the UAV mission will be performed, via the input-output moduleA (501), and processes it through a processor into the neural network-A (502) module. As output, the processor generates the geographical dataset (505), containing the atmospheric trend information of the region, which is to be used by the aircraft processing unit (101).

[0045] The aircraft processing unit (101) includes a neural network moduleB (303) trained using the embedded visual training dataset (305), which is created by combining regional flight data (location, altitude, time, speed, orientation, and environmental conditions) and flight imagery gathered from manned or unmanned flights in a specific region. During flight, real-time visual data from the onboard camera system (102), which enables the detection of air movements in the aircraft’s heading direction, is combined with data from the flight control unit (201) and sensors / actuators (202) (including physical flight parameters such as position, speed, orientation, acceleration, temperature, pressure), as well as data retrieved from the memory unit (301). These are collected by the input-output module-B (302) and processed via a processor by the trained neural network B (303) to generate information of the likelihood and direction of rising air currents along the UAV’s route and camera field of view. This core output is shared with the UAV flight control unit (201) via the input-output module-B (302). Whether this air movement prediction data, provided by the aircraft processing unit (101) — which is part of the UAV subsystem — is used in route configuration during the flight is left to the UAV flight control unit. Therefore, the system proposed in this invention can be defined as a guidance system.

[0046] In this way, regardless of the variation in route change limitations across different systems, the invention can be used as a guiding / r ecommending system for both civilian and military UAVs.

[0047] The working method of the UAV subsystem explained in this invention can be summarized as follows: • Training of neural network-A (502) via supervised learning module-A (503) using the geographical training dataset (504),

[0048] • Transmission of the geographical region data (506), including satellite images and topographic information of the flight area, via a processor through input-output module-A (501) to neural network module-A (502),

[0049] • Creation of the geographical dataset (505) by neural network module-A (502) for use by the aircraft processing unit (101),

[0050] • Storing of the created geographical dataset (505) in the memory unit (301) of the aircraft processing unit (101) via a connection,

[0051] • During the flight, real-time data obtained from the onboard camera system (102) and sensors (202) is collected via input-output module-B (302) and combined with the geographical dataset (505) retrieved from memory unit (301),

[0052] • Transmission of the combined data via a processor to the neural network module- B (303),

[0053] • Estimation of the likelihood and direction of rising air currents along the route by neural network-B (303) using a processor,

[0054] • Transmission of the resulting guidance data via a processor to the UAV flight control unit (201).

[0055] A UAV subsystem has been developed that can detect ridge, mountain wave, and thermal air movements using artificial intelligence before the aircraft enters their influence zone. The system operates with artificial intelligence algorithms capable of multidimensional analysis of environmental and atmospheric data and can evaluate data from various sources in an integrated manner. In this scope, elevation maps, real-time sensor data from the UAV, live camera images, satellite imagery, and meteorological maps (such as pressure, humidity, and temperature) have been successfully integrated into the system. In addition, atmospheric indicators such as cloud density and solar position are also considered in the prediction of updraft formation, increasing the system’s accuracy rate. In this way, UAVs can predict potential air currents in advance, develop proactive flight strategies, and optimize flight time through unpowered energy gain.

[0056] The invention is particularly suitable for use by manufacturers and users of fixed-wing, gliding UAVs and unpowered manned aircraft.

Claims

CLAIMS1. An artificial intelligence-based unmanned aerial vehicle subsystem for the remote sensing of air currents, characterized in comprising;• A geographical processing unit (401) that includes a neural network module- A (502) trained by a supervised learning module-A (503) using a geographical training dataset (504), transfers geographical region data (506) to the neural network module-A (502) via an input-output module-A (501), and generates as output a geographical dataset (505) containing atmospheric trend information of the region for use by the aircraft processing unit (101),• An aircraft processing unit (101) that includes a neural network module-B (303) trained by a supervised learning module-B (304) using an embedded visual training dataset (305) created by combining regional flight data and flight images, and during flight, collects images from the onboard camera system (102), sensor and actuator data (202), and atmospheric trend data retrieved from the memory unit (301) via the input-output module-B (302), processes this data through the neural network module-B (303) to generate air movement prediction information, and transmits this information to the UAV flight control unit (201).

2. The unmanned aerial vehicle subsystem according to claim 1, wherein the geographical training dataset (504) further comprises flight records, satellite images, and topographic information obtained from unpowered manned aircraft and flight tests conducted in the same geographical region.

3. The unmanned aerial vehicle subsystem according to claim 1, wherein the camera system (102) provides real-time visual data during flight and enables the detection of air movements in the direction of the aircraft's orientation.

4. The unmanned aerial vehicle subsystem according to claim 1, wherein the regional flight data are position, altitude, speed, orientation, and environmental condition data collected from flights performed by manned or unmanned aircraft in a specific geographical region.

5. The unmanned aerial vehicle subsystem according to claim 1, wherein the sensor and actuator data (202) comprises physical flight parameter data such as the aircraft’s position, time, speed, orientation, acceleration, temperature, and pressure.

6. A method of operation for the unmanned aerial vehicle subsystem according to claim 1, comprising the steps of:• Training neural network-A (502) using the geographical training dataset (504) via supervised learning module-A (503),• Transmitting geographical region data (506), including satellite images and topographic information of the flight area, to neural network module-A (502) via a processor and input-output module-A (501),• Generating a geographical dataset (505) by neural network-A (502) for use by the aircraft processing unit (101),• Storing the created geographical dataset (505) in the memory unit (301) of the aircraft processing unit (101) via a processor,• During flight, collecting real-time data from the onboard camera system (102) and sensors (202) via input-output module-B (302), and combining it with the geographical dataset (505) retrieved from the memory unit (301),• Transmitting the combined data to neural network module-B (303) via a processor,• Estimating the likelihood and direction of rising air currents along the route using neural network-B (303) via a processor,• Transmitting the guidance information produced from the prediction to the UAV flight control unit (201) via a processor.

Images