Intelligent outdoor flying waistcoat based on deep learning

Abstract

The invention relates to the technical field of deep learning, aircrafts and the like, and provides an intelligent outdoor flight waistcoat based on deep learning, the intelligent outdoor flight waistcoat is provided with a waistcoat body and an intelligent flight control system, the intelligent flight control system comprises a deep learning control module, a main flight power device, an attitude auxiliary adjusting device and an environment sensing system; the environment sensing system comprises a visual sensor, a laser radar sensor, an ultrasonic sensor, a height sensor and a GPS positioning module and is used for collecting outdoor environment data in real time and transmitting the outdoor environment data to the deep learning control module, and the deep learning control module analyzes the environment data in real time and automatically optimizes the flight state through an autonomous learning algorithm. The main flight power device is controlled to provide lift force and propulsion power, and the attitude auxiliary adjusting device is utilized to automatically correct the flight attitude, so that the flight waistcoat is convenient to wear, intelligent in environmental perception, high in self-adaptability, safe and stable, and meets diversified outdoor flight and rescue task requirements in the future.

Description

Technical Field

[0001] The present invention relates to the technical fields of deep learning, aircraft, etc., and particularly relates to an intelligent outdoor flight vest based on deep learning. Background Art

[0002] As an emerging means of transportation, personal aircraft are gradually becoming an important development direction for future means of transportation because they can get rid of ground traffic congestion and quickly shuttle through complex urban environments and special outdoor environments. However, the existing personal aircraft generally have the following technical defects: First, the existing flight devices are large in volume, with a complex overall structure, difficult to wear, and not easy for ordinary users to quickly wear and use; Second, the existing aircraft usually rely on manual experience for control, with difficult attitude control during flight, insufficient safety and stability, and are prone to safety accidents especially when encountering complex terrains or harsh climate environments; Third, the existing aircraft lack intelligent environmental perception and adaptive capabilities, cannot achieve automated route planning and real-time obstacle avoidance during flight, have a slow response speed in the face of unknown environments or emergencies, and have obvious safety hazards; In addition, the existing aircraft lack the ability to effectively interact with the external environment in real time and cooperate in flight, and cannot meet the requirements of complex outdoor tasks, such as special application scenarios like high-rise building escape, disaster search and rescue, or emergency material transportation. Therefore, in view of the above problems, it is urgent to develop an intelligent outdoor flight vest with convenient wearing, intelligent environmental perception, strong self-adaptability, and safety and stability to meet the needs of future diversified outdoor flight and rescue tasks. Summary of the Invention

[0003] Aiming at the deficiencies of the above-mentioned existing technologies, the present invention provides an intelligent outdoor flight vest based on deep learning, so that the flight vest can achieve convenient wearing, intelligent environmental perception, strong self-adaptability, and safety and stability, and meet the needs of future diversified outdoor flight and rescue tasks.

[0004] The intelligent outdoor flight vest based on deep learning provided by the present invention includes: A vest body for a user to wear, and an intelligent flight control system is arranged on the vest body. The intelligent flight control system includes a deep learning control module, a main flight power device, an attitude auxiliary adjustment device, and an environmental perception system; The deep learning control module is connected to the main flight power device, the attitude auxiliary adjustment device, and the environment perception system through a data communication line; the environment perception system includes a vision sensor, a lidar sensor, an ultrasonic sensor, an altitude sensor, and a GPS positioning module; the vision sensor, the lidar sensor, the ultrasonic sensor, the altitude sensor, and the GPS positioning module are all used to collect outdoor environment data in real time and transmit it to the deep learning control module; the deep learning control module analyzes the environment data in real time, automatically optimizes the flight state through an autonomous learning algorithm, controls the main flight power device to provide lift and propulsion power, and automatically corrects the flight attitude by using the attitude auxiliary adjustment device.

[0005] Compared with the prior art, the beneficial effects of the present invention are as follows: The intelligent outdoor flight vest based on deep learning provided by the present invention is provided with a vest body for the user to wear. An intelligent flight control system is arranged on the vest body. The intelligent flight control system includes a deep learning control module, a main flight power device, an attitude auxiliary adjustment device, and an environment perception system. The deep learning control module is connected to the main flight power device, the attitude auxiliary adjustment device, and the environment perception system through a data communication line. The environment perception system includes a vision sensor, a lidar sensor, an ultrasonic sensor, an altitude sensor, and a GPS positioning module. The vision sensor, the lidar sensor, the ultrasonic sensor, the altitude sensor, and the GPS positioning module are all used to collect outdoor environment data in real time and transmit it to the deep learning control module. The deep learning control module analyzes the environment data in real time, automatically optimizes the flight state through an autonomous learning algorithm, controls the main flight power device to provide lift and propulsion power, and automatically corrects the flight attitude by using the attitude auxiliary adjustment device, so that the flight vest is convenient to wear, has intelligent environment perception, strong autonomous adaptability, and is safe and stable, meeting the requirements of future diversified outdoor flight and rescue tasks. Brief Description of the Drawings

[0006] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. Some specific embodiments of the present invention will be described in detail later with reference to the drawings in an exemplary rather than restrictive manner. The same reference numerals in the drawings denote the same or similar components or parts. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings: Figure 1 is a schematic structural diagram of an intelligent outdoor flight vest based on deep learning in an embodiment of the present invention. Detailed Embodiments

[0007] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0008] See Figure 1 , this embodiment provides an intelligent outdoor flying vest based on deep learning, including: A vest body for the user to wear, and an intelligent flight control system is arranged on the vest body. The intelligent flight control system includes a deep learning control module, a main flight power device, an attitude auxiliary adjustment device, and an environment perception system; The deep learning control module is connected to the main flight power device, the attitude auxiliary adjustment device, and the environment perception system through a data communication line; the environment perception system includes a vision sensor, a lidar sensor, an ultrasonic sensor, an altitude sensor, and a GPS positioning module; the vision sensor, the lidar sensor, the ultrasonic sensor, the altitude sensor, and the GPS positioning module are all used to collect outdoor environment data in real time and transmit it to the deep learning control module; the deep learning control module analyzes the environment data in real time, automatically optimizes the flight state through an autonomous learning algorithm, controls the main flight power device to provide lift and propulsion power, and uses the attitude auxiliary adjustment device to automatically correct the flight attitude.

[0009] It should be noted that existing personal aircraft are often large in size and complex in structure, making it difficult to achieve rapid wearing similar to clothing, which poses a significant obstacle to the use of ordinary users. In this embodiment, various flight components are integrated on the vest body, adopting a compact design. The main flight power device, attitude auxiliary adjustment device, deep learning control module, and environmental perception system are all placed in the wearable vest, significantly reducing the overall volume and weight. Users only need to simply wear and fix it like an ordinary vest to complete the configuration of the flight equipment. Additionally, traditional personal aircraft usually rely on the pilot's experience for manual control and lack a systematic self-learning ability, resulting in difficulty in maintaining a stable attitude in complex terrains or adverse weather conditions and prone to safety risks. In this embodiment, by setting up a deep learning control module, on the one hand, the deep learning control module receives multi-source information (such as data from vision, lidar, ultrasonic, altitude, GPS positioning, etc.) collected in real time by the environmental perception system, and on the other hand, uses self-learning algorithms to extract features, predict states, and adaptively optimize flight strategies for the current environment. Through the self-learning, self-adaptive, and rapid decision-making capabilities of the deep learning algorithm, the flight vest can automatically adjust the flight attitude according to complex environmental changes and avoid potential risks, thus greatly enhancing the safety and stability of the flight process. Additionally, in this embodiment, various sensors such as a vision sensor, a lidar sensor, an ultrasonic sensor, an altitude sensor, and a GPS positioning module are integrated on the vest body, which can form an environmental perception system for multi-source data fusion. Among them, the vision sensor can capture environmental images for identifying obstacles or key terrain features; the lidar sensor can accurately measure distances and construct a three-dimensional model of the surrounding environment; the ultrasonic sensor plays an important role in detecting close-range obstacles and navigating through narrow spaces; the altitude sensor ensures that the flight vest maintains an appropriate altitude under various terrain drops and monitors the ascent / descent speed; the GPS positioning module provides accurate coordinate information for the flight vest to meet the needs of long-distance flight or autonomous cruising. The data fusion between multiple types of sensors enables the deep learning control module to quickly model, analyze, and assess risks for the current flight environment, achieving immediate response to complex terrains and emergencies. The main flight power device of the flight vest is responsible for providing the main lift and propulsion power, which can help the flight vest fly autonomously at different heights and in different environments; the attitude auxiliary adjustment device eliminates unstable factors caused by wind, airflow, or attitude changes by making real-time fine adjustments to the thrust direction or wing surface structure. In this embodiment, the deep learning control module combines the feedback data of various sensors, analyzes the current attitude of the flight vest, and automatically sends adjustment instructions to the attitude auxiliary adjustment device, thereby quickly correcting and compensating for the flight angle, azimuth, and posture. This process does not require a large amount of manual intervention by the user, can significantly reduce the difficulty of flight operation, and effectively improve the flight stability and safety factor.In addition, in this embodiment, the vest-like shape combined with the flexible and maneuverable characteristics brought by intelligent deep learning control can adapt to narrow flight corridors as well as complex terrains such as the wild and disaster areas. Through the real-time fusion of sensor data and the autonomous planning of the target flight path, the vest can quickly complete operations such as takeoff, landing, hovering, obstacle avoidance, and fixed-point search in case of emergencies, thus better meeting the requirements of rescue and emergency applications.

[0010] In some preferred embodiments, the main flight power device includes a multi-rotor electric engine group and a power battery group installed on the back of the vest body; the multi-rotor electric engine group includes at least four electrically-driven rotors symmetrically distributed horizontally; the power battery group is located at the rear side of the vest body and is electrically connected to the multi-rotor electric engine group to continuously supply electrical energy to the multi-rotor electric engine group. It should be noted that in this embodiment, at least four electrically-driven rotors are installed on the back of the vest body and are distributed symmetrically horizontally to provide sufficient lift and good controllability. The multi-rotor system has a relatively simple mechanical structure, a high redundancy, and a relatively low maintenance cost, and the electrically-driven rotors are superior to fuel engines in terms of noise control and environmental adaptability. The power battery group is located at the rear side of the vest body and directly supplies continuous electrical energy to the multi-rotor electric engine group, facilitating the overall layout and centralized management, effectively improving the flight stability and endurance, and thus solving the defects of the traditional personal aircraft power system such as large volume, complex oil circuit, and difficult maintenance.

[0011] In some preferred embodiments, the attitude auxiliary adjustment device includes vector thrust engines provided on both shoulders of the vest body and an attitude stabilizing fin group provided at the waist position of the vest; the vector thrust engines adjust the attitude of the flying vest in real time by changing the thrust direction; the attitude stabilizing fin group fine-tunes the flight direction in real time through the principle of aerodynamics to ensure the attitude stability during flight. It should be noted that in this embodiment, vector thrust engines are added to both shoulders of the vest body, and an attitude stabilizing fin group is provided at the waist position. The vector thrust engines can quickly adjust the roll, pitch, and yaw angles of the flying vest by changing the thrust direction; the attitude stabilizing fin group fine-tunes the flight direction in real time relying on the aerodynamic design, so that even under high-speed or heavy-load flight conditions, a relatively stable flight posture can be maintained, making up for the deficiencies of the conventional multi-rotor system in terms of rapid lateral movement and lateral attitude control, improving the adaptability of the aircraft to variable wind directions and various sudden flight environments, and meeting the higher requirements for attitude control.

[0012] In some preferred embodiments, an intelligent flight control terminal is provided at the front chest position of the vest body; the intelligent flight control terminal includes a touch display screen, a voice interaction module, and a wireless data transmission module; the intelligent flight control terminal displays the flight altitude, speed, power, and navigation path in real time for the user to interactively control the flight state in real time. It should be noted that in this embodiment, the intelligent flight control terminal is provided at the front chest position of the vest body. The terminal includes a touch display screen, a voice interaction module, and a wireless data transmission module, which can display important information such as flight altitude, speed, power, and navigation path in real time during flight and allow the user to quickly operate or switch the flight mode by touch. Combining with the automatic analysis and decision-making ability of the deep learning control module, the intelligent flight control terminal can also receive the warning information or flight suggestions output by the algorithm in real time, providing an intuitive intervention channel for the user, thereby improving the usability and human-computer interaction level of the flight vest and avoiding the deficiencies such as complex operation, lack of feedback information, and difficulty in timely intervention in the prior art.

[0013] In some preferred embodiments, the voice interaction module includes a microphone array, a speaker, and a voice processing chip, and the voice processing chip is connected to the deep learning control module to support the user to perform real-time regulation on the flight path, speed, and attitude through voice commands. It should be noted that traditional personal aircraft rely on manual control, which is obviously inconvenient when used in high-speed or high-noise environments and is not conducive to dealing with sudden situations. By capturing multi-angle voice signals through the microphone array and using the voice processing chip combined with the deep learning control module for voice recognition and command parsing, the flight vest can accurately identify the user's password and execute the regulation command in real time in a noisy environment. The speaker is used to immediately broadcast the flight state or warning information, thereby significantly reducing the occupation of the user's hands, making the flight operation more efficient and convenient, and also providing a more flexible command input method in emergency tasks such as rescue and reconnaissance, significantly enhancing the practicality and safety of the system.

[0014] In some preferred embodiments, the vision sensor includes at least two high-definition wide-angle cameras, which are respectively installed on the left and right shoulders of the vest body. The high-definition wide-angle cameras have a 360-degree panoramic image stitching function, and can collect dynamic images of the surrounding environment in real time and input them into the deep learning control module for environmental analysis. It should be noted that in practical applications, it is often difficult for existing personal flying vehicles to achieve full-range environmental monitoring. Especially in densely populated areas or complex terrains, more comprehensive dynamic perception of the surrounding environment is required. In this embodiment, at least two high-definition wide-angle cameras are respectively installed on the left and right shoulders of the vest body. By using their 360-degree panoramic image stitching function, dynamic images of the surrounding environment are collected in real time and input into the deep learning control module for environmental analysis. Through the multi-camera and multi-angle image acquisition method, the probability of the appearance of visual blind spots can be significantly reduced, and the safety and mobility in narrow spaces or urban three-dimensional traffic can be improved. At the same time, the obtained image information can be fused with other sensor data to provide accurate environmental perception input for the deep learning control module, achieving a more perfect autonomous flight and obstacle avoidance effect, thereby overcoming the problems of single information collection and many blind spots of traditional flying devices in complex environments.

[0015] In some preferred embodiments, the lidar sensor and the ultrasonic sensor are respectively installed in front of and below the vest body, and are used to detect the position, distance and movement state of obstacles on the flight path in real time. The data of the position, distance and movement state of the obstacles on the flight path detected are fed back to the deep learning control module for real-time path planning and active obstacle avoidance. It should be noted that when flying in narrow urban streets, jungle areas or disaster sites, timely detection and avoidance of obstacles in front of and below are the keys to ensuring safety. In this embodiment, the lidar sensor and the ultrasonic sensor are respectively installed in front of and below the vest body to detect the position, distance and movement state of obstacles on the flight path in real time. The various types of perception data obtained can be quickly fed back to the deep learning control module to realize multi-level monitoring of the space environment and obstacle recognition. Since the lidar sensor has a higher positioning accuracy in the medium and long distances, and the ultrasonic sensor is more adept at quickly detecting obstacles at close range, the collaborative work of the two can significantly improve the real-time performance and accuracy of flight path planning and active obstacle avoidance. Thus, the collision risk caused by complex environmental interference, sensor blind spots or distance recognition errors can be effectively reduced, making the flying vest more adaptable to the changeable and potentially dangerous outdoor scenarios.

[0016] In some preferred embodiments, the height sensor adopts a combination form of a laser rangefinder and a barometric altimeter. The laser rangefinder measures the near-ground height data in real time, and the barometric altimeter measures the relative altitude data in real time. The near-ground height data and the relative altitude data are jointly fed back to the deep learning control module for precise flight height control. It should be noted that when flying in an area with a large terrain height difference, relying solely on a single barometric altimeter or laser ranging sensor may result in measurement errors, thereby causing problems with inaccurate height control. In this embodiment, the laser rangefinder and the barometric altimeter are used in combination, and the obtained near-ground height data and relative altitude data are jointly fed back to the deep learning control module for comprehensive analysis. The laser rangefinder can achieve high-precision ground distance measurement at close range, while the barometric altimeter can obtain the height information of the aircraft relative to the sea level according to the change in atmospheric pressure. The combination of these two data sources can significantly enhance the robustness and accuracy of height control, providing a safer and more stable flight guarantee for the flight vest in a changing outdoor environment. At the same time, this method of dual-sensor fusion height measurement can also provide better dynamic adaptability when the wind direction suddenly changes or the atmospheric pressure fluctuates, meeting the flight requirements in multiple scenarios.

[0017] In some preferred embodiments, the GPS positioning module has the function of RTK real-time differential positioning. By combining satellite positioning signals with base station differential correction signals, it can calculate the precise spatial position of the aircraft in real time, ensuring accurate navigation and path tracking. It should be noted that for high-precision navigation and precise flight path tracking, general GPS positioning is prone to large positioning errors in areas with dense urban high-rise buildings or signal interference environments, making it difficult to meet the requirements of safe flight and complex tasks. In this embodiment, by introducing the RTK real-time differential positioning function into the GPS positioning module and combining the base station differential correction signals with satellite positioning signals, the accuracy of the position data can be improved to the centimeter level. This high-precision positioning method can not only ensure stable and accurate spatial coordinate information in both urban canyon and open field scenarios, but also work in coordination with the deep learning control module to achieve refined control of the aircraft's current coordinates, path planning, and target tracking. With the help of RTK technology, this embodiment can better meet the needs of emergency rescue, precise delivery, or automatic cruise in complex environments, providing a more accurate position reference for the flight vest to avoid obstacles autonomously in narrow spaces or areas with high-rise buildings.

[0018] In some preferred embodiments, a fireproof, wear-resistant, and waterproof coating is provided on the surface of the vest body, and a shock-absorbing and buffering structure is provided inside the vest body. It should be noted that when flying in an outdoor environment, the flight vest may face various adverse factors, such as high temperature, open fire, friction, wind and rain, and impact. If the material strength of the vest body is insufficient or there is a lack of protective measures, it is extremely easy to cause equipment damage or safety risks to the user. In this embodiment, a fireproof, wear-resistant, and waterproof coating is provided on the surface of the vest body, and a shock-absorbing and buffering structure is provided inside it. This structure can not only delay the invasion of fire in case of a sudden fire or high-temperature scenario and reduce the damage of high temperature to the power device and electronic components, but also prevent surface wear or water immersion when passing through dangerous environments such as forests, shrubs, or collapsed building ruins. The internal shock-absorbing and buffering structure can more effectively absorb the vibration and impact during takeoff, landing, and flight, providing a relatively stable working environment for the deep learning control module and various sensors, and further improving the overall safety and durability of the flight vest.

[0019] In some preferred embodiments, the deep learning control module has a complex environment flight model database built in, and the real-time environmental data is compared with the data in the complex environment flight model database to perform environmental prediction, self-optimization of flight strategies, and active early warning of abnormal states. It should be noted that during the flight process, various complex environmental factors (such as air flow disturbance, multi-obstacle scenarios, meteorological mutations, etc.) may pose challenges to the safety and stability of the flight vest. Making judgments based solely on real-time perception data is prone to limitations and it is difficult to handle potential non-conventional scenarios in a timely manner. In this embodiment, a complex environment flight model database is built in the deep learning control module, enabling the system to compare the currently detected environmental data with the reference data in historical big data or professional models, and make predictions in advance for possible extreme flight situations. Through self-optimization using this database, the deep learning control module can not only dynamically update flight strategies, but also identify abnormal states and give active early warnings, thus effectively avoiding flight hazards and improving the system's autonomous decision-making ability and safety redundancy in different scenarios.

[0020] In some preferred embodiments, the intelligent flight control system further includes an emergency landing module; when the deep learning control module detects engine failure, insufficient battery power or environmental risk exceeding the safety threshold during flight, the emergency landing module automatically activates the safe landing procedure to ensure the safety of personnel. It should be noted that when the personal aircraft is flying in urban or wild environments, engine failure, insufficient battery power and external environmental risks may all occur instantaneously. Without a perfect early warning and emergency mechanism, the safety of users and surrounding personnel will face great threats. In this embodiment, an emergency landing module is added to the intelligent flight control system, so that when the deep learning control module detects an abnormal or high-risk state of key components, the safe landing procedure is automatically triggered, so that a relatively safe position can be intelligently selected and quickly landed in the shortest time, effectively avoiding serious accidents such as hard landings of the flight vest.

[0021] In some preferred embodiments, an intelligent load monitoring module is provided inside the vest body. The intelligent load monitoring module measures the aircraft load data in real time and feeds it back to the deep learning control module to automatically adjust the flight parameters and engine output power to ensure flight stability under load. It should be noted that when performing rescue, material delivery or multi-user collaborative tasks, the flight vest needs to carry various loads with different weights and volumes, such as rescue equipment or items. Without real-time monitoring and dynamic adjustment of the current load, the flight attitude and power control will face the risk of imbalance. In this embodiment, an intelligent load monitoring module is configured inside the vest body to detect the load information of the aircraft in real time and send it to the deep learning control module. Through comprehensive analysis of the load data by deep learning algorithms, the power of the main flight power device or the thrust of the attitude auxiliary adjustment device can be adjusted in time to maintain balance and efficiency, ensuring stable and safe flight under various load conditions.

[0022] In some preferred embodiments, the intelligent flight control system is provided with a data wireless transmission module to realize data sharing and collaborative control with the ground control station or other aircraft, meeting the complex task requirements during group flight and collaborative rescue. It should be noted that a single aircraft can have a certain degree of autonomy in a narrow environment, but in complex scenarios such as large-scale search and rescue, group collaborative operations or urban air traffic dispatching, information sharing between multiple aircraft and unified command of the ground control station are required. In this embodiment, a data wireless transmission module is provided in the intelligent flight control system to realize real-time data exchange and collaborative control with external platforms (such as ground control stations, other aircraft), allowing multiple flight vests to share information such as obstacles, weather or mission status, improving the overall mission efficiency and group operation safety, and meeting the higher-level collaborative flight requirements under multiple applications.

[0023] In some preferred embodiments, the intelligent flight control system is provided with an automatic return module. When there is a data communication interruption, a navigation signal loss, or the deep learning control module fails to obtain stable data input for the flight vest, the automatic return module automatically activates the return mode and safely returns to the takeoff point according to the recorded initial path. It should be noted that during the actual use of the flight vest, sudden situations such as signal interference, data communication interruption, the deep learning control module crashing, or navigation signal loss may occur. If normal control cannot be restored in time, safety accidents are very likely to occur. To reduce such out-of-control risks, in this embodiment, an automatic return module is set up to automatically perform a return operation based on the initial trajectory or a pre-recorded safe path when a data communication interruption or abnormal navigation signal is detected. This function can maximize the protection of personnel and equipment safety in a lost connection or failure state, thereby effectively improving the system fault tolerance rate and further improving the emergency response ability in unpredictable situations.

[0024] In some preferred embodiments, when the deep learning control module automatically optimizes the flight state, controls the main flight power device to provide lift and propulsion power, and uses the attitude auxiliary adjustment device to automatically correct the flight attitude, it includes: the deep learning control module adopts a flight control algorithm based on reinforcement learning, and pre-sets the state space, action space, and reward function in the algorithm initialization stage. The state space includes multi-source information (data from visual sensors, lidar sensors, ultrasonic sensors, altitude sensors, and GPS positioning modules) acquired by the environmental perception system. The action space includes adjustment instructions for the output power of the main flight power device and the vector direction of the attitude auxiliary adjustment device; during the flight, the deep learning control module sends a power adjustment signal to the main flight power device according to the current state to provide appropriate lift and propulsion power, and sends a vector angle correction signal to the attitude auxiliary adjustment device to automatically correct the current flight attitude. When it is detected that the flight attitude stability, energy consumption, obstacle distance, or flight path deviation meets the preset requirements, a relatively high reward is given to the corresponding reward value. Thus, through continuous iterative learning and exploration, the flight stability and energy-saving efficiency are gradually improved, and the automatic optimization of the flight state in different outdoor environments is realized. It should be noted that in the prior art, personal aircraft usually rely on fixed control strategies or manual operations to achieve the adjustment of flight attitude and the allocation of power output, and it is difficult to maintain safe, stable, energy-saving, and efficient flight in changing outdoor environments or complex task scenarios. In this embodiment, through the flight control algorithm of reinforcement learning, the definitions of the state space, action space, and reward function are clarified in the algorithm initialization stage; the state space covers the input of multi-source information composed of visual sensors, lidar sensors, ultrasonic sensors, altitude sensors, GPS positioning modules, etc., and the action space includes real-time adjustment instructions for the main flight power device and the attitude auxiliary adjustment device. By dynamically evaluating important indicators such as flight attitude stability, energy consumption, obstacle distance, and path deviation during the flight, and setting the reward value corresponding to excellent performance as a high reward, the system can spontaneously optimize the strategy in continuous iterative experiments and explorations, and gradually improve its adaptability to various emergencies, terrain changes, and environmental interferences. At the same time, the algorithm guides the deep learning control module to balance multiple requirements such as rapid maneuverability, energy consumption efficiency, and safety factor through an internal reward mechanism, so as to achieve the automatic optimization of the flight state and the stable correction of the flight attitude.

[0025] In some preferred embodiments, when the deep learning control module automatically optimizes the flight state, controls the main flight power device to provide lift and propulsion power, and uses the attitude auxiliary adjustment device to automatically correct the flight attitude, it includes: the deep learning control module adopts an adaptive flight algorithm based on neural network predictive control. First, a multi-layer perceptron (MLP) or a convolutional neural network (CNN) is used to extract features and estimate the state of the real-time data collected by the environmental perception system, and a comprehensive evaluation vector of the flight attitude, wind direction and speed, obstacle distribution, and target navigation information is obtained; Subsequently, a recurrent neural network (RNN) or a long short-term memory network (LSTM) including time series analysis is used to predict the state changes at consecutive moments to obtain an estimated value of the future flight state; According to the difference between the estimated value and the desired flight target, power allocation instructions for the main flight power device and thrust vector correction instructions for the attitude auxiliary adjustment device are generated, so as to ensure the stable operation of the flight vest in a dynamic environment; During the flight, the neural network weight parameters will be adaptively updated with data accumulation and error feedback to continuously optimize the flight trajectory and attitude adjustment strategy, and achieve high-precision active obstacle avoidance, automatic cruise, and flight energy consumption management. It should be noted that in a complex and changing outdoor flight environment, the aircraft needs to make quick and accurate judgments and responses to dynamic external information (including airflow changes, obstacle distribution, and target navigation requirements, etc.). Traditional control methods often have deficiencies in dealing with multi-sensor, large-scale non-linear data, and fast time-varying states, and it is difficult to achieve effective prediction and adaptive control of future states. In this embodiment, an adaptive flight algorithm based on neural network predictive control is adopted, and various deep learning structures such as a multi-layer perceptron (MLP) or a convolutional neural network (CNN) and a recurrent neural network (RNN) or a long short-term memory network (LSTM) are integrated in the implementation steps to give full play to their respective advantages in feature extraction and time series prediction. By extracting features and estimating the state of real-time environmental data, a comprehensive evaluation vector of flight attitude, wind direction and speed, obstacle distribution, and target navigation information is formed, and a time series model is used to predict the flight state for a period of time in the future, so as to formulate power allocation and thrust vector correction strategies in advance. At the same time, the neural network weights will be continuously accumulated with data and gradually corrected with error feedback to form an adaptive update mechanism, so that the flight vest can still maintain good route control and attitude stability in a changing external environment. Through this method, the tolerance and response speed to sudden environmental disturbances can be greatly improved, and the core capabilities such as active obstacle avoidance, automatic cruise, and energy consumption management during flight can be significantly enhanced, and finally a high-precision, safe and stable flight effect can be achieved.

[0026] It should be noted that the above embodiments are only preferred specific embodiments of the present invention, and the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. The protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A smart outdoor flight vest based on deep learning, characterized in that: include: A vest body, which is provided to a user for wearing, and an intelligent flight control system is arranged on the vest body, and the intelligent flight control system includes a deep learning control module, a main flight power device, an attitude auxiliary adjustment device and an environmental perception system; The deep learning control module is connected to the main flight power unit, the attitude auxiliary adjustment device and the environmental perception system through a data communication line; the environmental perception system includes a visual sensor, a lidar sensor, an ultrasonic sensor, an altitude sensor and a GPS positioning module; the visual sensor, the lidar sensor, the ultrasonic sensor, the altitude sensor and the GPS positioning module are all used to collect outdoor environmental data in real time and transmit them to the deep learning control module; the deep learning control module analyzes the environmental data in real time, automatically optimizes the flight state through an autonomous learning algorithm, controls the main flight power unit to provide lift and propulsion power, and uses the attitude auxiliary adjustment device to automatically correct the flight attitude.

2. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: The main flight power unit includes a multi-rotor electric engine group and a power battery pack installed on the back of the vest body; the multi-rotor electric engine group includes at least four horizontally symmetrically distributed electric rotors; the power battery pack is located on the rear side of the vest body and is electrically connected to the multi-rotor electric engine group to continuously provide electrical energy to the multi-rotor electric engine group.

3. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: The attitude auxiliary adjustment device includes vector thrust engines arranged on the shoulders of both sides of the vest body and an attitude stabilizing wing group arranged at the waist position of the vest; the vector thrust engine adjusts the attitude of the flying vest in real time by changing the thrust direction; the attitude stabilizing wing group fine-tunes the flight direction in real time through the principles of aerodynamics to ensure attitude stability during flight.

4. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: An intelligent flight control terminal is arranged at the front chest position of the vest body; the intelligent flight control terminal includes a touch display screen, a voice interaction module and a wireless data transmission module; the intelligent flight control terminal displays the flight altitude, speed, power and navigation path in real time, allowing users to interactively control the flight status in real time.

5. The deep learning-based intelligent outdoor flight vest according to claim 4 is characterized in that: The voice interaction module includes a microphone array, a speaker and a voice processing chip. The voice processing chip is connected to the deep learning control module to support users to control the flight path, speed and posture in real time through voice commands.

6. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: The visual sensor includes at least two high-definition wide-angle cameras, which are respectively installed on the left and right shoulders of the vest body. The high-definition wide-angle camera has a 360-degree panoramic image stitching function, which collects dynamic images of the surrounding environment in real time and inputs them into the deep learning control module for environmental analysis.

7. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: The laser radar sensor and the ultrasonic sensor are respectively installed in front of and below the vest body to detect the position, distance and movement status of obstacles on the flight path in real time. The detected obstacle position, distance and movement status data on the flight path are fed back to the deep learning control module for real-time path planning and active obstacle avoidance.

8. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: The altitude sensor is a combination of a laser rangefinder and a barometric altimeter. The laser rangefinder measures the ground altitude data in real time, and the barometric altimeter measures the relative altitude data in real time. The ground altitude data and the relative altitude data are jointly fed back to the deep learning control module for precise flight altitude control.

9. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: The GPS positioning module has an RTK real-time differential positioning function, which calculates the precise spatial position of the aircraft in real time through satellite positioning signals and base station differential correction signals to ensure accurate navigation and path tracking.

10. The deep learning-based intelligent outdoor flight vest according to claim 1, characterized in that: The surface of the vest body is provided with a fireproof, wear-resistant and waterproof coating, and the interior of the vest body is provided with a shock-absorbing and buffering structure.

Images