Autonomous navigation and obstacle avoidance method and system for intelligent aircraft

Abstract

The invention discloses an autonomous navigation and obstacle avoidance method for an intelligent aircraft. The autonomous navigation and obstacle avoidance method comprises the following steps: S1, multi-modal data acquisition; s2, generating a motion path; s3, predicting an obstacle path; s4, flight obstacle avoidance; the invention discloses an autonomous navigation and obstacle avoidance system for an intelligent aircraft. The system comprises a sensor group, an edge calculation unit, an execution mechanism, an environment sensing and positioning module, a decision control module and a communication module, the autonomous navigation and obstacle avoidance capabilities of the aircraft are remarkably improved through multi-modal data fusion, three-dimensional environment map generation, dynamic obstacle prediction, path planning and attitude control; semantic tags are introduced to identify obstacle types, geometric consistency scores ensure authenticity and reliability of obstacles, and radar reflection intensity and point cloud density provide spatial distribution and material information; and the dynamic obstacle prediction model dynamically adjusts an obstacle avoidance strategy in combination with semantic tags, geometric consistency scores, radar reflection intensity and point cloud density.

Description

Technical Field

[0001] The present invention relates to obstacle avoidance for intelligent aircraft, and more specifically, to a method for autonomous navigation and obstacle avoidance of intelligent aircraft, and to a system for autonomous navigation and obstacle avoidance of intelligent aircraft. Background Art

[0002] Intelligent aircraft are a peak achievement of modern technology. They represent the progress of human science and technology, bringing us an unprecedented flying experience and convenience. The emergence of intelligent aircraft enables us to break through geographical limitations, explore unknown areas, and realize our flying dreams.

[0003] First of all, an intelligent aircraft refers to an aircraft that can fly autonomously, perceive the environment, and make corresponding decisions. Different from traditional aircraft, intelligent aircraft are equipped with various high-tech devices, such as sensors, cameras, global positioning systems, etc., which can obtain environmental information in real time and make intelligent responses. It analyzes and processes data through a built-in computer system, enabling it to perform tasks more intelligently.

[0004] The existing aircraft navigation and obstacle avoidance technologies have the following problems:

[0005] Insufficient environmental perception: Relying on a single sensor is vulnerable to interference and difficult to handle complex dynamic scenarios.

[0006] Poor real-time performance: Traditional obstacle avoidance algorithms lag in response in dynamic obstacle scenarios.

[0007] Rigid path planning: The lack of coordination between global planning and local obstacle avoidance causes the aircraft to frequently fall into local optima or repeatedly adjust the path. Summary of the Invention

[0008] An object of the present invention is to provide a new technical solution for a method and system for autonomous navigation and obstacle avoidance of intelligent aircraft.

[0009] According to a first aspect of the present invention, there is provided a method for autonomous navigation and obstacle avoidance of intelligent aircraft, including the following steps:

[0010] S1. Multi-modal data collection: Collect multi-modal data, preprocess the multi-modal data, and then fuse the data to generate a three-dimensional environmental map;

[0011] S2. Generate a motion path: Receive the three-dimensional environmental map and generate a preset path on the three-dimensional environmental map;

[0012] S3. Predict the obstacle path: Identify the obstacles and predict the obstacle path through an obstacle motion prediction model;

[0013] S4, Flight Obstacle Avoidance: Determine the collision intersection point between the preset path and the obstacle path, and dynamically adjust the pitch angle, roll angle, and yaw angle of the aircraft. Achieve smooth maneuvering through PID control to avoid sudden stops or jitters.

[0014] An intelligent aircraft autonomous navigation and obstacle avoidance system, including a sensor group, an edge computing unit, an actuator, an environment perception and positioning module, a decision-making control module, and a communication module;

[0015] The sensor group includes a binocular vision camera and an integrated lidar. The binocular vision camera is used to collect visual data, and the integrated lidar is used to collect radar data to obtain environmental data and obstacle data;

[0016] The edge computing unit performs computational analysis processing of various algorithms, including a processing module for preprocessing multi-modal data, fusing the data to generate a three-dimensional environmental map, and an environment perception and positioning module for generating a preset path and an obstacle path, determining the collision intersection point between the preset path and the obstacle path, and dynamically adjusting the pitch angle, roll angle, and yaw angle of the aircraft, and a decision-making control module for formulating an avoidance path;

[0017] The actuator includes a brushless motor and a servo controller. The brushless motor and the servo controller are used to control and adjust the flight angle, direction, and speed;

[0018] The communication module supports data information transmission between multiple aircraft and the edge computing unit, and realizes data information transmission between the aircraft and the remote controller.

[0019] Advantages of the present invention:

[0020] Multi-modal data fusion and three-dimensional environmental map generation: Collect data information through an integrated lidar and binocular vision, and fuse the radar data and laser data to achieve the construction of a high-precision three-dimensional environmental map, which can provide accurate geographical information in a complex dynamic environment and lay a solid foundation for path planning and obstacle recognition;

[0021] Introduce semantic tags to help identify the type of obstacle, and the geometric consistency score ensures the authenticity and reliability of the obstacle. The radar reflection intensity and point cloud density provide spatial distribution and material information; The dynamic obstacle prediction model combines semantic tags and geometric consistency scores to ensure the accuracy and reliability of the predicted obstacle trajectory. The radar reflection intensity and point cloud density provide real-time status information of the obstacle, and dynamically adjust the obstacle avoidance strategy; Based on the predicted obstacle path, the path planning of the aircraft can be dynamically adjusted to avoid high-risk areas. The semantic tags and geometric consistency scores ensure the accuracy of obstacle recognition, and the reflection intensity and point cloud density help the aircraft determine whether the obstacle is a traversable area;

[0022] Dynamic obstacle path prediction and collision avoidance decision-making: Based on the Markov chain model, dynamically predict the movement trajectory of obstacles, simulate the behavior of obstacles through the state transition probability matrix, predict their future positions and paths, and adjust the collision avoidance strategy in real time;

[0023] Adaptive attitude control and smooth maneuvering: According to the collision intersection points between the preset path and the obstacle path, dynamically adjust the pitch angle, roll angle, and yaw angle of the aircraft, and achieve smooth maneuvering through the PID controller to avoid sudden stops or jitters; Adaptive attitude control can ensure the stability of the aircraft during collision avoidance and improve the overall flight performance;

[0024] Efficient implementation of the system architecture: The system adopts a modular architecture, including a sensor group, an edge computing unit, an actuator, and a communication module; The sensor group provides multi-source data input, the edge computing unit processes the data and generates a three-dimensional environment map, the actuator controls the actions of the aircraft, and the communication module ensures data sharing and command and control; It ensures the real-time performance and reliability of the system, enabling the aircraft to execute tasks safely and efficiently in complex environments;

[0025] In summary, the present invention improves the autonomous navigation and collision avoidance capabilities of the aircraft through multi-modal data fusion, three-dimensional environment map generation, dynamic obstacle prediction, path planning, and attitude control; It realizes efficient and safe flight in various complex environments such as dense urban environments, forest areas, and dynamic traffic scenarios.

[0026] Through the following detailed description of the exemplary embodiments of the present invention with reference to the accompanying drawings, other features and advantages of the present invention will become clear. Brief Description of the Drawings

[0027] The accompanying drawings incorporated in and constituting a part of this specification illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention.

[0028] Figure 1 It is a schematic diagram of the step flow of a method for autonomous navigation and collision avoidance of an intelligent aircraft in one embodiment;

[0029] Figure 2 It is a schematic diagram of the generation steps of a preset path of a method for autonomous navigation and collision avoidance of an intelligent aircraft in another embodiment;

[0030] Figure 3 It is a schematic diagram of the system structure of a system for autonomous navigation and collision avoidance of an intelligent aircraft in still another embodiment. Detailed Description of the Embodiments

[0031] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that: unless otherwise specifically stated, the relative arrangements, numerical expressions, and numerical values of the components and steps set forth in these embodiments do not limit the scope of the present invention.

[0032] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way a limitation on the present invention or its application or use.

[0033] Technologies, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such technologies, methods, and devices should be regarded as part of the specification.

[0034] In all the examples shown and discussed herein, any specific values should be construed as merely exemplary and not as a limitation. Accordingly, other examples of the exemplary embodiments may have different values.

[0035] As Figure 1-2 shown, an intelligent aircraft autonomous navigation and obstacle avoidance method includes the following steps:

[0036] S1. Multimodal data acquisition: Acquire multimodal data, preprocess the multimodal data, and then fuse the data to generate a three-dimensional environmental map;

[0037] S2. Generate a motion path: Receive the three-dimensional environmental map and generate a preset path on the three-dimensional environmental map;

[0038] S3. Predict the obstacle path: Identify the obstacles and predict the obstacle path through an obstacle motion prediction model;

[0039] S4. Flight obstacle avoidance: Determine the collision intersection point between the preset path and the obstacle path, and dynamically adjust the pitch angle, roll angle, and yaw angle of the aircraft, and achieve smooth maneuvering through PID control to avoid sudden stops or jitters.

[0040] In this embodiment, preferably, the acquisition of multimodal data in S1 includes visual data and radar data. The visual data is captured by a binocular vision camera, and the radar data is acquired by an integrated lidar;

[0041] The multimodal data fuses the visual data and radar data in real time through an extended Kalman filter to construct a high-precision three-dimensional environmental map;

[0042] It should be noted that the multimodal data is acquired through a binocular vision camera and a lidar, and the multimodal data fuses the three-dimensional environmental map in real time through an extended Kalman filter, which is convenient for generating the path of the intelligent aircraft and the path of the obstacles.

[0043] In this embodiment, preferably, the visual data includes the pixel coordinates, depth, semantic label, and geometric consistency score of the feature points, and the pixel coordinates, depth, semantic label, and geometric consistency score are used to generate a visual data group:

[0044]

[0045] Among them, Z vision represents the actual measurement value of the binocular vision camera, h vision (X) represents the predicted visual measurement value according to state X, and state X represents the pose and motion information of the aircraft, v vision represents the visual observation noise, u and v represent the projection coordinates of the feature point in the binocular vision image, d represents the depth value calculated by the binocular disparity, s c represents the confidence score of the semantic label, and Confidence represents the geometric consistency score of the feature match;

[0046] It should be noted that the environmental information is provided in real time to support the recognition and status update of dynamic obstacles, and based on the predicted obstacle path, the smoothness of the path is optimized. The semantic label and geometric consistency score ensure the accuracy of obstacle recognition.

[0047] In this embodiment, preferably, the s c represents the confidence score of the semantic label and is calculated as follows:

[0048] s c = Softmax(W c x + b c );

[0049] Among them, W c and b c represent the weight and bias parameters related to the semantic label c, x represents the feature vector in the view, s c represents the confidence score of the semantic label c under the input image, and Softmax represents a function used to normalize the confidence;

[0050] The geometric consistency score of the feature match of the Confidence is calculated as follows:

[0051]

[0052] Among them, σ represents a parameter that controls the error tolerance. The smaller the error, the closer the confidence is to 1; the larger the error, the closer the confidence approaches 0, and error represents the reprojection error of the homography matrix;

[0053] It should be noted that semantic tags are introduced to help identify the types of obstacles, geometric consistency scoring ensures the authenticity and reliability of obstacles, and the dynamic obstacle prediction model combines semantic tags and geometric consistency scoring to ensure the accuracy and reliability of the predicted obstacle trajectories.

[0054] In this embodiment, preferably, the radar data includes three-dimensional point cloud coordinates, reflection intensity, and point cloud density, and the three-dimensional point cloud coordinates, reflection intensity, and point cloud density are used to generate a laser data set:

[0055]

[0056] Among them, Z lidar represents the actual laser measurement value, h lidar (X) represents the predicted laser measurement value according to state X, and state X represents the pose and motion information of the aircraft, v lidar represents the laser measurement noise, x L , y L , z L represent the coordinates of the point cloud in the global coordinate system, I represents the reflection intensity value, and through normalization mapping, it is in [0,1], represents the number of point clouds per unit volume and is used to determine whether the obstacle is a penetrable object;

[0057] It should be noted that providing three-dimensional point cloud coordinates, reflection intensity, and point cloud density can accurately measure the position, shape, and material properties of obstacles, provide real-time state information of dynamic obstacles, such as pedestrians walking and vehicles moving, assist in dynamic obstacle avoidance decisions, provide environmental information for path planning, and ensure that the aircraft can avoid obstacles.

[0058] In this embodiment, preferably, the calculation of the reflection intensity value I is as follows:

[0059]

[0060] Among them, R represents the absolute value of the reflection intensity, with a range of 0 < R < 1, d represents the distance from the laser beam to the target object, and the reflection intensity represents the magnitude of the electromagnetic energy reflected by the laser beam back to the sensor and is used to identify the material properties of the object;

[0061] The number of point clouds per unit volume is calculated as follows:

[0062]

[0063] Among them, N represents the number of point clouds in the area; A represents the area of the area and is used to evaluate the aggregation degree of the obstacles and assist the UAV in making dynamic obstacle avoidance decisions;

[0064] It should be noted that the radar reflection intensity and point cloud density provide spatial distribution and material information; the radar reflection intensity and point cloud density provide real-time status information of obstacles, and the reflection intensity and point cloud density help the aircraft determine whether the obstacle is a traversable area.

[0065] In this embodiment, preferably, the steps for generating the preset path in S2 are as follows:

[0066] Initial state: Starting from the starting point, gradually expand the surrounding flyable area;

[0067] Node generation: Generate multiple possible nodes within the expanded area;

[0068] Path evaluation: Evaluate the pros and cons of each node through a heuristic function and select the next step with the lowest path cost;

[0069] Dynamic obstacle handling: Use the dynamic window method to adjust the path in real time, exclude obstacles in the generated path, and re-evaluate the path cost;

[0070] Path optimization: Combine the dynamic constraints of the aircraft to optimize the continuity and smoothness of the path to ensure that the aircraft can fly smoothly along the generated path;

[0071] It should be noted that the preset path of the intelligent aircraft is generated through a high-precision three-dimensional environmental map, and the obstacles in the generated path are excluded, the path cost is re-evaluated, and the continuity and smoothness of the path are optimized in combination with the dynamic constraints of the aircraft to ensure the smooth flight of the intelligent aircraft.

[0072] In this embodiment, preferably, the obstacle motion prediction model in S3 uses a Markov chain-based obstacle path and is calculated as follows:

[0073] The state of the dynamic obstacle is defined as follows:

[0074] Position state (x, y), velocity state (v x , v y ), acceleration state (a x , a y ), direction state θ, behavior state s e , representing the behavior pattern of the obstacle,

[0075] State space S = [x, y, v x , v y , a x , a y , θ, s e ;

[0076] π t+1 (S') = π t(S')P + Q(S'),

[0077] where P represents the state transition probability matrix, and π t (S') represents the state distribution at time t, and Q(S') represents the noise term for the robustness of the model, and π t+1 (S') represents the state distribution at time t + 1;

[0078] If there is an intersection between the obstacle path calculated by π t+1 (S') and the preset path, then the PID control is triggered to adjust the flight route of the aircraft. If there is no intersection between the obstacle path calculated by π t+1 (S') and the preset path, then the aircraft flies according to the preset path;

[0079] It should be noted that according to the collision intersection point between the preset path and the obstacle path, the pitch angle, roll angle and yaw angle of the aircraft are dynamically adjusted, and smooth maneuvering is achieved through the PID controller to avoid sudden stops or jitters; the adaptive attitude control can ensure the stability of the aircraft during obstacle avoidance and improve the overall flight performance.

[0080] In this embodiment, preferably, the PID control in S4 is used for the dynamic adjustment of the aircraft attitude, and the goal is to map the intersection position error to the attitude adjustment amount;

[0081]

[0082] where e p represents the error between the attitude adjustment target and the current attitude, and K p , K i , K d represent the proportional, integral and differential coefficients of the PID controller, represents the attitude adjustment amount output by the controller;

[0083] The detection of the attitude adjustment target is determined by the reflection intensity value I and the number of point clouds per unit volume If I > 0.7 AND ρ > 50, then it means that there is no gap in the obstacle and it is not traversable;

[0084] If I < 0.7 AND ρ < 50, then it means that there is a gap in the obstacle and it is traversable;

[0085]

[0086] where Δθ 仰 , Δθ 滚 , Δθ 航 , Δθ 位 , Δθ 速They are respectively represented as the pitch difference angle, roll difference angle, yaw difference angle for the attitude adjustment of the aircraft, the three-dimensional distance difference of the aircraft position, and the difference in flight speed. λ1, λ2, λ3, λ4, and λ5 are respectively represented as the weight values of the pitch difference angle, roll difference angle, yaw difference angle, the three-dimensional distance difference of the aircraft position, and the difference in flight speed, and λ1 + λ2 + λ3 + λ4 + λ5 = 1. For example, they can all be represented as 0.2, or can be set according to the actual situation, such as 0.18, 0.23, 0.14, 0.19, 0.26;

[0087] It should be noted that according to the collision intersection point of the preset path and the obstacle path, the pitch angle, roll angle, and yaw angle of the aircraft are dynamically adjusted, and smooth maneuvering is achieved through a PID controller to avoid sudden stops or jitters; the adaptive attitude control can ensure that the aircraft remains stable during obstacle avoidance and improve the overall flight performance.

[0088] Reference Figure 3 , a system for autonomous navigation and obstacle avoidance of an intelligent aircraft, including a sensor group, an edge computing unit, an actuator, an environment perception and positioning module, a decision-making control module, and a communication module;

[0089] The sensor group includes a binocular vision camera and an integrated lidar. The binocular vision camera is used to collect visual data, and the integrated lidar is used to collect radar data to obtain environmental data and obstacle data;

[0090] The edge computing unit performs computational analysis processing of various algorithms, including a processing module for preprocessing multi-modal data, fusing the data to generate a three-dimensional environmental map, and an environment perception and positioning module for generating a preset path and an obstacle path, determining the collision intersection point of the preset path and the obstacle path, and dynamically adjusting the pitch angle, roll angle, and yaw angle of the aircraft, and a decision-making control module for formulating an avoidance path;

[0091] The actuator includes a brushless motor and a servo controller. The brushless motor and the servo controller are used to control and adjust the flight angle, direction, and speed;

[0092] The communication module supports data information transmission between multiple aircraft and the edge computing unit, and realizes data information transmission between the aircraft and the remote controller;

[0093] It should be noted that the efficient implementation of the system architecture: The system adopts a modular architecture, including a sensor group, an edge computing unit, an actuator, and a communication module; the sensor group provides multi-source data input, the edge computing unit processes the data and generates a three-dimensional environmental map, the actuator controls the actions of the aircraft, and the communication module ensures data sharing and command and control; ensuring the real-time performance and reliability of the system, enabling the aircraft to safely and efficiently perform tasks in a complex environment.

[0094] Specific operation process of this application:

[0095] Step 1, Multimodal data collection: Collect multimodal data, preprocess the multimodal data, and then fuse the data to generate a three-dimensional environmental map;

[0096] Step 2, Generate a motion path: Receive the three-dimensional environmental map and generate a preset path on the three-dimensional environmental map;

[0097] Step 3, Predict the obstacle path: Identify the obstacles and predict the obstacle path through the obstacle motion prediction model;

[0098] Step 4, Flight obstacle avoidance: Determine the collision intersection point between the preset path and the obstacle path, and dynamically adjust the pitch angle, roll angle and yaw angle of the aircraft, and achieve smooth maneuvering through PID control to avoid sudden stop or jitter.

[0099] Although some specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for illustration purposes and not for limiting the scope of the present invention. Those skilled in the art should understand that the above embodiments can be modified without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. An autonomous navigation and obstacle avoidance method for an intelligent aircraft, characterized in that, It includes the following steps: S1. Multimodal data acquisition: Acquire multimodal data, preprocess the multimodal data, and then fuse the data to generate a three-dimensional environmental map; S2. Generate a motion path: Receive the three-dimensional environmental map and generate a preset path on the three-dimensional environmental map; S3. Predict the obstacle path: Identify obstacles and predict the obstacle path through an obstacle motion prediction model; S4. Flight obstacle avoidance: Determine the collision intersection point between the preset path and the obstacle path, and dynamically adjust the pitch angle, roll angle, and yaw angle of the aircraft, and achieve smooth maneuvering through PID control to avoid sudden stops or jitters.

2. The autonomous navigation and obstacle avoidance method of an intelligent aircraft according to claim 1, characterized in that: The acquisition of multimodal data in S1 includes visual data and radar data. The visual data is captured by a binocular vision camera, and the radar data is acquired by an integrated lidar; The multimodal data fuses visual data and radar data in real time through an extended Kalman filter to construct a high-precision three-dimensional environmental map.

3. The autonomous navigation and obstacle avoidance method of an intelligent aircraft according to claim 2, characterized in that: The visual data includes the pixel coordinates, depth, semantic labels, and geometric consistency scores of feature points, and the pixel coordinates, depth, semantic labels, and geometric consistency scores are used to generate a visual data group.

4. An autonomous navigation and obstacle avoidance method for an intelligent aircraft according to claim 3, characterized in that: The radar data includes the three-dimensional coordinates of point clouds, reflection intensity, and point cloud density, and the three-dimensional coordinates of point clouds, reflection intensity, and point cloud density are used to generate a laser data group.

5. A method for autonomous navigation and obstacle avoidance of an intelligent aircraft, characterized in that: The generation steps of the preset path in S2 are as follows: Initial state: Start from the starting point and gradually expand the flyable area around; Node generation: Generate multiple possible nodes within the expanded area; Path evaluation: Evaluate the advantages and disadvantages of each node through a heuristic function and select the next step with the lowest path cost; Dynamic obstacle handling: Use the dynamic window method to adjust the path in real time, exclude obstacles in the generated path, and re-evaluate the path cost; Path optimization: Combine the dynamic constraints of the aircraft to optimize the continuity and smoothness of the path to ensure that the aircraft can fly smoothly along the generated path.

6. The autonomous navigation and obstacle avoidance method of an intelligent aircraft according to claim 1, characterized in that: The obstacle motion prediction model in S3 uses an obstacle path based on a Markov chain.

7. A method for autonomous navigation and obstacle avoidance of an intelligent aircraft according to claim 1, characterized in that: The PID control in S4 is used for the dynamic adjustment of the aircraft attitude, and the goal is to map the intersection position error to the attitude adjustment amount; The detection of the attitude adjustment target is determined by the reflection intensity value I and the number of point clouds per unit volume for determination.

8. A system for autonomous navigation and obstacle avoidance of an intelligent aircraft, which system implements the method as described in claim 1, characterized in that: It includes a sensor group, an edge computing unit, an actuator, an environment perception and positioning module, a decision control module, and a communication module; The sensor group includes a binocular vision camera and an integrated lidar. The binocular vision camera is used to acquire visual data, and the integrated lidar is used to acquire radar data to obtain environmental data and obstacle data; The edge computing unit performs computational analysis processing of various algorithms, including a processing module for preprocessing multimodal data, fusing the data to generate a three-dimensional environmental map, and an environment perception and positioning module for generating a preset path and an obstacle path, determining the collision intersection point between the preset path and the obstacle path, and dynamically adjusting the pitch angle, roll angle, and yaw angle of the aircraft, and a decision control module for formulating an avoidance path; The actuator includes a brushless motor and a servo controller, and the brushless motor and the servo controller are used to control and adjust the flight angle, direction, and speed; The communication module supports data information transmission between multiple aircraft and the edge computing unit.

Images