A multi-rotor unmanned aerial vehicle surveying and mapping operation control method and system

Abstract

This invention relates to the field of multi-rotor UAV mapping technology, and discloses a multi-rotor UAV mapping control method and system. By constructing a segment description table around the sequence of flight path points, the geometric relationships, turning angle relationships, and mapping orientation relationships of each segment are unified and describable. Subsequently, during flight, the temporal changes in attitude angles and the lateral changes in the mapped image are simultaneously extracted, allowing the image results to enter the attitude control link. Then, a reference attitude sequence is generated based on the attitude change trends of the current segment, the next segment, and future times, and shooting interval data and image classification results are generated based on attitude consistency and image offset. Finally, the reference attitude data for the next segment is corrected using the statistical results of the current segment, thereby achieving continuous control and result determination between multiple segments. Therefore, this invention can solve the problem of maintaining image coverage continuity, lateral overlap stability, and imaging consistency between segments in narrow, continuous, and close-angle mapping scenarios.

Description

Technical Field

[0001] This invention relates to the field of multi-rotor unmanned aerial vehicle (UAV) surveying technology, and in particular to a multi-rotor UAV surveying operation control method and system. Background Technology

[0002] Multi-rotor drones, due to their flexible takeoff and landing, strong low-altitude close-range operation capabilities, and suitability for close-range high-precision image acquisition, have been widely used in scenarios such as campus building surveying, ancient building facade acquisition, park facility surveying, close-range photogrammetry, and digital modeling. In recent years, with the popularization of drone surveying teaching applications, multi-rotor drones have also been extensively used for flight path planning demonstrations, facade surveying training, and close-range modeling process demonstrations. Typically, teaching demonstrations tend to be conducted in open areas, so conventional control methods are mostly formed under regular flight segments and ordinary turning conditions.

[0003] However, in practical engineering applications of UAV surveying teaching, the surveying objects are often located in narrow courtyards between teaching buildings, continuous corridors, building corners, eaves edges, and other complex areas close to the facade. In these scenarios, the UAV not only needs to maintain a close flight distance along the facade but also needs to complete multiple directional changes within a short distance. Since the aircraft's roll, pitch, and yaw states directly affect the camera's field of view and image coverage area, a lack of coordination between attitude changes and imaging requirements can easily lead to problems such as the image subject shifting to one side, unstable lateral overlap of adjacent images, and local missed measurements in corner areas.

[0004] Existing technologies typically employ methods such as flight based on waypoints, following in a fixed direction, or capturing images at a fixed pace. While these methods can accomplish flight and data acquisition tasks in general scenarios, their control logic revolves primarily around the flight path itself, failing to directly incorporate the deviation of the surveyed object from the image center as input for subsequent attitude adjustments. Furthermore, existing technologies rarely quantify the spatial compactness between the current and next flight segments, making it difficult for the system to pre-allocate attitude changes before entering corners, thus hindering stable imaging results in narrow, continuous corner scenarios.

[0005] Therefore, how to provide a control method for multi-rotor UAV mapping operations that enables it to form a control link with continuous data transmission logic by utilizing the geometric relationship of flight segments, the temporal changes of attitude angles, and the image bias in close-range mapping scenarios with narrow and continuous turning angles, thereby improving the continuity of imaging coverage and the consistency of mapping operations, remains a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] This invention provides a multi-rotor UAV mapping operation control method and system to at least solve the problems of existing technologies in maintaining image coverage continuity, lateral overlap stability and imaging consistency between flight segments in narrow, continuous, and close-to-the-map mapping scenarios.

[0007] To achieve the above objectives, the present invention provides a control method for surveying and mapping operations using a multi-rotor unmanned aerial vehicle (UAV), the method comprising the following steps: Obtain the route point sequence corresponding to the surveying task, calculate the route direction data, route length data, adjacent route angle data and route compactness data, and generate a route description table; Based on the attitude angle time series data and mapping image sequence of multi-rotor UAVs during the mapping operation, attitude change intensity data and image lateral offset sequence are generated respectively. Combined with flight segment compactness data, scene compactness data and corner lead data are generated. Based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time, generate the attitude prediction sequence and reference attitude sequence for future time. Based on the reference attitude sequence, the current real-time attitude data, the image lateral offset sequence, and the flight segment compactness data, a shooting interval sequence is generated. The mapping image sequence is classified into main mapping frames and transition frames to form flight segment statistical results and update the reference attitude data for the next flight segment. Based on the updated reference attitude data for the next segment, control of the subsequent segments is continuously executed, and a consistency determination is performed after all segments are completed. When the determination meets the preset conditions, the mapping operation completion result is output.

[0008] Optionally, obtain the route point sequence corresponding to the surveying task, calculate the route direction data, route length data, adjacent route angle data, and route compactness data, and generate a route description table, specifically including: Read the spatial coordinate data of each route point, form a segment by combining two adjacent route points, and calculate the segment direction data and segment length data of each segment based on the spatial coordinate data of two adjacent route points. The turning angle data of adjacent segments is calculated based on the segment direction data of two adjacent segments, and the segment compactness data is calculated by combining the current segment length data and the next segment length data. The baseline yaw angle data for each flight segment is generated based on the flight segment direction data, lateral mapping relationship data, lateral distance data, and expected imaging altitude data. The baseline pitch angle data for each flight segment is generated based on the expected imaging altitude data, flight altitude data, and lateral distance data. Write the segment direction data, segment length data, adjacent segment turning angle data, segment compactness data, and reference attitude data including reference pitch angle and reference roll angle into the segment description table.

[0009] Optionally, based on the attitude angle time-series data and mapping image sequence of the multi-rotor UAV during the mapping operation, attitude change intensity data and image lateral offset sequence are generated respectively. Combined with flight segment compactness data, scene compactness data and corner lead data are generated, specifically including: The roll angle, pitch angle and yaw angle data are read according to the preset sampling period to form real-time attitude vector data, and the attitude increment data of the current sampling time relative to the previous sampling time is calculated. The attitude change intensity data at the current moment is calculated based on the attitude increment data at the current sampling moment and the attitude change intensity data at the previous moment. The texture change statistics of the mapping image corresponding to the current moment are performed according to the image column direction to form column energy distribution data, and normalized column energy data is calculated based on the column energy distribution data. Image column centroid data is calculated based on normalized column energy data, and image lateral offset data is calculated based on image column centroid data; Scene compactness data is calculated based on the current segment compactness data, the current attitude change intensity data, and the current image lateral offset data. The turn lead data is calculated based on the segment compactness data, the image lateral offset data, and the attitude change intensity data.

[0010] Optionally, based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time, a future time attitude prediction sequence and a reference attitude sequence are generated, specifically including: The scene-level data for the current moment is determined based on the scene compactness data and image lateral offset data corresponding to the current moment; Calculate the baseline attitude difference between adjacent segments based on the current segment's baseline attitude data and the next segment's baseline attitude data; For roll angle data, pitch angle data, and yaw angle data, future attitude recursion processing is performed based on scene-level data and baseline attitude difference data to obtain the future attitude prediction sequence. Calculate the transition mixing coefficient between the current segment reference attitude data and the next segment reference attitude data based on the remaining distance data from the current moment to the end of the current segment and the turning lead data; A reference attitude sequence is generated based on the transition mixing coefficient and the attitude prediction sequence at future time points.

[0011] Optionally, a reference attitude sequence is generated based on the transition mixing coefficient and the attitude prediction sequence at future time points, specifically including: The current scene compactness data is used to generate weight decay scale data, and the weight distribution data of the pose prediction sequence at future time is generated based on the weight decay scale data. Based on the current flight segment reference attitude data, the next flight segment reference attitude data, the transition mixing coefficient, the current real-time attitude data, and the prediction weight data, the attitude prediction sequence for future time moments is fused to obtain a reference attitude sequence.

[0012] Optionally, based on the reference attitude sequence, the current real-time attitude data, the image lateral offset sequence, and the flight segment compactness data, a shooting interval sequence is generated to classify the mapping image sequence into main mapping frames and transition frames, specifically including: The attitude consistency data at the current moment is calculated based on the reference attitude sequence and real-time attitude data; The shooting distance data at the current moment is generated based on the segment compactness data, image lateral offset data, and attitude consistency data; The current image is classified based on the attitude consistency data and the image lateral offset data. When the attitude consistency data is not greater than the first threshold and the absolute value of the image lateral offset data is not greater than the second threshold, the current image is written into the main mapping frame set; otherwise, the current image is written into the transition frame set. The main mapping frame index table is formed by establishing an index association between the shooting time data, reference posture data, real-time posture data, image lateral offset data, and shooting distance data corresponding to the main mapping frame set.

[0013] Optionally, update the reference attitude data for the next flight segment, specifically including: The average image lateral offset data and average attitude deviation data of the main mapping frame set of the current flight segment are statistically averaged to correct the reference attitude data of the next flight segment. The corrected reference attitude data for the next segment is written back to the segment description table and used as the data input for generating the reference attitude sequence for subsequent segments.

[0014] Optionally, a consistency check is performed after all flight segments are completed. If the check meets preset conditions, the surveying operation completion result is output, specifically including: After all flight segments are completed, the set of master mapping frames corresponding to each flight segment is extracted according to the master mapping frame index table, and the final inspection status data corresponding to each master mapping frame is collected; wherein, the final inspection status data includes at least the final inspection plane position deviation data, the final inspection attitude deviation data, and the final inspection image lateral offset data. For each flight segment, generate flight segment consistency evaluation data. When the consistency evaluation data of all flight segments meet the preset judgment conditions, output the surveying and mapping operation completion result; otherwise, output the verification mark result.

[0015] Optionally, the method further includes performing continuity constraint processing on the image lateral offset data and flight segment consistency evaluation data, specifically including: Image offset change rate data is generated based on image lateral offset data at adjacent sampling times, and segment connection change rate data is generated based on consistency evaluation data of adjacent segments. When the image offset change rate data is greater than the preset image change threshold, the image lateral offset data at the current moment is subjected to smooth replacement processing, and the smoothed replacement image lateral offset data is rewritten into the image lateral offset sequence; When the segment connection change rate data is greater than the preset segment change threshold, the reference attitude data of the next segment is corrected again, and the original reference attitude data of the next segment is replaced with the corrected reference attitude data of the next segment. Based on the updated image lateral offset sequence and the replaced reference attitude data for the next segment, the reference attitude sequence generation process and consistency determination process are re-executed.

[0016] Furthermore, to achieve the above objectives, the present invention also provides a multi-rotor unmanned aerial vehicle (UAV) mapping operation control system, comprising: The acquisition module is used to acquire the route point sequence corresponding to the surveying and mapping task, calculate the route direction data, route length data, adjacent route angle data and route compactness data, and generate a route description table. The generation module is used to generate attitude change intensity data and image lateral offset sequence based on the attitude angle time series data and mapping image sequence of multi-rotor UAV during the mapping operation, and to generate scene compactness data and corner lead data by combining the flight segment compactness data. The generation module is used to generate future attitude prediction sequences and reference attitude sequences based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time. The classification module is used to generate a shooting interval sequence based on the reference attitude sequence, the current real-time attitude data, the image lateral offset sequence, and the flight segment compactness data, to perform main mapping frame classification and transition frame classification on the mapping image sequence, to form flight segment statistical results, and to update the reference attitude data for the next flight segment. The output module is used to continuously perform control of subsequent segments based on the updated reference attitude data of the next segment, and to perform a consistency determination after all segments are completed. When the determination meets the preset conditions, the module outputs the mapping operation completion result.

[0017] The beneficial effects of this invention are as follows: It proposes a multi-rotor UAV mapping operation control method and system. By constructing a segment description table around the flight path point sequence, the geometric relationships, turning angle relationships, and mapping orientation relationships of each segment are unified and describable. Subsequently, during flight, the temporal changes in attitude angles and the lateral changes in the mapped image are extracted simultaneously, allowing the image results to enter the attitude control link. Then, a reference attitude sequence is generated based on the attitude change trends of the current segment, the next segment, and future times, and shooting interval data and image classification results are generated based on attitude consistency and image offset. Finally, the reference attitude data of the next segment is corrected based on the statistical results of the current segment, thereby achieving continuous control and result determination between multiple segments. Thus, this invention can solve the problem of maintaining image coverage continuity, lateral overlap stability, and imaging consistency between segments in narrow, continuous, and close-angle mapping scenarios. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the method of an embodiment of the present invention; Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] This invention provides a control method for surveying and mapping operations using a multi-rotor unmanned aerial vehicle (UAV), referring to... Figure 1 , Figure 1 This is a flowchart illustrating the multi-rotor UAV mapping operation control method according to an embodiment of the present invention.

[0021] In this embodiment, a multi-rotor UAV mapping operation control method includes the following steps: S1: Obtain the route point sequence corresponding to the surveying task, calculate the route direction data, route length data, adjacent route angle data and route compactness data, and generate a route description table.

[0022] Specifically, the spatial coordinate data of each route point is read, and two adjacent route points are combined to form a flight segment. Based on the spatial coordinate data of two adjacent route points, the flight segment direction data and flight segment length data of each flight segment are calculated. Based on the flight segment direction data of two adjacent flight segments, the turning angle data of adjacent flight segments are calculated. The flight segment compactness data is calculated by combining the current flight segment length data and the next flight segment length data. Based on the flight segment direction data, lateral mapping relationship data, lateral distance data, and expected imaging altitude data, the reference yaw angle data of each flight segment is generated. Based on the expected imaging altitude data, flight altitude data, and lateral distance data, the reference pitch angle data of each flight segment is generated. The flight segment direction data, flight segment length data, adjacent flight segment turning angle data, flight segment compactness data, and reference attitude data including the reference pitch angle and reference roll angle are written into the flight segment description table.

[0023] In this embodiment of the invention, the waypoint sequence is not simply a discrete set of positions used for navigation, but rather the fundamental source of all subsequent data processing in the entire control method. Specifically, this invention first converts the waypoint sequence into a flight segment description table, rather than directly entering the flight control processing.

[0024] In practical applications, all route points in the surveying task can be acquired first and arranged according to the task execution order. Two adjacent route points can be defined as a flight segment. To obtain the planar extension direction of the current flight segment, this invention first calculates the flight segment direction data based on the planar coordinate difference between two adjacent route points, as shown in the following formula: ; in, Indicates the first Segment direction data for each flight segment; The arctangent function, which represents quadrant information, is used to accurately distinguish the quadrant of a direction angle in a plane coordinate system. Indicates the first lateral coordinate data of each route point; Indicates the first Vertical coordinate data of each route point; Indicates the first lateral coordinate data of each route point; Indicates the first Vertical coordinate data of each route point; Indicates the flight segment number.

[0025] It should be noted that the segment orientation data is used to directly represent the horizontal orientation of the segment by utilizing the coordinate difference between adjacent route points. This process not only clarifies which direction the current segment should fly in, but also provides the foundational data for subsequently generating segment turning angle data, reference yaw angle data, and the directional transition relationship between the current segment and the next segment.

[0026] Furthermore, in order to measure the range of distances available for attitude adjustment in the current flight segment, this invention also needs to calculate the flight segment length data, the calculation formula of which is as follows: ; in, Indicates the first Segment length data for each flight segment; , , , The meaning is the same as above.

[0027] As is easily understood, segment length data reflects the actual distance between the start and end points of the current segment on a plane. In this invention, segment length data is used, on the one hand, to calculate the remaining distance of the current segment, and on the other hand, to form segment compactness data together with the turning angle data of adjacent segments. That is, segment length data determines how much space the aircraft has to complete attitude changes.

[0028] After obtaining the directional data for each flight segment, this invention further calculates the directional change between adjacent flight segments to form the turning angle data for adjacent flight segments. The calculation formula is as follows: ; in, Indicates the first The first segment and the first Turning angle data between adjacent flight segments; This represents the angle normalization processing function, used to normalize the angle difference to a preset range; Indicates the first Segment direction data for each flight segment; Indicates the first Segment direction data for each flight segment.

[0029] It should be noted that the adjacent segment turning angle data is used to characterize the magnitude of the directional change that the aircraft needs to complete when entering the next segment after the current segment ends. If only the magnitude of the directional change is known, without considering the lengths of the preceding and following segments, it is still impossible to determine whether the scenario is truly complex. Therefore, this invention further constructs segment compactness data based on this, and its calculation formula is as follows: ; in, Indicates the first Segment compactness data for each flight segment; Indicates the first The first segment and the first Turning angle data between adjacent flight segments; Indicates the first Segment length data for each flight segment; Indicates the first Segment length data for each flight segment; This represents a positive correction term to prevent the denominator from being zero.

[0030] As is easily understood, the above segment compactness data is used to characterize the extent of directional switching required within a limited distance by using the ratio of the turning angle to the total length of the preceding and following segments. When the turning angle is larger and the preceding and following segments are shorter, the segment compactness data is larger, indicating that the current scene is less conducive to achieving stable imaging using conventional fixed attitude methods.

[0031] After completing the calculation of the geometric relationships of each flight segment, this invention further requires generating reference attitude data by combining the lateral mapping relationships of each flight segment. It should be noted that the reference attitude data in this invention does not simply set the aircraft orientation to the flight segment direction, but is generated comprehensively based on the left-right relationship of the measured elevation, lateral mapping requirements, lateral distance, and desired imaging height. The purpose is to ensure that each flight segment has a basic attitude reference for the mapped object before entering subsequent complex control, rather than just for path navigation itself.

[0032] In one executable implementation, the baseline yaw angle data can be generated as follows: ; in, Indicates the first Reference yaw angle data for each flight segment; Indicates the first Segment direction data for each flight segment; Indicates the first Lateral mapping relationship data for each flight segment, used to indicate whether the measured elevation is located to the left or right of the flight direction; This indicates the preset side view data.

[0033] It should be noted that the technical principle of the baseline yaw angle data lies in superimposing a side view related to the left-right relationship of the elevation onto the flight segment direction data. This ensures that the aircraft's yaw direction no longer simply follows the flight segment direction, but rather incorporates a stable lateral observation component. Therefore, this invention enables the camera's field of view to more stably cover the measured elevation, rather than leaving the elevation permanently in the image's edge region.

[0034] Furthermore, the reference pitch angle data can be generated as follows: ; in, Indicates the first Reference pitch angle data for each flight segment; Represents the arctangent function; Indicates the first Expected imaging altitude data for each flight segment; Indicates the first Flight altitude data for each flight segment; Indicates the first Lateral distance data for each flight segment; This represents a positive correction term to prevent the denominator from being zero.

[0035] It should be noted that the technical principle behind the reference pitch angle data lies in determining the aircraft's pitch direction using the geometric relationship between height difference and lateral distance, making it easier for the camera to align vertically with the desired imaging area of ​​the measured facade. In this technical solution, this formula serves to provide an initial pitch reference consistent with the facade height requirements for subsequent reference attitude generation.

[0036] In this embodiment of the invention, the reference roll angle data can be set in the following form: ; in, Indicates the first The baseline roll angle data for each flight segment.

[0037] It should be noted that setting the initial reference roll angle data to zero does not mean that the invention does not process the roll direction, but rather that under ideal static geometry, the aircraft does not need to pre-apply roll. The roll correction truly related to image lateral deviation will be completed by the fusion process of direction response data and reference attitude in subsequent steps. After completing the above processing, the segment direction data, segment length data, adjacent segment turn angle data, segment compactness data, and reference attitude data can be uniformly written into the segment description table as the data basis for subsequent steps.

[0038] S2: Based on the attitude angle time series data and mapping image sequence of the multi-rotor UAV during the mapping operation, attitude change intensity data and image lateral offset sequence are generated respectively. Combined with the flight segment compactness data, scene compactness data and turning lead data are generated.

[0039] Specifically, roll angle data, pitch angle data, and yaw angle data are read according to a preset sampling period to form real-time attitude vector data, and the attitude increment data of the current sampling time relative to the previous sampling time is calculated; the attitude change intensity data of the current time is calculated based on the attitude increment data of the current sampling time and the attitude change intensity data of the previous time; texture change statistics are performed on the mapping image corresponding to the current time according to the image column direction to form column energy distribution data, and normalized column energy data is calculated based on the column energy distribution data; image column centroid data is calculated based on the normalized column energy data, and image lateral offset data is calculated based on the image column centroid data; scene compactness data is calculated based on the segment compactness data of the current flight segment, the attitude change intensity data of the current time, and the image lateral offset data of the current time; and turn lead data is calculated based on the segment compactness data, the image lateral offset data, and the attitude change intensity data.

[0040] In this embodiment of the invention, considering that the problem in continuous turning scenarios is not simply whether the current attitude value deviates, but whether the current attitude is in a process of accelerating or continuously accumulating change, the present invention does not directly use a certain instantaneous attitude value as the sole basis for subsequent control. Instead, it first extracts attitude change intensity data through attitude angle time series data to reflect the overall change trend of the body in a short period of time.

[0041] In practical applications, roll angle, pitch angle, and yaw angle data can be continuously read according to a preset sampling period and compared with the corresponding data from the previous moment to form attitude increment data. To avoid the system becoming overly sensitive to single-frame changes, this invention further adopts a weighted combination of the current increment and historical intensity to generate attitude change intensity data, the calculation formula of which is as follows: ; in, This represents the intensity of attitude change at the current moment; This represents the intensity of the attitude change at the previous moment; Indicates the smoothing coefficient; This represents the current roll angle increment data; This represents the pitch angle increment data at the current moment; This represents the incremental yaw angle data at the current moment; Indicates the sampling time.

[0042] It should be noted that the calculation of attitude change intensity data retains both the real-time information of the attitude change at the current moment and the cumulative characteristics of the attitude change over a recent period, thereby generating a data volume that better represents the overall trend of change. The purpose of this data is that if the attitude change intensity data begins to increase significantly before the UAV has reached its actual turning position, the system can identify that the current state is transitioning from a normal state to a sensitive state.

[0043] On the other hand, this invention also directly performs image lateral offset extraction processing on the surveyed image sequence. It should be noted that this invention does not rely on complex external equipment to determine whether the measured facade is biased to one side of the image; instead, it utilizes the texture variation information of the image itself to calculate the degree of bias of the facade in the image. This approach is adopted because the edges of doors and windows, wall outlines, and texture structures of building facades typically produce a clear column-direction energy distribution in the image, which can effectively reflect whether the main area is located on the left, middle, or right side of the image.

[0044] In one executable implementation, the column energy distribution data of the current image is first calculated column by column, as follows: ; in, Indicates the current time of the image. Column energy data; This indicates that the image at the current time is at column coordinates. Row coordinates are Brightness data at the location; Represents image height data; Represents the column coordinates of the image; Represents the row coordinates of the image.

[0045] It should be noted that the column energy distribution data is obtained by summing the absolute values ​​of the brightness differences between adjacent pixels along the image height direction to obtain the intensity of texture variations contained in each column. If a column contains more facade edges, window frame lines, or brick joints, the energy of that column is usually higher. The purpose of this data is to transform the original image into a data foundation that can characterize the lateral distribution of the main subject area.

[0046] After obtaining the column energy distribution data, the present invention further normalizes it, and the calculation formula is as follows: ; in, Indicates the current time of the image. Normalized column energy data; Indicates the current time of the image. Column energy data; The variable representing the summation of column coordinates in the image; This represents the image width data; This represents a positive correction term to prevent the denominator from being zero.

[0047] It should be noted that the purpose of normalization is to eliminate the influence of differences in overall texture strength between different images on subsequent calculations, so that the column energy distribution of each frame can be compared on a uniform scale. In other words, the above formula is not simply numerical normalization, but rather provides a basis for subsequent calculations of image column centroid data that is comparable in total amount and interpretable in distribution.

[0048] Based on this, the present invention continues to calculate the column centroid data of the current image, and the calculation formula is as follows: ; in, This represents the column centroid data of the image at the current moment; Represents the column coordinates of the image; Indicates the current time of the image. Normalized column energy data; This represents the image width data.

[0049] It should be noted that the technical principle of column centroid data is similar to that of weighted centroid calculation, that is, using the normalized energy proportion of each column as a weight to obtain the lateral position of the main texture distribution in the current image. This column centroid data does not directly represent the position of the facade in physical space, but rather the center position of the main texture distribution within the image plane. It is an important intermediate quantity for quantifying the current imaging state from the perspective of image results.

[0050] Subsequently, this invention generates image lateral offset data based on image column centroid data, and the calculation formula is as follows: ; in, This represents the lateral offset data of the image at the current moment; This represents the column centroid data of the image at the current moment; This represents the image width data.

[0051] It should be noted that image lateral offset data normalizes the deviation of image column centroid data from the image center to a uniform scale, allowing the left- or right-biased orientation of the image subject, and the degree of deviation, to be characterized by a signed continuous data set. This data directly transforms the question of whether the mapping result has been skewed into a controllable amount of data, enabling subsequent attitude processing to respond directly to the imaging results, rather than solely based on flight status.

[0052] Furthermore, in this invention, in order to determine the sensitivity of the current operating scenario to the control system, the segment compactness data of the current flight segment, the attitude change intensity data at the current moment, and the image lateral offset data at the current moment are combined and processed to form scene compactness data, the calculation formula of which is as follows: ; in, This represents the scene compactness data at the current moment; , , Indicates the weighting coefficient; This indicates the segment compactness data for the current flight segment; This represents the intensity of attitude change at the current moment; This represents the lateral offset data of the image at the current moment.

[0053] It should be noted that scene compactness data compresses three types of information—the geometric compactness of the flight segment, the current trend of aircraft changes, and the degree of current image lateral deviation—into a unified scene evaluation metric. If only one type of data increases, it does not necessarily mean that the current scene has entered a highly sensitive state; however, when all three types of data increase together, the system can reliably identify that the current scene is in a state with higher requirements for attitude control and image continuity.

[0054] After generating scene compactness data, this invention further generates corner lead data, the calculation formula of which is as follows: ; in, This represents the lead time for turning at the current moment; Indicates the baseline lead time data; , , Indicates the weighting coefficient; This indicates the segment compactness data for the current flight segment; This represents the lateral offset data of the image at the current moment; This represents the intensity of attitude change at the current moment.

[0055] It should be noted that the angle lead data is not a simple fixed constant, but varies with the complexity of the current scenario. When the flight segment is more compact, the image is more lateral, and the attitude change is more significant, the system needs to start transitioning the reference attitude from the current flight segment to the next flight segment earlier. Therefore, the angle lead data will increase accordingly, and this data will directly participate in the reference attitude generation in the next step.

[0056] S3: Based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time, generate the attitude prediction sequence and reference attitude sequence for future time.

[0057] Specifically, the scene-level data for the current moment is determined based on the scene compactness data and image lateral offset data corresponding to the current moment; the reference attitude difference data between adjacent segments is calculated based on the reference attitude data of the current segment and the reference attitude data of the next segment; for roll angle data, pitch angle data, and yaw angle data, future moment attitude recursion processing is performed based on the scene-level data and the reference attitude difference data to obtain the future moment attitude prediction sequence; the transition mixing coefficient between the current segment reference attitude data and the next segment reference attitude data is calculated based on the remaining distance data from the current moment to the end of the current segment and the turning lead data; and a reference attitude sequence is generated based on the transition mixing coefficient and the future moment attitude prediction sequence.

[0058] Furthermore, weight decay scale data is generated based on the scene compactness data at the current moment, and weight distribution data for the attitude prediction sequence at future moments is generated based on the weight decay scale data. Based on the current flight segment reference attitude data, the next flight segment reference attitude data, the transition mixing coefficient, the current real-time attitude data, and the prediction weight data, the attitude prediction sequence at future moments is fused to obtain a reference attitude sequence.

[0059] In this embodiment of the invention, the control challenge in continuous turning scenarios lies in the fact that the system cannot wait until it actually reaches the turning point before initiating significant attitude changes, nor can it prematurely switch to the attitude of the next flight segment while still far from the turning point. Therefore, this invention does not directly generate a single-moment control variable, but instead first generates a sequence of predicted attitudes for future moments, and then combines this with the baseline attitude relationship between the current flight segment and the next flight segment to form a reference attitude sequence. In this way, the system can incorporate the state it will enter in the future into the current moment, thereby improving transition continuity.

[0060] In practical applications, the scene-level data for the current moment can be determined first based on the scene compactness data and image lateral offset data. This scene-level data is used to distinguish whether the current control is in a normal state, a transitional state, or an enhanced state. Although the specific rules for this division can be implemented in different ways, their common technical purpose is to call different parameter sets for attitude recursion processing in different states, thereby enabling the system to adapt to different scenes.

[0061] Furthermore, it is also necessary to generate reference attitude difference data between adjacent segments based on the reference attitude data of the current segment and the reference attitude data of the next segment. It should be noted that the reference attitude difference data is used to characterize how much the ideal attitude requirement has changed when entering the next segment after the current segment ends. Its role in this invention is to incorporate the requirements of the next segment into the attitude processing process at the current moment in advance, rather than allowing the system to only make local corrections based on the current historical attitude.

[0062] Furthermore, in one executable implementation, future attitude recursion processing is performed on the roll angle data, pitch angle data, and yaw angle data respectively to obtain the future attitude prediction sequence, the recursive formula of which is as follows: ; in, This represents the attitude prediction data for the next moment; This represents any one of the roll angle, pitch angle, or yaw angle data; Indicates the order of revisiting; Indicates the order of recurrence; , , , , , This represents the recursive coefficients related to scene-level data; This represents the scene-level data at the current moment. This represents the scene compactness data at the current moment; This represents the lateral offset data of the image at the current moment; This represents the baseline attitude difference data between adjacent flight segments; This represents the attitude increment data at the current moment; This represents the directional response data generated from the image lateral offset data.

[0063] It should be noted that the technical principle of the future attitude prediction sequence lies in unifying three types of information from different sources into the attitude recursive processing. The first type is historical attitude time-series information, i.e., attitude values ​​from several past moments, used to reflect the current motion trend of the aircraft. The second type is current scene state and image lateral information, i.e., scene compactness data and image lateral offset data, used to indicate whether the current application scenario is special and whether the imaging results have deviated. The third type is flight segment transition information, i.e., the baseline attitude difference data between the current flight segment and the next flight segment, used to reflect the impact of future flight segments on current control. Through the above unified recursion, this invention can generate a future attitude prediction sequence more suitable for the current mapping scenario. This sequence can provide forward-looking correction information for subsequent reference attitude fusion.

[0064] After obtaining the attitude prediction sequence for future moments, this invention also needs to determine the extent to which the current segment's reference attitude should be retained at the current moment, and the extent to which it should transition to the reference attitude for the next segment. To this end, this invention constructs a transition mixing coefficient using the relationship between the remaining distance data from the current moment to the end of the current segment and the turning lead data. The calculation formula is as follows: ; in, Indicates the transition mixing coefficient; Represents an exponential function; Indicates the slope coefficient; This represents the lead time for turning at the current moment; This represents the remaining distance from the current moment to the end of the current flight segment.

[0065] It's important to note that the technical principle behind the transition blending coefficient lies in using a smooth curve rather than a hard handover to complete the attitude transition for each flight segment. When the remaining distance for the current segment is still relatively large, the transition blending coefficient is small, indicating that the current reference attitude is mainly determined by the baseline attitude for the current segment. As the remaining distance gradually decreases and approaches the turn lead data, the transition blending coefficient increases rapidly, indicating that the current reference attitude needs to move more significantly closer to the baseline attitude for the next segment. This transforms the timing and extent of the transition into a continuously changing data set, avoiding the imaging instability issues caused by sudden attitude changes near a waypoint in traditional methods.

[0066] To further enable image lateral offset to directly influence the attitude recursion direction, this invention also generates directional response data based on the image lateral offset data. Specifically, the roll direction response data is in signed square root form, while the pitch and yaw direction response data are in linear form. Their generation formulas are as follows: ; in, This indicates the roll direction response data; This represents the sign determination function; This represents the lateral offset data of the image at the current moment.

[0067] It should be noted that the roll direction response data is as follows: when the image offset is small, the square root form can provide high sensitivity, allowing the system to start correcting in the early offset stage; when the image offset is large, the square root growth rate is lower than the linear growth rate, which helps to avoid excessive roll correction.

[0068] Correspondingly, the pitch response data can be expressed as: ; in, This represents the pitch response data; This represents the lateral offset data of the image at the current moment.

[0069] Yaw response data can be expressed as: ; in, This indicates the yaw direction response data; This represents the lateral offset data of the image at the current moment.

[0070] It should be noted that the pitch and yaw response data are presented in a linear form, primarily to maintain a direct correspondence with the image offset direction, ensuring that the correction direction is applied in the direction the image shifts. These three types of directional response data collectively participate in the subsequent fusion of the reference attitude.

[0071] Furthermore, in order to control the influence of different prediction steps in the future pose prediction sequence on the current reference pose, this invention also generates weight decay scale data based on the scene compactness data at the current moment, and its calculation formula is as follows: ; in, This represents the weight decay scale data at the current moment; Represents basic scale data; This represents the scene compactness data at the current moment.

[0072] It should be noted that the calculation of the weight decay scale data is used to characterize the fact that when the current scene is more compact, the system needs to pay more attention to the attitude changes in the near future. Therefore, the weight decay scale data will be reduced accordingly, so that the predictions closer to the current moment will receive greater weight. When the scene is relatively stable, the weight decay scale data is larger, so that the influence of multiple future moment predictions is more balanced.

[0073] After obtaining the weight decay scale data, this invention further calculates the weight distribution data for each future prediction time, as shown in the following formula: ; in, Indicates the future number Predicted weight data corresponding to each time point; Indicates a future time number; This represents the weight decay scale data at the current moment; Represents the variable to be summed; This represents the number of prediction steps.

[0074] It should be noted that the weight distribution data enables the data from multiple future prediction times to form a normalized and comparable weight sequence, thereby ensuring that different prediction steps can participate in the reference attitude fusion in the expected proportion, without causing imbalance in dimensions or total amount due to different prediction steps.

[0075] Based on this, the present invention finally generates the reference attitude data at the current moment, and its fusion formula is as follows: ; in, Represents the reference attitude data at the current moment; Indicates the transition mixing coefficient; This represents the reference attitude data for the current flight segment; This indicates the reference attitude data for the next flight segment; Indicates the future number Predicted weight data corresponding to each time point; Indicates the future number Attitude prediction data at each time step; Represents the real-time attitude data at the current moment; This represents the number of prediction steps.

[0076] It should be noted that the reference attitude data is used to unify and integrate the ideal requirements of the current flight segment, the ideal requirements of the next flight segment, and future attitude change trends. This generates reference attitude data that neither deviates from the current actual attitude state nor fails to anticipate the imaging requirements of subsequent flight segments. Subsequent shooting control and image classification will be based on this reference attitude data.

[0077] S4: Based on the reference attitude sequence, the current real-time attitude data, the image lateral offset sequence, and the flight segment compactness data, generate a shooting interval sequence, perform main mapping frame classification and transition frame classification on the mapping image sequence, form flight segment statistical results, and update the reference attitude data for the next flight segment.

[0078] Specifically, the attitude consistency data at the current moment is calculated based on the reference attitude sequence and real-time attitude data; the shooting distance data at the current moment is generated based on the segment compactness data, image lateral offset data, and attitude consistency data; the current image is classified based on the attitude consistency data and image lateral offset data, and when the attitude consistency data is not greater than a first threshold and the absolute value of the image lateral offset data is not greater than a second threshold, the current image is written into the main mapping frame set; otherwise, the current image is written into the transition frame set; an index is established to associate the shooting time data, reference attitude data, real-time attitude data, image lateral offset data, and shooting distance data corresponding to the main mapping frame set, forming a main mapping frame index table.

[0079] Based on this, the average image lateral offset data and average attitude deviation data are statistically analyzed for the main mapping frame set of the current flight segment, and the reference attitude data for the next flight segment is corrected. The corrected reference attitude data for the next flight segment is written back to the flight segment description table and used as the data input for generating the reference attitude sequence for subsequent flight segments.

[0080] In this embodiment of the invention, the generated reference attitude data is not only used for flight control execution, but also needs to be directly linked to the image acquisition process. This is because, in continuous cornering scenarios, if the acquisition rhythm still uses a fixed interval method, even if the attitude has been improved to some extent, uneven local coverage may occur due to insufficient image sampling in key areas. Therefore, this invention dynamically generates shooting interval data based on the current attitude execution status and the current image eccentricity, and further classifies the images into main mapping frames and transition frames.

[0081] In practical applications, attitude consistency data can be generated first based on the difference between the current reference attitude data and the real-time attitude data. The calculation formula is as follows: ; in, This represents the attitude consistency data at the current moment; This represents the roll angle component in the reference attitude data at the current moment; This represents the pitch angle component in the reference attitude data at the current moment; This represents the yaw angle component in the reference attitude data at the current moment; This represents the roll angle data in the real-time attitude data at the current moment; This represents the pitch angle data in the real-time attitude data at the current moment; This represents the yaw angle data in the real-time attitude data at the current moment.

[0082] It should be noted that attitude consistency data is used to compress the three-axis difference between the current real-time attitude and the ideal reference attitude into a continuously variable evaluation quantity. This data is not equivalent to the control deviation within the flight controller, but rather is used to characterize whether the current image acquisition moment has approached the attitude state desired by this invention. This establishes a relationship between attitude execution performance and the shooting strategy.

[0083] After obtaining attitude consistency data, this invention further combines flight segment compactness data and image lateral offset data to generate shooting distance data, the calculation formula of which is as follows: ; in, This represents the shooting distance data at the current moment; This represents the basic shooting distance data; , , Indicates the weighting coefficient; This indicates the segment compactness data for the current flight segment; This represents the lateral offset data of the image at the current moment; This represents the attitude consistency data at the current moment; This represents the minimum shooting distance data; This represents the maximum shooting distance data; This represents the amplitude limiting function.

[0084] It should be noted that the shooting interval data is used to characterize the following: when the current flight segment geometry is more compact, the current image is more skewed to the side, and the current attitude consistency is worse, the system needs to increase the sampling density of key areas, thus reducing the shooting interval data; when the above factors are relatively stable, the shooting interval data will return to a larger normal value. This coordinates the shooting rhythm with the attitude control results, thereby improving the image coverage sufficiency of corner-sensitive areas.

[0085] Furthermore, in this invention, images are not uniformly treated as data of the same importance level for subsequent processing. Instead, they are classified according to the current attitude consistency data and the image lateral offset data. Specifically, when the attitude consistency data is not greater than a preset first threshold and the absolute value of the image lateral offset data is not greater than a preset second threshold, it indicates that the body attitude at the corresponding moment of the current image is relatively close to the reference attitude, and the lateral offset of the image body is small. Therefore, the image can be written into the main mapping frame set; otherwise, the image is written into the transition frame set.

[0086] It should be noted that the division between the main mapping frame and the transition frame does not mean that the transition frame is invalid. On the contrary, the transition frame can still be used to improve the image coverage redundancy in key areas. However, in subsequent flight segment statistics and final inspection, the main mapping frame is more reflective of the overall quality level of the mapping operation. Through this classification process, the present invention retains the necessary high-density sampling in complex scenes while avoiding the indiscriminate inclusion of all transition images into the main evaluation sequence.

[0087] It should be noted that after forming the master mapping frame set, the corresponding shooting time data, reference attitude data, real-time attitude data, image lateral offset data, and shooting distance data are indexed and associated to form a master mapping frame index table. This provides a traceable data mapping for subsequent flight segment statistics and consistency determination, ensuring that each master mapping image corresponds to its attitude and shooting control state at the time of acquisition.

[0088] Based on this, the present invention further calculates the average image lateral offset data for the current flight segment, and the calculation formula is as follows: ; in, This represents the average image lateral offset data for the current flight segment; Indicates the number of main mapping frames for the current flight segment; This represents the set of main mapping frames for the current flight segment; This represents the lateral offset data of the image at the current moment.

[0089] It should be noted that the average image lateral offset data is used to obtain the overall lateral offset trend of the current flight segment at the image level by averaging the main mapping frame set, rather than being influenced by local anomalies in a single instantaneous frame. This provides segment-level feedback at the image result level for the baseline attitude correction of the next flight segment.

[0090] Similarly, the present invention also statistically analyzes the average attitude deviation data for the current flight segment, and the calculation formula is as follows: ; in, This represents the average attitude deviation data for the current flight segment; Indicates the number of main mapping frames for the current flight segment; This represents the set of main mapping frames for the current flight segment; Represents the reference attitude data at the current moment; This represents the real-time attitude data at the current moment.

[0091] It should be noted that the average attitude deviation data is used to characterize the overall degree to which the aircraft follows the reference attitude at the current flight segment scale. If the aircraft exhibits an average deviation in a certain direction for an extended period during the current flight segment, similar problems may recur if the original reference attitude is still used in subsequent flight segments. Therefore, this invention uses the aforementioned average image lateral offset data and average attitude deviation data together to correct the reference attitude data for the next flight segment, and the correction formula is as follows: ; in, This indicates the corrected reference attitude data for the next flight segment; This indicates the reference attitude data for the next flight segment before correction; and This indicates that the weighted data has been adjusted. This represents the average attitude deviation data for the current flight segment; This represents the average image lateral offset data for the current flight segment; This represents the sign determination function.

[0092] It should be noted that the reference attitude data for the next flight segment transforms the systematic attitude deviation trend and systematic image lateral trend already exhibited in the current flight segment into pre-correction values ​​for the reference attitude data of the next flight segment. This establishes a continuous data transfer relationship between adjacent flight segments, making the entire multi-segment operation process no longer an independent segmented control, but a continuous control process with segment-level experience inheritance characteristics.

[0093] S5: Based on the updated reference attitude data for the next segment, continue to perform control of the subsequent segments, and perform consistency determination after all segments are completed. When the determination meets the preset conditions, output the mapping operation completion result.

[0094] Specifically, after all flight segments are completed, the set of master mapping frames corresponding to each flight segment is extracted according to the master mapping frame index table, and the final inspection status data corresponding to each master mapping frame is collected; the final inspection status data includes at least the final inspection plane position deviation data, the final inspection attitude deviation data, and the final inspection image lateral offset data; for each flight segment, flight segment consistency evaluation data is generated. When the consistency evaluation data of all flight segments meet the preset judgment conditions, the surveying and mapping operation completion result is output; otherwise, the verification mark result is output.

[0095] In a preferred embodiment, the method further includes performing continuity constraint processing on the image lateral offset data and the segment consistency evaluation data, specifically including: generating image offset change rate data based on the image lateral offset data at adjacent sampling times, and generating segment connection change rate data based on the consistency evaluation data of adjacent segments; when the image offset change rate data is greater than a preset image change threshold, performing smooth replacement processing on the image lateral offset data at the current time, and rewriting the smoothed replacement image lateral offset data into the image lateral offset sequence; when the segment connection change rate data is greater than a preset segment change threshold, performing a second correction processing on the reference attitude data of the next segment, and replacing the original reference attitude data of the next segment with the second corrected reference attitude data of the next segment; and re-performing the reference attitude sequence generation processing and consistency determination processing based on the updated image lateral offset sequence and the replaced reference attitude data of the next segment.

[0096] In this embodiment of the invention, the completion of flight control is not directly considered as the completion of mapping. This is because the present invention aims to ensure the consistency of the entire multi-segment mapping results in terms of image coverage, attitude execution, and segment connections. Therefore, after all segments are completed, a consistency determination is required based on the final inspection status data corresponding to the main mapping frame and the segment description table to confirm whether the entire mapping operation has met the preset requirements.

[0097] In practical applications, the set of master mapping frames corresponding to each flight segment can be extracted according to the master mapping frame index table, and the final inspection status data corresponding to each flight segment can be collected. The final inspection status data includes at least final inspection plane position deviation data, final inspection attitude deviation data, and final inspection image lateral offset data. It should be noted that the method of obtaining the final inspection status data is not limited to a single implementation path; it can come from flight records, control records, image statistical results, or a combination of the above data, as long as it can reflect the final execution status of the flight segment.

[0098] To provide a unified evaluation of the final state of each flight segment, this invention constructs flight segment consistency evaluation data, the calculation formula of which is as follows: ; in, Indicates the first Consistency evaluation data for each flight segment; , , This represents the evaluation weight data; Indicates the first Statistical data on the final inspection plane position deviation of each flight segment; Indicates the first Planar position reference data for each flight segment; Indicates the first Statistical data on attitude deviation at the final inspection of each flight segment; Indicates the first Attitude reference data for each flight segment; Indicates the first Statistical data on lateral offset of final inspection images for each flight segment; Indicates the first Image offset reference data for each flight segment; This represents a positive correction term to prevent the denominator from being zero.

[0099] It should be noted that the consistency evaluation data is constructed by using three dimensions—positional deviation, attitude deviation, and image offset—to create a unified, normalized evaluation metric, thereby avoiding misjudgments caused by a single evaluation indicator. For example, considering only flight positional deviation may not reflect the problem of subject slant in the image; considering only image offset may not reflect the persistence of attitude execution deviation. Therefore, the joint evaluation of these three types of statistics is more suitable for the surveying and mapping operation control scenario targeted by this invention.

[0100] In this invention, when the consistency evaluation data for all flight segments meets the preset judgment conditions, the mapping operation completion result can be output; otherwise, the verification mark result can be output. Furthermore, in one executable implementation, continuity constraint processing can be performed on the image lateral offset data and the flight segment consistency evaluation data. For example, between adjacent sampling times, when the image lateral offset data changes too rapidly and exceeds a preset threshold, the current image lateral offset data can be smoothly replaced to reduce the impact of occasional abnormal frames on the overall judgment; between adjacent flight segments, when the consistency evaluation change rate increases abnormally, the reference attitude data for the next flight segment can be corrected again. Through the above additional processing, the stability and robustness of the entire control method in complex scenarios can be further enhanced.

[0101] To further illustrate the operation of this invention, a typical application scenario is described below. The surveying object is defined as the exterior facade of a campus teaching building and the corner area of ​​the connecting corridor. A multi-rotor UAV flies close to the facade, passing two corners consecutively within a short distance. First, the system generates a flight segment description table based on the flight path point sequence, thereby identifying which flight segments are shorter and have larger adjacent corners, giving these segments higher flight segment compactness data. Subsequently, during flight and data acquisition, the system continuously extracts attitude change intensity data and image lateral offset data. As the UAV gradually approaches the first corner, if the image offset increases and the attitude change intensity increases simultaneously, the scene compactness data and corner lead data will increase accordingly. At this time, the reference generation module advances the current reference attitude data to meet the attitude requirements of the next flight segment, rather than waiting until the corner is actually reached before abruptly changing. The shooting control module automatically reduces the shooting interval based on the attitude consistency data and image offset data, and writes the more stable image into the main survey frame set. After the first leg is completed, the leg correction module uses the average image offset data and average attitude deviation data from the main mapping frame set of that leg to correct the reference attitude data for the next leg, thus making the attitude control in the second turning area more in line with the actual imaging requirements on site. Finally, after all legs are completed, the judgment output module uses consistency evaluation data to make a unified judgment on the entire set of mapping results, and outputs the final mapping operation completion result.

[0102] It should be noted that the above examples are only used to assist in understanding the data processing logic of this invention and do not constitute a limitation on the applicable objects of this invention. For other scenarios with continuous turning angles, short flight segments, and close proximity to surveying and mapping characteristics, the corresponding technical effects can be obtained as long as the data link and processing method described in this invention are adopted.

[0103] Reference Figure 2 , Figure 2 This is a schematic diagram of the structure of the multi-rotor UAV mapping operation control system according to an embodiment of the present invention.

[0104] like Figure 2 As shown, the multi-rotor UAV mapping operation control system proposed in this embodiment of the invention includes: The acquisition module 10 is used to acquire the route point sequence corresponding to the surveying and mapping task, calculate the route direction data, route length data, adjacent route angle data and route compactness data, and generate a route description table. The forming module 20 is used to generate attitude change intensity data and image lateral offset sequence based on the attitude angle time series data and mapping image sequence of the multi-rotor UAV during the mapping operation, and generate scene compactness data and turning lead data by combining the flight segment compactness data. The generation module 30 is used to generate a future attitude prediction sequence and a reference attitude sequence based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time. The classification module 40 is used to generate a shooting interval sequence based on the reference attitude sequence, the current real-time attitude data, the image lateral offset sequence and the flight segment compactness data, perform main mapping frame classification and transition frame classification on the mapping image sequence, form flight segment statistical results and update the reference attitude data for the next flight segment; The output module 50 is used to continuously perform subsequent segment control based on the updated reference attitude data of the next segment, and to perform consistency determination after all segments are completed. When the determination meets the preset conditions, the output module 50 outputs the mapping operation completion result.

[0105] Other embodiments or specific implementations of the multi-rotor UAV mapping operation control system of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0106] It is understood that in the description of this specification, references to terms such as "one embodiment," "another embodiment," "other embodiments," or "first embodiment to Nth embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0107] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0108] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A control method for surveying and mapping operations using a multi-rotor unmanned aerial vehicle (UAV), characterized in that, The method includes the following steps: Obtain the route point sequence corresponding to the surveying task, calculate the route direction data, route length data, adjacent route angle data and route compactness data, and generate a route description table; Based on the attitude angle time series data and mapping image sequence of multi-rotor UAVs during the mapping operation, attitude change intensity data and image lateral offset sequence are generated respectively. Combined with flight segment compactness data, scene compactness data and corner lead data are generated. Based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time, generate the attitude prediction sequence and reference attitude sequence for future time. Based on the reference attitude sequence, the current real-time attitude data, the image lateral offset sequence, and the flight segment compactness data, a shooting interval sequence is generated. The mapping image sequence is classified into main mapping frames and transition frames to form flight segment statistical results and update the reference attitude data for the next flight segment. Based on the updated reference attitude data for the next segment, control of the subsequent segments is continuously executed, and a consistency determination is performed after all segments are completed. When the determination meets the preset conditions, the mapping operation completion result is output.

2. The multi-rotor UAV mapping operation control method as described in claim 1, characterized in that, Obtain the route point sequence corresponding to the surveying task, calculate the route direction data, route length data, adjacent route angle data, and route compactness data, and generate a route description table, specifically including: Read the spatial coordinate data of each route point, form a segment by combining two adjacent route points, and calculate the segment direction data and segment length data of each segment based on the spatial coordinate data of two adjacent route points. The turning angle data of adjacent segments is calculated based on the segment direction data of two adjacent segments, and the segment compactness data is calculated by combining the current segment length data and the next segment length data. The baseline yaw angle data for each flight segment is generated based on the flight segment direction data, lateral mapping relationship data, lateral distance data, and expected imaging altitude data. The baseline pitch angle data for each flight segment is generated based on the expected imaging altitude data, flight altitude data, and lateral distance data. Write the segment direction data, segment length data, adjacent segment turning angle data, segment compactness data, and reference attitude data including reference pitch angle and reference roll angle into the segment description table.

3. The multi-rotor UAV mapping operation control method as described in claim 1, characterized in that, Based on the attitude angle time-series data and mapping image sequence of multi-rotor UAVs during surveying operations, attitude change intensity data and image lateral offset sequence are generated respectively. Combined with flight segment compactness data, scene compactness data and corner lead data are generated, specifically including: The roll angle, pitch angle and yaw angle data are read according to the preset sampling period to form real-time attitude vector data, and the attitude increment data of the current sampling time relative to the previous sampling time is calculated. The attitude change intensity data at the current moment is calculated based on the attitude increment data at the current sampling moment and the attitude change intensity data at the previous moment. The texture change statistics of the mapping image corresponding to the current moment are performed according to the image column direction to form column energy distribution data, and normalized column energy data is calculated based on the column energy distribution data. Image column centroid data is calculated based on normalized column energy data, and image lateral offset data is calculated based on image column centroid data; Scene compactness data is calculated based on the current segment compactness data, the current attitude change intensity data, and the current image lateral offset data. The turn lead data is calculated based on the segment compactness data, the image lateral offset data, and the attitude change intensity data.

4. The multi-rotor UAV mapping operation control method as described in claim 1, characterized in that, Based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time, a future time attitude prediction sequence and a reference attitude sequence are generated, specifically including: The scene-level data for the current moment is determined based on the scene compactness data and image lateral offset data corresponding to the current moment; Calculate the baseline attitude difference between adjacent segments based on the current segment's baseline attitude data and the next segment's baseline attitude data; For roll angle data, pitch angle data, and yaw angle data, future attitude recursion processing is performed based on scene-level data and baseline attitude difference data to obtain the future attitude prediction sequence. Calculate the transition mixing coefficient between the current segment reference attitude data and the next segment reference attitude data based on the remaining distance data from the current moment to the end of the current segment and the turning lead data; A reference attitude sequence is generated based on the transition mixing coefficient and the attitude prediction sequence at future time points.

5. The multi-rotor UAV mapping operation control method as described in claim 4, characterized in that, A reference attitude sequence is generated based on the transition mixing coefficient and the attitude prediction sequence at future time points, specifically including: The current scene compactness data is used to generate weight decay scale data, and the weight distribution data of the pose prediction sequence at future time is generated based on the weight decay scale data. Based on the current flight segment reference attitude data, the next flight segment reference attitude data, the transition mixing coefficient, the current real-time attitude data, and the prediction weight data, the attitude prediction sequence for future time moments is fused to obtain a reference attitude sequence.

6. The multi-rotor UAV mapping operation control method as described in claim 1, characterized in that, Based on the reference attitude sequence, current real-time attitude data, image lateral offset sequence, and flight segment compactness data, a shooting interval sequence is generated. The mapping image sequence is then classified into main mapping frames and transition frames, specifically including: The attitude consistency data at the current moment is calculated based on the reference attitude sequence and real-time attitude data; The shooting distance data at the current moment is generated based on the segment compactness data, image lateral offset data, and attitude consistency data; The current image is classified based on the attitude consistency data and the image lateral offset data. When the attitude consistency data is not greater than the first threshold and the absolute value of the image lateral offset data is not greater than the second threshold, the current image is written into the main mapping frame set; otherwise, the current image is written into the transition frame set. The main mapping frame index table is formed by establishing an index association between the shooting time data, reference posture data, real-time posture data, image lateral offset data, and shooting distance data corresponding to the main mapping frame set.

7. The multi-rotor UAV mapping operation control method as described in claim 6, characterized in that, Update the reference attitude data for the next flight segment, specifically including: The average image lateral offset data and average attitude deviation data of the main mapping frame set of the current flight segment are statistically averaged to correct the reference attitude data of the next flight segment. The corrected reference attitude data for the next segment is written back to the segment description table and used as the data input for generating the reference attitude sequence for subsequent segments.

8. The multi-rotor UAV mapping operation control method as described in claim 1, characterized in that, After all flight segments are completed, a consistency check is performed. If the check meets preset conditions, the surveying and mapping operation completion results are output, including: After all flight segments are completed, the set of master mapping frames corresponding to each flight segment is extracted according to the master mapping frame index table, and the final inspection status data corresponding to each master mapping frame is collected; wherein, the final inspection status data includes at least the final inspection plane position deviation data, the final inspection attitude deviation data, and the final inspection image lateral offset data. For each flight segment, generate flight segment consistency evaluation data. When the consistency evaluation data of all flight segments meet the preset judgment conditions, output the surveying and mapping operation completion result; otherwise, output the verification mark result.

9. The multi-rotor UAV mapping operation control method as described in claim 8, characterized in that, The method further includes performing continuity constraint processing on the image lateral offset data and flight segment consistency evaluation data, specifically including: Image offset change rate data is generated based on image lateral offset data at adjacent sampling times, and segment connection change rate data is generated based on consistency evaluation data of adjacent segments. When the image offset change rate data is greater than the preset image change threshold, the image lateral offset data at the current moment is subjected to smooth replacement processing, and the smoothed replacement image lateral offset data is rewritten into the image lateral offset sequence; When the segment connection change rate data is greater than the preset segment change threshold, the reference attitude data of the next segment is corrected again, and the original reference attitude data of the next segment is replaced with the corrected reference attitude data of the next segment. Based on the updated image lateral offset sequence and the replaced reference attitude data for the next segment, the reference attitude sequence generation process and consistency determination process are re-executed.

10. A multi-rotor unmanned aerial vehicle (UAV) surveying and mapping operation control system, characterized in that, The system includes: The acquisition module is used to acquire the route point sequence corresponding to the surveying and mapping task, calculate the route direction data, route length data, adjacent route angle data and route compactness data, and generate a route description table. The generation module is used to generate attitude change intensity data and image lateral offset sequence based on the attitude angle time series data and mapping image sequence of multi-rotor UAV during the mapping operation, and to generate scene compactness data and corner lead data by combining the flight segment compactness data. The generation module is used to generate future attitude prediction sequences and reference attitude sequences based on the segment description table corresponding to the current segment and the attitude angle time series data corresponding to the current time. The classification module is used to generate a shooting interval sequence based on the reference attitude sequence, the current real-time attitude data, the image lateral offset sequence, and the flight segment compactness data, to perform main mapping frame classification and transition frame classification on the mapping image sequence, to form flight segment statistical results, and to update the reference attitude data for the next flight segment. The output module is used to continuously perform control of subsequent segments based on the updated reference attitude data of the next segment, and to perform a consistency determination after all segments are completed. When the determination meets the preset conditions, the module outputs the mapping operation completion result.

Images