Heavy load type AMR control method and system based on eVTOL driving technology

Abstract

The invention discloses a heavy load type AMR control method and system based on an eVTOL driving technology, relates to the technical field of heavy load type autonomous mobile robot control, and is used for solving the problem of high energy consumption caused by load shaking due to attitude measurement element data drift of a heavy load type AMR in a complex industrial environment. The method comprises the following steps: acquiring inertial measurement unit data of the heavy load type AMR, camera array image information of a surrounding environment of the heavy load type AMR and a magnetic field intensity sensor array monitoring result of a local magnetic field around the heavy load type AMR; according to the camera array image information, the relative pose of the heavy load type AMR is calculated; according to the attitude information of the heavy load type AMR and the physical characteristics of the heavy load, the attitude correction torque of the heavy load type AMR is calculated; and decomposing the attitude correction torque into a thrust adjustment instruction of the eVTOL driving unit, and performing energy consumption optimization thrust distribution of the eVTOL driving unit based on the thrust adjustment instruction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of heavy-load autonomous mobile robot control, and particularly relates to a heavy-load AMR control method and system based on eVTOL driving technology. BACKGROUND

[0002] When a conventional existing heavy-load autonomous mobile robot (AMR) operates in a complex industrial environment, especially in an area with local strong magnetic field interference, the attitude measurement element (such as a magnetometer in an inertial measurement unit IMU) is easily affected, causing subtle but continuous drift of the attitude data. This unnoticed attitude deviation, when combined with the irregular physical characteristics of heavy loads, can cause the load to sway slightly in the air. In order to correct this sway, the control system is forced to frequently and substantially adjust the thrust output of the eVTOL driving unit, causing the motor to operate in an inefficient manner, consuming a large amount of electrical energy, and possibly triggering an unplanned emergency landing, increasing safety risks and economic losses. SUMMARY

[0003] The present application provides a heavy-load AMR control method and system based on eVTOL driving technology, aiming to solve the problem of attitude measurement element data drift of heavy-load AMR in a complex industrial environment, especially under strong magnetic field interference, leading to load sway, and further causing high energy consumption, premature battery depletion, and unplanned emergency landing.

[0004] In a first aspect, to solve the above technical problems, the present application provides a heavy-load AMR control method based on eVTOL driving technology, which comprises: acquiring inertial measurement unit data of the heavy-load AMR, camera array image information of the environment around the heavy-load AMR, and magnetic field strength sensor array monitoring results of the local magnetic field around the heavy-load AMR; calculating the relative pose of the heavy-load AMR according to the camera array image information; adjusting the weight of the magnetometer data in the fusion algorithm according to the magnetic field strength sensor array monitoring results, and fusing the inertial measurement unit data and the relative pose according to the adjusted fusion algorithm to obtain the attitude information of the heavy-load AMR; monitoring the physical characteristics of the heavy load carried by the heavy-load AMR, including the center of gravity position and the moment of inertia; calculating the attitude correction moment of the heavy-load AMR according to the attitude information of the heavy-load AMR and the physical characteristics of the heavy load; decomposing the attitude correction moment into thrust adjustment instructions of the eVTOL driving unit, and performing energy consumption optimization thrust distribution of the eVTOL driving unit based on the thrust adjustment instructions.

[0005] In a second aspect, the application provides a heavy-load AMR control system based on eVTOL driving technology, which comprises: a data acquisition module for acquiring inertial measurement unit data of the heavy-load AMR, camera array image information of the environment around the heavy-load AMR, and magnetic field strength sensor array monitoring results of the local magnetic field around the heavy-load AMR; a relative pose calculation module for calculating the relative pose of the heavy-load AMR according to the camera array image information; an attitude information fusion module for adjusting the weight of the magnetometer data in the fusion algorithm according to the magnetic field strength sensor array monitoring results, and fusing the inertial measurement unit data and the relative pose according to the adjusted fusion algorithm to obtain the attitude information of the heavy-load AMR; a load physical property perception module for monitoring the physical properties of the heavy load carried by the heavy-load AMR, including the center of gravity position and the moment of inertia; an attitude correction moment calculation module for calculating the attitude correction moment of the heavy-load AMR according to the attitude information of the heavy-load AMR and the physical properties of the heavy load; a thrust distribution and energy consumption optimization module for decomposing the attitude correction moment into thrust adjustment instructions of the eVTOL driving units, and performing energy consumption optimization thrust distribution of the eVTOL driving units based on the thrust adjustment instructions.

[0006] The application has at least the following beneficial effects: the heavy-load AMR control method based on eVTOL driving technology disclosed in the application effectively solves the problem of inaccurate attitude information caused by the deviation of magnetometer data under strong magnetic field interference in the prior art by acquiring inertial measurement unit data, camera array image information, and magnetic field strength sensor array monitoring results, and dynamically adjusting the weight of magnetometer data in the fusion algorithm according to the magnetic field strength sensor array monitoring results. This dynamic weight adjustment mechanism enables the fusion algorithm to more accurately fuse the inertial measurement unit data and the relative pose, thereby obtaining more accurate AMR attitude information and avoiding attitude drift caused by magnetic field interference.

[0007] On this basis, the application further calculates the attitude correction moment of the AMR by monitoring the physical properties such as the center of gravity position and the moment of inertia of the heavy load carried by the heavy-load AMR, and combining the accurate attitude information. This step effectively solves the load shaking problem caused by the combination of attitude deviation and heavy load characteristics in the prior art. By accurately calculating the attitude correction moment, the AMR can actively offset the load shaking and maintain flight stability.

[0008] Finally, the attitude correction moment is decomposed into thrust adjustment instructions of the eVTOL driving units, and energy consumption optimized thrust distribution is performed based on the instructions. This technical solution overcomes the problem of frequent and large adjustment of thrust in the prior art to correct load sway, resulting in low efficiency of the motor and high energy consumption. Through energy consumption optimized thrust distribution, the AMR can maximize energy consumption while maintaining stable attitude, significantly extending the battery endurance time, avoiding unplanned emergency landing due to premature power consumption, thereby improving the operation efficiency and safety of the AMR, and reducing potential equipment damage and personnel safety risks.

[0009] In summary, through a series of technical means such as multi-source data fusion, dynamic weight adjustment, load physical property perception, accurate attitude correction and energy consumption optimized thrust distribution, the core technical problems of attitude measurement inaccuracy, load sway, high energy consumption and unplanned emergency landing of heavy-duty AMR in complex industrial environments, especially under strong magnetic field interference, are effectively solved. The control accuracy, operation stability, energy efficiency and overall safety of the heavy-duty AMR are significantly improved, which has significant progress and practical value. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 is a flowchart of the control method of the heavy-duty AMR based on the eVTOL driving technology provided by the present application. DETAILED DESCRIPTION

[0011] The technical solutions in the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the present application described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0012] It should be noted that: similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present application, the terms "first", "second" and the like are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0013] In modern industrial production, heavy-duty autonomous mobile robots (AMRs) combined with electric vertical take-off and landing (eVTOL) technology play an increasingly important role in the field of material handling. Such robots can carry several tons of industrial components and transport them across areas and floors in complex production environments, greatly improving logistics efficiency. However, the application of eVTOL technology to heavy-duty AMRs and their operation in complex industrial environments also presents unique control challenges.

[0014] In practical applications, the industrial environment in which heavy-duty AMRs operate is often challenging, for example, there are local strong magnetic field interferences. When the core attitude measurement elements of a heavy-duty AMR, such as its high-precision inertial measurement unit (IMU), enter the influence range of such a strong magnetic field, its performance will be subtly affected. The IMU usually contains magnetometers to provide an absolute heading reference and assist in correcting the drift of the gyroscope. A strong external magnetic field can introduce bias to the readings of these magnetometers. This bias in turn can cause subtle, non-linear drift in the attitude data output by the IMU, i.e., the pitch, roll, and yaw angles of the AMR. This means that the self-reported attitude information of the AMR will gradually deviate from its true physical attitude. This bias can be subtle enough, or develop slowly enough, that it fails to reach the pre-set fault detection threshold within the AMR control system. Therefore, the system does not trigger a direct "magnetic field interference" or "IMU failure" alarm. Instead, it continues to operate, processing the attitude data that it considers valid but slightly abnormal. This creates a hidden discrepancy between the internal model of the AMR's state in the control system and the actual physical state, forming a silent error that can accumulate over time.

[0015] As the AMR continues to climb upwards and prepares to transition in mid-air to the upper floor, the cumulative effect of these small and uncorrected attitude biases becomes more pronounced, combined with the characteristics of the heavy load it carries. The heavy load itself has an irregular shape and its center of gravity may be high or off-center. Even if the AMR body has a small unperceived tilt angle (resulting from subtle errors in attitude data), it can generate an unbalanced force on the heavy load with significant inertial mass. This unbalanced force, combined with the inertia of the load itself and the small accelerations during flight, can cause the load to start to sway slightly but continuously in the air.

[0016] To correct this detected load sway, the control system initiates a series of corrective actions. The main goal is to restore stability and maintain the predetermined flight trajectory. To do this, the system must frequently and substantially adjust the thrust output of each eVTOL drive unit. This constant, high-frequency thrust modulation means that the drive motors are operating in a highly dynamic regime, constantly accelerating and decelerating. This rapid load fluctuation causes the motors to consume electrical energy in an inefficient manner, generating a large amount of heat and consuming electrical energy at a rate far exceeding that required for steady flight or hovering. During such corrective operations, energy conversion efficiency is significantly reduced.

[0017] Further complexity arises from the fact that industrial production processes can require the AMR to enter a waiting state upon arrival at the target floor. During this extended hovering phase, the AMR continues to be influenced by the strong magnetic field. This constant interference means that the core attitude measurement elements continue to provide data with subtle drift, resulting in the persistence of hidden attitude errors. As a result, the heavy load continues to sway slightly, and the control system is forced to maintain high-intensity, frequent thrust adjustments throughout the waiting period. This abnormal, high-power consumption state caused by constant combat against load instability causes the on-board power battery pack to deplete at a rate much faster than expected for normal hovering.

[0018] Due to the unexpectedly high power consumption during the long, unstable hovering period, the AMR's power battery pack reaches a critically low level much earlier than planned by the task. This triggers the system's low-power alarm and initiates an automatic emergency landing procedure according to its preset emergency protocol. However, since the AMR is currently in an unplanned waiting area, the "nearest open area" it identifies for emergency landing may not be the best or safest location. An unplanned landing in such an environment not only increases the risk of damage to the AMR or its valuable load, but also may pose potential safety hazards to nearby workers or other automated equipment.

[0019] In view of the above problems, the present application provides a heavy-load AMR control method based on eVTOL drive technology, which dynamically adjusts the weight of magnetometer data in the fusion algorithm by introducing the monitoring results of the magnetic field strength sensor array, effectively suppressing the interference of the strong magnetic field on attitude measurement, and improving the accuracy of attitude information. At the same time, combined with real-time monitoring of the physical characteristics of the heavy load, the attitude correction torque can be more accurately calculated, and the energy consumption of the drive unit can be significantly reduced through energy consumption optimization of thrust distribution, improving the running stability, safety and endurance of the heavy-load AMR in complex industrial environments.

[0020] The heavy-load AMR control method and system based on eVTOL drive technology provided by the embodiments of the present application will be described in detail below through the following specific embodiments.

[0021] Referring toFigure 1 This application provides a heavy-duty AMR control method based on eVTOL drive technology, which may include the following steps:

[0022] S1. Acquire inertial measurement unit data of the heavy-duty AMR, camera array image information of the environment around the heavy-duty AMR, and magnetic field strength sensor array monitoring results of the local magnetic field around the heavy-duty AMR.

[0023] Among them, heavy-duty AMR refers to autonomous mobile robots that can carry a large weight (usually hundreds of kilograms to several tons) and have autonomous navigation, obstacle avoidance and task execution capabilities.

[0024] eVTOL drive technology refers to the use of electric vertical take-off and landing (eVTOL) technology, which uses multiple electric rotors or ducted fans to provide lift and thrust, enabling vertical take-off and landing and aerial maneuvering.

[0025] Inertial measurement unit (IMU) data typically includes raw or pre-processed data from accelerometers, gyroscopes, and magnetometers, used to sense the robot's linear acceleration, angular velocity, and geomagnetic field information.

[0026] Camera array image information refers to image data captured by an array of multiple cameras, used to provide visual information about the environment and relative motion information of the robot.

[0027] The monitoring results of a magnetic field strength sensor array refer to the local magnetic field strength data measured by an array of multiple magnetic field sensors, which are used to assess the interference of the ambient magnetic field on the magnetometer.

[0028] In some embodiments, inertial measurement unit (IMU) data can be directly acquired using IMU sensors mounted on a heavy-duty AMR, which typically includes a three-axis accelerometer, a three-axis gyroscope, and a three-axis magnetometer. Camera array image information can be captured in real time using multiple vision sensors (e.g., stereo cameras, monocular camera arrays, or panoramic cameras) deployed on the heavy-duty AMR. These cameras can be configured to continuously acquire images at a fixed frame rate and transmit the image data to the processing unit. Magnetic field strength sensor array monitoring results are acquired by installing multiple magnetic field strength sensors at key locations on the heavy-duty AMR (e.g., near the IMU or drive unit), which periodically measure local magnetic field strength and aggregate the data.

[0029] S2. Calculate the relative pose of the heavy-duty AMR based on the image information from the camera array.

[0030] Relative pose refers to the three-dimensional position and attitude information of a heavy-duty AMR relative to its surrounding environment or a specific reference point.

[0031] In some embodiments, the relative pose of the heavy-duty AMR is calculated based on camera array image information, which can be achieved using various visual odometry (VO) or visual-inertial odometry (VIO) algorithms. For example, a feature point matching-based method can be used to extract and match feature points from consecutive image frames, and then calculate the motion of the heavy-duty AMR between adjacent frames through geometric transformations, thereby accumulating the relative pose. Another approach is based on a direct method, estimating motion by minimizing image pixel brightness errors.

[0032] S3. Based on the monitoring results of the magnetic field strength sensor array, adjust the weight of the magnetometer data in the inertial measurement unit data in the fusion algorithm, and fuse the inertial measurement unit data with the relative pose according to the adjusted fusion algorithm to obtain the attitude information of the heavy-duty AMR.

[0033] Attitude information refers to the orientation and tilt angle of a heavy-duty AMR in space, typically represented by pitch angle, roll angle, and yaw angle.

[0034] In some embodiments, when the magnetic field strength sensor array detects an abnormal increase in local magnetic field strength, it indicates that the magnetometer data may be interfered with. In this case, the fusion algorithm dynamically reduces the weight of magnetometer data in attitude estimation, relying instead more on gyroscope and accelerometer data, as well as relative pose information provided by visual odometry. For example, an extended Kalman filter (EKF) or an unscented Kalman filter (UKF) can be used as the fusion algorithm, where the covariance matrix of the magnetometer data is adjusted in real time based on the magnetic field strength monitoring results to reflect changes in its reliability.

[0035] S4. Monitor the physical characteristics of the heavy load carried by the heavy-duty AMR, including the center of gravity position and inertia.

[0036] The physical characteristics of heavy loads include the location of the center of gravity and the moment of inertia, which are crucial for accurately assessing the impact of loads on AMR attitude.

[0037] For example, force sensors or pressure sensor arrays can be installed on the support structure of heavy loads to estimate the center of gravity of the heavy load in real time by measuring the force at each support point. For inertia monitoring, the minute vibrations or attitude changes generated by the heavy-duty AMR during motion can be used, combined with IMU data and the geometric model of the heavy load, to estimate its inertia parameters through system identification methods.

[0038] S5. Calculate the attitude correction torque of the heavy-load AMR based on the attitude information and physical characteristics of the heavy load.

[0039] Among them, attitude correction torque refers to the torque required to correct attitude deviations of heavy-duty AMRs.

[0040] Once accurate attitude information and the physical characteristics of the heavy-load AMR are obtained, a precise dynamic model can be established. This model can predict the gravitational force, inertial force, and additional torque generated by load swaying experienced by the heavy-load AMR under the current attitude and load conditions. By comparing the deviation between the actual attitude and the desired attitude, and combining this with the center of gravity position and moment of inertia of the heavy load, the attitude correction torque that needs to be applied to the heavy-load AMR can be accurately calculated to counteract the unbalanced torque and restore stability.

[0041] S6. Decompose the attitude correction torque into thrust adjustment commands for the eVTOL drive unit, and optimize the thrust allocation of the eVTOL drive unit based on the thrust adjustment commands.

[0042] In this context, the eVTOL drive unit refers to the electric rotor or ducted fan that provides thrust, along with its associated motors, ESCs, and other components. The thrust adjustment command is the control signal that adjusts the thrust output of the eVTOL drive unit. Energy-optimized thrust allocation refers to distributing thrust among the drive units in a way that minimizes energy consumption, while still meeting attitude correction requirements, through an optimization algorithm.

[0043] In some embodiments, the attitude correction torque needs to be translated into specific thrust adjustments for each eVTOL drive unit. This is typically achieved through a control allocation matrix that decomposes the total torque and thrust requirements to each drive unit. During this decomposition, energy efficiency optimization algorithms can be introduced. For example, the thrust allocation strategy can be dynamically adjusted based on the current operating state (e.g., motor speed, current, temperature) and efficiency curve of each drive unit to minimize overall energy consumption while meeting the total thrust and attitude correction torque requirements. This can be achieved by solving a constrained optimization problem, such as a quadratic programming (QP) problem, to find the optimal thrust allocation scheme.

[0044] The proposed implementation effectively addresses the issue of decreased attitude measurement accuracy in heavy-duty AMRs operating in complex industrial environments, particularly under strong magnetic field interference, by integrating multi-source sensor data and introducing a dynamic weight adjustment mechanism. Traditional methods often rely on single IMUs or fixed-weight sensor fusion; when the magnetometer is disturbed, attitude estimation errors accumulate, leading to load swaying. This application utilizes a magnetic field strength sensor array to monitor the local magnetic field in real time and dynamically adjusts the weight of magnetometer data in the fusion algorithm accordingly, significantly improving the robustness and accuracy of attitude information.

[0045] Furthermore, this application considers the impact of the physical characteristics of heavy loads (center of gravity position and inertia) on attitude control. In traditional solutions, the physical characteristics of heavy loads are usually treated as fixed values ​​or estimated through preset models, which can lead to inaccurate calculations of attitude correction torque when the load is irregular or dynamically changing. This application, by monitoring the physical characteristics of heavy loads in real time, makes the calculation of attitude correction torque more accurate, thereby more effectively suppressing load sway.

[0046] More importantly, this application introduces an energy consumption optimization mechanism in the thrust allocation stage. Traditional methods often focus only on the timeliness and accuracy of attitude correction, neglecting the energy efficiency of the drive units. When load fluctuations lead to frequent and significant thrust adjustments, traditional methods result in inefficient motor operation and excessive energy consumption. This application decomposes the attitude correction torque into thrust adjustment commands and allocates thrust based on an energy consumption optimization algorithm, enabling each drive unit to operate with minimal energy consumption while meeting attitude correction requirements. This not only extends the endurance of heavy-duty AMRs and reduces operating costs but also reduces motor heat generation and wear, improving system reliability and safety.

[0047] In summary, this application significantly improves the attitude control accuracy, stability, and energy efficiency of heavy-duty AMRs in complex industrial environments through innovative technologies such as multi-sensor fusion, dynamic weight adjustment, heavy load physical characteristic perception, and energy consumption optimization thrust allocation. It effectively solves key problems such as attitude drift, load sway, and high energy consumption in traditional solutions, providing a more reliable and efficient control method for the widespread application of heavy-duty AMRs.

[0048] In some embodiments, this application further proposes the above-mentioned method of fusing inertial measurement unit data and relative pose according to the adjusted fusion algorithm to obtain the attitude information of a heavy-duty AMR, including: filtering the inertial measurement unit data and interpolating the relative pose; and fusing the filtered inertial measurement unit data and the interpolated relative pose according to the adjusted fusion algorithm to obtain the attitude information of the heavy-duty AMR.

[0049] Specifically, filtering inertial measurement unit (IMU) data aims to eliminate or significantly reduce sensor noise, improving data purity and reliability. IMUs typically output data at high frequencies, but this data often contains random noise and drift. If this noise is not processed, it will directly affect the accuracy of attitude estimation. Filtering can be implemented using various algorithms, such as Kalman filtering, complementary filtering, moving average filtering, or Gaussian filtering, to select the appropriate filtering strategy based on specific noise characteristics and real-time requirements.

[0050] The purpose of interpolating relative pose data is to address the inconsistency in sampling frequencies between different sensors and to ensure temporal alignment. Camera array image information typically provides relative pose data at a relatively low frequency, while inertial measurement unit (IMU) data provides it at a high frequency. To effectively combine these two different frequency data in the fusion algorithm, the low-frequency relative pose data needs to be interpolated to synchronize it temporally with the high-frequency IMU data. Interpolation methods can include linear interpolation, spline interpolation, or polynomial interpolation, to ensure data smoothness while restoring the original data's variation trend as much as possible.

[0051] The proposed solution effectively suppresses inherent sensor noise by filtering the inertial measurement unit (IMU) data, resulting in cleaner and more reliable inertial data input to the fusion algorithm. Simultaneously, interpolation of the relative pose data resolves the issues of sampling frequency mismatch and time asynchrony between different sensor data sources, ensuring precise temporal alignment between the IMU data and the relative pose data. These preprocessing steps enable the fusion algorithm to receive high-quality, time-synchronized input data, thereby more accurately estimating the attitude information of the heavy-duty AMR during the fusion process. Filtering improves the short-term accuracy and stability of the inertial data, while interpolation ensures the temporal continuity and synchronization of the relative pose data with the inertial data. Together, they provide the optimal data foundation for the fusion algorithm.

[0052] In some embodiments, this application further proposes the above-mentioned method for monitoring the physical characteristics of the heavy load carried by the heavy-duty AMR, including: monitoring the motion state of the heavy-duty AMR and the internal state and structural deformation of the heavy load carried by the heavy-duty AMR; and dynamically updating the center of gravity position and moment of inertia of the heavy load carried by the heavy-duty AMR based on the internal state, structural deformation and motion state of the heavy load, to obtain the physical characteristics of the heavy load carried by the heavy-duty AMR.

[0053] Specifically, monitoring the motion state of a heavy-duty AMR refers to acquiring information such as its position, velocity, acceleration, and angular velocity in three-dimensional space. This data is typically provided by an inertial measurement unit (IMU), a global positioning system (GPS), or other navigation sensors. The internal state of the heavy load carried by the AMR refers to changes in the mass distribution within the load. For example, for liquids or bulk materials, this includes changes in internal sloshing, flow, or accumulation; for deformable goods, it includes the distribution of internal stress or pressure. These internal states can be monitored by deploying pressure sensors, level sensors, ultrasonic sensors, or image sensors inside the load. Structural deformation refers to changes in the geometry of the load itself or its connection to the AMR, such as bending, twisting, or localized deformation due to uneven stress. Structural deformation can be sensed using strain gauges, laser displacement sensors, or visual measurement systems.

[0054] The dynamic updating of the center of gravity and inertia of the heavy load carried by the heavy-duty AMR refers to using the monitored internal state, structural deformation, and motion state data of the heavy load, and employing specific algorithmic models (e.g., physics-based estimators, Kalman filters, particle filters, or machine learning models) to recalculate the center of gravity and inertia of the heavy load in real time or periodically. The purpose is to ensure that the physical characteristic data of the heavy load used remains consistent with the actual load conditions of the heavy-duty AMR.

[0055] This application's solution comprehensively captures the dynamic changes of heavy-duty AMRs during transportation by real-time monitoring of their motion state, internal load state, and structural deformation. Specifically, the motion state of the heavy-duty AMR provides information on the carrier's own motion, while the internal load state (e.g., liquid sloshing, bulk material movement) and structural deformation (e.g., container deformation, cargo displacement) directly reflect the real-time changes in the load's mass distribution. By comprehensively analyzing and dynamically updating this multi-source information, the center of gravity and inertia of the heavy load can be accurately obtained. Therefore, subsequent calculations of attitude correction moments can be based on the most accurate physical characteristics of the load, ensuring a high degree of match between the calculated attitude correction moments and the actual load conditions of the heavy-duty AMR, thus avoiding control deviations caused by changes in load characteristics.

[0056] In some embodiments, this application further proposes the steps of decomposing the attitude correction torque into thrust adjustment commands for the eVTOL drive unit and performing energy-optimized thrust allocation for the eVTOL drive unit based on the thrust adjustment commands, including: real-time monitoring of the operating status parameters of the eVTOL drive unit, including motor current, motor speed, motor temperature, and fan vibration frequency; evaluating the thrust output efficiency and thrust response characteristics of the eVTOL drive unit based on the operating status parameters; dynamically adjusting the drive unit efficiency parameters and response weights in the energy-optimized thrust allocation algorithm based on the evaluated thrust output efficiency and thrust response characteristics; decomposing the attitude correction torque into thrust adjustment commands for the eVTOL drive unit based on the adjusted drive unit efficiency parameters and response weights, and performing energy-optimized thrust allocation for the eVTOL drive unit based on the thrust adjustment commands.

[0057] Specifically, real-time monitoring of the operating status parameters of the eVTOL drive unit refers to continuously collecting key operating data by deploying corresponding sensors on each eVTOL drive unit. For example, motor current can reflect the real-time load and energy consumption of the drive unit; motor speed is directly related to the magnitude of thrust output; motor temperature is an important indicator for assessing the health status of the drive unit and potential overload risks; and fan vibration frequency can be used to detect wear or failure of internal mechanical components of the drive unit. The real-time acquisition of these parameters provides the basic data for subsequent performance evaluation.

[0058] The evaluation of the thrust output efficiency and thrust response characteristics of the eVTOL drive unit based on operating status parameters can be understood as using this real-time data, combined with a pre-defined drive unit performance model, to calculate the efficiency of each drive unit in converting electrical energy into thrust under the current operating conditions, as well as its response speed and accuracy to changes in thrust commands. Thrust output efficiency typically refers to the magnitude of thrust generated at a given input power, while thrust response characteristics describe the time and dynamic process required for the actual thrust output to stabilize from receiving the command.

[0059] In practical applications, dynamically adjusting the drive unit efficiency parameters and response weights in the energy-optimized thrust allocation algorithm based on the evaluated thrust output efficiency and thrust response characteristics means using the real-time evaluated efficiency and response characteristics as input to update the internal parameters of the energy-optimized thrust allocation algorithm. For example, if the efficiency of a drive unit decreases due to increased temperature, its weight will be reduced when allocating thrust, or it will be preferentially allocated to drive units with higher efficiency. If a drive unit has a slow response speed, its response weight in thrust allocation will be adjusted accordingly when rapid attitude adjustments are needed to ensure a rapid response of the overall system.

[0060] Therefore, based on the adjusted drive unit efficiency parameters and response weights, the attitude correction torque is decomposed into thrust adjustment commands for the eVTOL drive units, and energy-efficient thrust allocation is performed based on these thrust adjustment commands. This means that thrust allocation is no longer static, but dynamically optimized according to the real-time performance of each drive unit. The attitude correction torque is decomposed into thrust commands for each drive unit, which are further optimized to minimize overall energy consumption while ensuring attitude stability and precise control of the heavy-duty AMR, taking into account the current efficiency and response capability of each drive unit.

[0061] This application's solution addresses the problems of static allocation and inability to adapt to real-time operating conditions inherent in traditional energy-optimized thrust allocation by introducing a real-time monitoring and evaluation mechanism for the eVTOL drive unit's operating status. Specifically, when a heavy-duty AMR is performing a task, its eVTOL drive unit's operating status parameters, such as motor current, motor speed, motor temperature, and fan vibration frequency, are continuously acquired. These parameters directly reflect the drive unit's current load, health status, and performance. Based on this real-time data, the thrust output efficiency and thrust response characteristics of each drive unit can be accurately evaluated. For example, when the motor temperature is too high, its efficiency may decrease, and its response speed may slow down. By feeding these real-time evaluation results back into the energy-optimized thrust allocation algorithm, the algorithm can dynamically adjust the drive unit's efficiency parameters and response weights. This means that when decomposing attitude correction torque, the system no longer simply allocates based on preset values ​​but prioritizes the use of currently efficient and fast-responding drive units, or, when necessary, adjusts the allocation strategy to avoid over-reliance on drive units with degraded performance. This dynamic adjustment mechanism ensures that the attitude correction torque can be converted into thrust command most effectively under any operating condition, thereby minimizing overall energy consumption and guaranteeing the attitude control accuracy and response speed of heavy-duty AMRs.

[0062] In some embodiments, this application further proposes a method for real-time monitoring of the operating status parameters of the aforementioned eVTOL drive unit, which specifically includes: deploying a multimodal sensor redundancy array on the eVTOL drive unit, the array containing at least two sensors based on different physical principles for monitoring the same operating status parameters; continuously receiving data from the multimodal sensor redundancy array; when data from any sensor becomes abnormal, initiating a sensor data cross-validation and adaptive fusion mechanism; comparing the same parameter readings from sensors based on different physical principles, and dynamically evaluating the reliability of each sensor by combining historical data trends and the physical model of the drive unit; adjusting the measurement error covariance of each sensor in real time according to environmental conditions and the aging model of the sensor itself; and fusing the effective data from the redundant sensors through a weighted fusion algorithm to obtain the operating status parameters.

[0063] Specifically, a multimodal sensor redundancy array refers to configuring multiple sensors in key components of the eVTOL drive unit. These sensors are not only redundant in number, but more importantly, they employ at least two different physical principles to measure the same operating parameters. For example, for monitoring motor temperature, a thermistor and an infrared temperature sensor can be deployed simultaneously; for motor speed, a Hall effect sensor and an encoder can be used simultaneously. The purpose is to improve the accuracy and robustness of measurement results by cross-validating sensors based on different principles, and to reduce the risks associated with single sensor failure or limitations of a specific measurement principle.

[0064] Continuously receiving data from the multimodal sensor redundancy array means that the control system continuously acquires real-time measurement data from all deployed sensors, providing a continuous data stream for subsequent data processing and fusion.

[0065] In practical applications, when any sensor data becomes abnormal—for example, when the reading exceeds a preset threshold, the data fluctuates drastically, or deviates significantly from the readings of other similar sensors—the system will automatically activate the sensor data cross-validation and adaptive fusion mechanism. This mechanism aims to identify, diagnose, and process abnormal data through intelligent algorithms, ensuring the reliability of the final output operating status parameters.

[0066] Specifically, comparing the same parameter readings from sensors based on different physical principles, and combining historical data trends with the physical model of the drive unit, dynamically assesses the reliability of each sensor. This means the system compares the measurements of the same parameter from different types of sensors. For example, if a thermistor shows a motor temperature of 80°C, while an infrared sensor shows 60°C, the system will consider the motor's historical operating temperature curve, current load conditions, and the motor's thermodynamic model to comprehensively determine which sensor's data is more reliable, or whether there is a discrepancy between the two. In this way, a reliability weight can be dynamically assigned to each sensor.

[0067] Furthermore, the system dynamically adjusts the measurement error covariance of each sensor in real time based on environmental conditions and the sensor's own aging model. This means that, taking into account the performance differences of sensors under different environments (such as temperature, humidity, and vibration) and the aging effects that may result from long-term use, the system dynamically corrects the measurement uncertainty (i.e., error covariance) of each sensor based on real-time environmental data and a preset aging model. For example, in high-temperature environments, the accuracy of some sensors may decrease, and their error covariance will be increased accordingly.

[0068] Finally, a weighted fusion algorithm is used to fuse the valid data from redundant sensors to obtain the operating status parameters. This means that the system will use weighted fusion algorithms such as Kalman filtering, extended Kalman filtering, or unscented Kalman filtering, based on the dynamically evaluated sensor reliability and the adjusted measurement error covariance, to comprehensively process all reliable and calibrated sensor data, thereby outputting a highly accurate, robust, and high-confidence eVTOL drive unit operating status parameter.

[0069] This application's solution fundamentally addresses the limitations of single-sensor monitoring by deploying a multimodal sensor redundancy array. When a sensor fails or is interfered with, other redundant sensors based on different physical principles can still provide valid measurement data. Furthermore, by initiating sensor data cross-validation and adaptive fusion mechanisms, the system can dynamically evaluate the reliability of each sensor and make intelligent judgments based on historical data trends and the physical model of the driving unit. Thus, even in complex and changing environments or facing sensor aging, the system can adjust the measurement error covariance in real time to ensure the accuracy of data fusion. Finally, through a weighted fusion algorithm, the valid data from multiple redundant sensors are comprehensively processed to obtain highly accurate and robust operating status parameters. This mechanism significantly improves the reliability and accuracy of operating status parameter monitoring, providing a solid data foundation for subsequent thrust output efficiency and thrust response characteristic evaluation.

[0070] In some embodiments, this application further proposes to evaluate the thrust output efficiency and thrust response characteristics of the eVTOL drive unit based on the above-mentioned operating state parameters, including: acquiring the flight altitude, airspeed, and wind field sensor array monitoring results of the heavy-duty AMR; correcting the evaluation parameters of the drive unit's thrust output efficiency based on the flight altitude and airspeed; correcting the evaluation parameters of the drive unit's thrust response characteristics based on the wind field sensor array monitoring results; and evaluating the thrust output efficiency and thrust response characteristics of the eVTOL drive unit based on the corrected evaluation parameters of the drive unit's thrust output efficiency and the evaluation parameters of the drive unit's thrust response characteristics.

[0071] Specifically, acquiring the flight altitude, airspeed, and wind field monitoring results of heavy-duty AMRs aims to provide crucial environmental context information for the performance evaluation of the propulsion unit. Flight altitude and airspeed are directly related to air density and relative airflow velocity, which are important physical quantities affecting the thrust generation efficiency of propellers or ducted fans. The wind field sensor array monitoring results provide detailed information on the local airflow field where the propulsion unit is located, including wind speed and direction, which is essential for accurately assessing the dynamic response characteristics of the propulsion unit. This data can be acquired in real time through various methods such as airborne altimeters, pitot tubes, Doppler radar, lidar, or distributed anemometers.

[0072] The adjustment of the evaluation parameters for the thrust output efficiency of the drive unit based on flight altitude and airspeed refers to adjusting the theoretical thrust output efficiency of the drive unit according to the current flight altitude and airspeed using a known aerodynamic model or a pre-calibrated lookup table. For example, as flight altitude increases, air density decreases, resulting in less thrust at the same rotational speed, and thus a change in thrust output efficiency. By introducing these adjustments, the evaluation parameters can more accurately reflect the actual efficiency of the drive unit under the current environment.

[0073] In practical applications, adjusting the evaluation parameters of the drive unit's thrust response characteristics based on the monitoring results of the wind field sensor array means considering the impact of the external wind field on the thrust response speed and stability of the drive unit. For example, in strong gusts or turbulent environments, the drive unit needs a faster response speed to counteract external disturbances, or its thrust response may be delayed or overshooted due to airflow impact. By analyzing the monitoring results of the wind field sensor array, evaluation parameters such as the damping coefficient and inertial parameters in the thrust response model can be dynamically adjusted to better reflect actual dynamic response behavior.

[0074] Therefore, based on the revised evaluation parameters of the drive unit's thrust output efficiency and thrust response characteristics, the thrust output efficiency and thrust response characteristics of the eVTOL drive unit are evaluated, ensuring the accuracy and real-time nature of the evaluation results. These revised evaluation results will serve as input to the subsequent energy-optimized thrust allocation algorithm, enabling thrust allocation decisions to fully consider environmental factors and thus achieve more refined control.

[0075] This application's solution modifies the evaluation parameters for the thrust output efficiency and thrust response characteristics of the eVTOL drive unit by incorporating monitoring results from a heavy-duty AMR's flight altitude, airspeed, and wind field sensor array. The actual performance of the eVTOL drive unit is closely related to environmental conditions; for example, air density directly affects thrust magnitude and efficiency, while external airflow influences the drive unit's dynamic response. By acquiring and utilizing these environmental parameters, the drive unit's performance model can be calibrated in real time, enabling the evaluation results to more accurately reflect the drive unit's true performance under the current operating environment. This environment-aware dynamic correction mechanism effectively overcomes the limitations of relying solely on internal operating state parameters for evaluation, ensuring that subsequent thrust allocation algorithms can make decisions based on more accurate performance data.

[0076] In some embodiments, the above-mentioned evaluation parameters for correcting the thrust output efficiency of the drive unit based on flight altitude and airspeed specifically include: acquiring raw sensor data of flight altitude and airspeed; performing multi-source heterogeneous data redundancy verification on the raw sensor data, and adaptively correcting the raw sensor data in combination with the motion state and environmental characteristics of the heavy-duty AMR to obtain the corrected flight altitude and airspeed; and correcting the evaluation parameters for the thrust output efficiency of the drive unit based on the corrected flight altitude and airspeed.

[0077] Specifically, acquiring raw sensor data for flight altitude and airspeed refers to obtaining flight altitude data through various sensors deployed on heavy-duty AMRs, such as barometric altimeters, lidar, ultrasonic sensors, and GPS receivers, and airspeed data through pitot tubes, inertial measurement units (IMUs), or visual odometry. These sensors may be based on different physical principles to provide raw measurements of flight altitude and airspeed.

[0078] The first step, multi-source heterogeneous data redundancy verification of raw sensor data, involves cross-comparing and verifying identical or related parameter data from multiple sensors of different types or the same type but at different locations. For example, flight altitude data can be provided simultaneously by a barometric altimeter and a lidar; by comparing their readings, abnormal data can be detected and identified. The second step, adaptive correction of the raw sensor data based on the motion state and environmental characteristics of the heavy-load AMR, involves dynamically adjusting the parameters of the correction algorithm according to the AMR's current flight speed, acceleration, attitude angles, and other motion state information, as well as environmental characteristics such as temperature, humidity, air pressure, and wind speed. For example, during high-speed flight or violent maneuvers, the weight of inertial measurement unit (IMU) data can be increased; in complex terrain or environments with obstacles, lidar data can be relied upon more heavily. Adaptive correction aims to compensate for inherent sensor measurement errors, data deviations caused by environmental changes, and sensor aging, thereby obtaining more accurate and reliable corrected flight altitude and airspeed.

[0079] Therefore, based on the corrected flight altitude and airspeed, the evaluation parameters of the thrust output efficiency of the drive unit are adjusted, ensuring that the input data used for correction has higher accuracy and reliability, and providing a solid foundation for subsequent energy consumption optimization and thrust allocation.

[0080] This application's solution effectively solves the problem of thrust output efficiency evaluation parameter correction deviation caused by inaccurate or unreliable raw sensor data in traditional solutions by introducing a multi-source heterogeneous data redundancy verification and adaptive correction mechanism for raw flight altitude and airspeed sensor data. Specifically, the multi-source heterogeneous data redundancy verification mechanism utilizes the complementarity between different sensors to identify and eliminate abnormal data through cross-validation, thereby improving data robustness. Simultaneously, adaptive correction combined with the motion state and environmental characteristics of the heavy-duty AMR allows the correction process to be dynamically adjusted according to actual operating conditions, compensating for sensor measurement errors under different conditions and ensuring high-precision flight altitude and airspeed information can be obtained in various complex environments. It is precisely because of the rigorously verified and corrected flight altitude and airspeed data that the evaluation parameters of the drive unit's thrust output efficiency can be more accurately corrected, providing a more accurate input for subsequent energy consumption optimization and thrust allocation, avoiding control deviations and energy waste caused by inaccurate data.

[0081] In some embodiments, the evaluation parameters of the thrust response characteristics of the drive unit are corrected based on the monitoring results of the wind field sensor array, including: acquiring the flight attitude and motion speed of the heavy-load AMR; fitting the local airflow field model to the monitoring results of the wind field sensor array to obtain the local airflow velocity and direction at the location of the drive unit; and correcting the evaluation parameters of the thrust response characteristics of the drive unit based on the local airflow velocity and direction, combined with the flight attitude and motion speed of the heavy-load AMR.

[0082] Specifically, acquiring the flight attitude and velocity of a heavy-duty AMR involves fusing data from multiple sensors, such as inertial measurement unit (IMU) data, GPS data, and visual odometry, to obtain in real-time the AMR's attitude (e.g., pitch, roll, yaw angles) in three-dimensional space and its velocity relative to the ground (e.g., linear velocity and angular velocity). This information is crucial for understanding the interaction between the heavy-duty AMR and the surrounding airflow.

[0083] This process involves fitting a local airflow field model to the monitoring results of the wind field sensor array to obtain the local airflow velocity and direction at the location of the drive unit. This can be understood as using a computational fluid dynamics (CFD) model or a machine learning-based airflow prediction model, combined with monitoring data from the wind field sensor array (e.g., miniature Pitot tubes, hot-wire anemometers, or ultrasonic anemometers deployed at different locations on the AMR), to perform a refined modeling of the airflow field around the heavy-duty AMR. The aim is to overcome the limitations of traditional wind field sensors, which can only provide macroscopic or local point-based wind speed information. By fitting the data, the precise local airflow velocity and direction at the location of each eVTOL drive unit can be obtained, which is crucial for accurately evaluating the thrust response of the drive unit.

[0084] In practical applications, the evaluation parameters of the thrust response characteristics of the drive unit are modified based on the local airflow velocity and direction, combined with the flight attitude and velocity of the heavy-load AMR. Specifically, this involves comprehensively considering the fitted local airflow information with the motion state (flight attitude and velocity) of the heavy-load AMR itself. For example, when a heavy-load AMR flies at a specific speed and attitude, its drive unit may face complex situations such as oncoming airflow, crossflow, or wake. By incorporating these factors, relevant parameters in the thrust response model of the drive unit, such as the thrust coefficient, drag coefficient, or response time constant, can be dynamically adjusted to more accurately reflect the actual performance of the drive unit under the current airflow environment.

[0085] This application's approach first acquires the flight attitude and velocity of a heavy-load AMR, providing essential kinematic background information for subsequent airflow field analysis. Since the motion of the heavy-load AMR significantly affects its surrounding local airflow field, these motion parameters are fundamental to accurately evaluating the thrust response characteristics of the drive unit. Secondly, by fitting a local airflow field model to the monitoring results of the wind field sensor array, discrete sensor data can be transformed into continuous and precise local airflow velocity and direction information. This fitting process effectively compensates for the inability of a single or limited number of sensors to comprehensively capture complex airflow fields, allowing for precise quantification of the micro-airflow environment of each drive unit. Finally, combining this precise local airflow information with the flight attitude and velocity of the heavy-load AMR enables a more comprehensive and accurate correction of the evaluation parameters for the drive unit's thrust response characteristics. This comprehensive correction mechanism ensures that the thrust response evaluation parameters can adapt in real-time to the dynamic flight conditions and complex airflow environment of the heavy-load AMR, thus providing a more reliable basis for subsequent energy consumption optimization and thrust allocation.

[0086] In some embodiments, the above-mentioned fitting of the local airflow field model to the monitoring results of the wind field sensor array to obtain the local airflow velocity and direction at the location of the drive unit includes: acquiring the geometric model of the heavy-duty AMR and the three-dimensional map information of the surrounding environment of the heavy-duty AMR; dynamically adjusting the grid density of the computational fluid dynamics model according to the current position and motion state of the heavy-duty AMR, wherein the local grid density is increased when the heavy-duty AMR is close to an obstacle or in a high-gradient wind field region, and the grid density is decreased in open areas; adopting a parallel computing architecture to decompose the local airflow field model fitting task into multiple processing units for parallel execution; and fitting the local airflow field model to the monitoring results of the wind field sensor array based on the geometric model of the heavy-duty AMR, the three-dimensional map information of the surrounding environment of the heavy-duty AMR, the dynamically adjusted grid density, and the fitting task results after parallel execution to obtain the local airflow velocity and direction at the location of the drive unit.

[0087] Obtaining the geometric model of a heavy-duty AMR refers to acquiring detailed three-dimensional structural data of the AMR, such as CAD models or point cloud data. The purpose is to provide accurate boundary conditions for computational fluid dynamics (CFD) simulations, ensuring that the simulation results accurately reflect the interaction between the airflow and the AMR itself. The three-dimensional map information of the environment surrounding the heavy-duty AMR can be understood as precise three-dimensional spatial data of the AMR's operating area, including information on obstacles and terrain. Its purpose is to provide geometric constraints on the external environment for CFD simulations, enabling the local airflow field model to consider the influence of the environment on the airflow.

[0088] In practical applications, the mesh density of the computational fluid dynamics model is dynamically adjusted based on the current position and motion state of the heavy-duty AMR. For example, the fineness of the mesh can be adjusted in real time according to the distance between the AMR and obstacles, changes in the wind gradient, etc. Specifically, when the heavy-duty AMR is close to obstacles or in a high-gradient wind field region, the local mesh density is increased to capture complex airflow details more precisely; while in open areas, the mesh density is reduced to reduce unnecessary computation. The aim is to optimize the utilization efficiency of computing resources while ensuring simulation accuracy. In addition, a parallel computing architecture is adopted, decomposing the task of fitting the local airflow field model into multiple processing units for parallel execution. The purpose is to significantly shorten the computation time and meet the computational speed requirements of real-time control.

[0089] This application's solution provides more accurate and comprehensive physical boundary conditions for fitting the local airflow field model by acquiring the geometric model of the heavy-duty AMR and the 3D map information of its surrounding environment, enabling the fitted airflow field model to more realistically reflect the actual situation. Simultaneously, by dynamically adjusting the grid density of the computational fluid dynamics model based on the AMR's current position and motion state, intelligent allocation of computational resources is achieved. Specifically, the grid density is increased in areas requiring high-precision simulation (such as near obstacles or high-gradient wind fields) and decreased in areas with lower computational demands, thus significantly improving computational efficiency while maintaining simulation accuracy. Furthermore, a parallel computing architecture is employed to decompose the complex model fitting task and execute it in parallel by multiple processing units, greatly shortening the computation time and enabling real-time or near-real-time updates of the local airflow field model. Therefore, this application's solution can more accurately and efficiently obtain the local airflow velocity and direction at the location of the drive unit, providing a solid data foundation for subsequent thrust response characteristic evaluation.

[0090] This application also discloses a heavy-duty AMR control system based on eVTOL drive technology. The system includes: a data acquisition module for acquiring inertial measurement unit (IMU) data of the heavy-duty autonomous mobile robot (AMR), camera array image information of the environment surrounding the heavy-duty AMR, and magnetic field strength sensor array monitoring results of the local magnetic field around the heavy-duty AMR; a relative pose calculation module for calculating the relative pose of the heavy-duty AMR based on the camera array image information; and a pose information fusion module for adjusting the weight of magnetometer data in the IMU data within the fusion algorithm based on the magnetic field strength sensor array monitoring results, and adjusting accordingly. The subsequent fusion algorithm fuses the inertial measurement unit data with the relative pose to obtain the attitude information of the heavy-duty AMR; the load physical characteristic sensing module is used to monitor the physical characteristics of the heavy load carried by the heavy-duty AMR, the physical characteristics including the center of gravity position and inertia; the attitude correction torque calculation module is used to calculate the attitude correction torque of the heavy-duty AMR based on the attitude information of the heavy-duty AMR and the physical characteristics of the heavy load; the thrust distribution and energy consumption optimization module is used to decompose the attitude correction torque into thrust adjustment commands of the eVTOL drive unit, and perform energy consumption optimization thrust distribution of the eVTOL drive unit based on the thrust adjustment commands.

[0091] The control system of this application acquires multi-source sensor data through a data acquisition module, and the attitude information fusion module dynamically adjusts the weight of magnetometer data in the fusion algorithm based on the monitoring results of the magnetic field strength sensor array. This effectively suppresses the interference of strong magnetic fields on attitude measurement and improves the accuracy of attitude information. Simultaneously, the load physical characteristic sensing module monitors the physical characteristics of heavy loads in real time, and the attitude correction torque calculation module can calculate the attitude correction torque more accurately. Furthermore, the thrust distribution and energy consumption optimization module optimizes thrust distribution for energy consumption, significantly reducing the energy consumption of the eVTOL drive unit and improving the operational stability, safety, and endurance of heavy-duty AMRs in complex industrial environments.

[0092] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A heavy-duty AMR control method based on eVTOL drive technology, characterized in that, The method includes: Acquire inertial measurement unit data of the heavy-duty AMR, camera array image information of the environment around the heavy-duty AMR, and magnetic field strength sensor array monitoring results of the local magnetic field around the heavy-duty AMR; The relative pose of the heavy-duty AMR is calculated based on the image information from the camera array. Based on the monitoring results of the magnetic field strength sensor array, the weight of the magnetometer data in the inertial measurement unit data in the fusion algorithm is adjusted, and the inertial measurement unit data is fused with the relative pose according to the adjusted fusion algorithm to obtain the attitude information of the heavy-duty AMR. Monitor the physical characteristics of the heavy load carried by the heavy-duty AMR, including the center of gravity position and moment of inertia; Based on the attitude information and physical characteristics of the heavy-load AMR, the attitude correction torque of the heavy-load AMR is calculated. The attitude correction torque is decomposed into thrust adjustment commands for the eVTOL drive unit, and the energy consumption optimization thrust allocation of the eVTOL drive unit is performed based on the thrust adjustment commands.

2. The heavy-duty AMR control method based on eVTOL drive technology according to claim 1, characterized in that, The process of fusing the inertial measurement unit data with the relative pose according to the adjusted fusion algorithm to obtain the attitude information of the heavy-duty AMR includes: The inertial measurement unit data is filtered, and the relative pose is interpolated. According to the adjusted fusion algorithm, the filtered inertial measurement unit data and the differentially processed relative pose are fused to obtain the attitude information of the heavy-duty AMR.

3. The heavy-duty AMR control method based on eVTOL drive technology according to claim 2, characterized in that, The monitoring of the physical characteristics of the heavy load borne by the heavy-duty AMR includes: Monitor the motion state of the heavy-duty AMR and the internal state and structural deformation of the heavy load borne by the heavy-duty AMR; Based on the internal state of the heavy load, structural deformation, and motion state of the heavy-load AMR, the center of gravity position and moment of inertia of the heavy load carried by the heavy-load AMR are dynamically updated to obtain the physical characteristics of the heavy load carried by the heavy-load AMR.

4. The heavy-duty AMR control method based on eVTOL drive technology according to claim 1, characterized in that, The step of decomposing the attitude correction torque into thrust adjustment commands for the eVTOL drive unit, and performing energy-optimized thrust allocation for the eVTOL drive unit based on the thrust adjustment commands, includes: The operating status parameters of the eVTOL drive unit are monitored in real time, including motor current, motor speed, motor temperature and fan vibration frequency. Based on the operating status parameters, evaluate the thrust output efficiency and thrust response characteristics of the eVTOL drive unit; Based on the evaluated thrust output efficiency and thrust response characteristics, the drive unit efficiency parameters and response weights in the energy consumption optimization thrust allocation algorithm are dynamically adjusted. Based on the adjusted drive unit efficiency parameters and the response weights, the attitude correction torque is decomposed into thrust adjustment commands for the eVTOL drive unit, and energy consumption optimization thrust allocation for the eVTOL drive unit is performed based on the thrust adjustment commands.

5. The heavy-duty AMR control method based on eVTOL drive technology according to claim 4, characterized in that, The real-time monitoring of the operating status parameters of the eVTOL drive unit includes: A multimodal sensor redundancy array is deployed on the eVTOL drive unit. The multimodal sensor redundancy array contains at least two sensors based on different physical principles for monitoring the same operating status parameters. Continuously receive data from the multimodal sensor redundancy array; When data from any sensor becomes abnormal, the sensor data cross-validation and adaptive fusion mechanism is activated. By comparing the same parameter readings from sensors based on different physical principles and combining historical data trends with the physical model of the drive unit, the reliability of each sensor is dynamically evaluated. The measurement error covariance of each sensor is adjusted in real time based on environmental conditions and the aging model of the sensor itself. The operating status parameters are obtained by fusing effective data from redundant sensors using a weighted fusion algorithm.

6. The heavy-duty AMR control method based on eVTOL drive technology according to claim 4, characterized in that, The step of evaluating the thrust output efficiency and thrust response characteristics of the eVTOL drive unit based on the operating state parameters includes: The flight altitude, airspeed, and wind field sensor array monitoring results of the heavy-duty AMR were obtained. The evaluation parameters for the thrust output efficiency of the drive unit are adjusted based on the flight altitude and the airspeed. Based on the monitoring results of the wind field sensor array, the evaluation parameters of the thrust response characteristics of the drive unit are corrected; The thrust output efficiency and thrust response characteristics of the eVTOL drive unit are evaluated based on the revised evaluation parameters of the drive unit thrust output efficiency and the evaluation parameters of the drive unit thrust response characteristics.

7. The heavy-duty AMR control method based on eVTOL drive technology according to claim 6, characterized in that, The evaluation parameters for correcting the thrust output efficiency of the drive unit based on the flight altitude and the airspeed include: Acquire raw sensor data for the flight altitude and airspeed; The original sensor data is subjected to multi-source heterogeneous data redundancy verification, and adaptive correction is performed on the original sensor data in combination with the motion state and environmental characteristics of the heavy-duty AMR to obtain the corrected flight altitude and airspeed. The evaluation parameters for the thrust output efficiency of the drive unit are corrected based on the corrected flight altitude and airspeed.

8. The heavy-duty AMR control method based on eVTOL drive technology according to claim 6, characterized in that, The step of correcting the evaluation parameters of the thrust response characteristics of the drive unit based on the monitoring results of the wind field sensor array includes: The flight attitude and velocity of the heavy-duty AMR were obtained; The local airflow field model is fitted to the monitoring results of the wind field sensor array to obtain the local airflow velocity and direction at the location of the driving unit; Based on the local airflow velocity and direction, combined with the flight attitude and motion speed of the heavy-duty AMR, the evaluation parameters of the thrust response characteristics of the drive unit are corrected.

9. The heavy-duty AMR control method based on eVTOL drive technology according to claim 8, characterized in that, The step of fitting a local airflow field model to the monitoring results of the wind field sensor array to obtain the local airflow velocity and direction at the location of the driving unit includes: Obtain the geometric model of the heavy-duty AMR and the three-dimensional map information of the environment surrounding the heavy-duty AMR; Based on the current position and motion state of the heavy-duty AMR, the grid density of the computational fluid dynamics model is dynamically adjusted. Specifically, when the heavy-duty AMR is close to an obstacle or in a high-gradient wind field region, the local grid density is increased, and in open areas, the grid density is decreased. A parallel computing architecture is adopted to decompose the task of fitting the local airflow field model into multiple processing units for parallel execution; Based on the geometric model of the heavy-duty AMR, the three-dimensional map information of the environment surrounding the heavy-duty AMR, the dynamically adjusted grid density, and the fitting task results after parallel execution, the local airflow field model is fitted to the monitoring results of the wind field sensor array to obtain the local airflow velocity and direction at the location of the drive unit.

10. A heavy-duty AMR control system based on eVTOL drive technology, characterized in that, The system includes: The data acquisition module is used to acquire inertial measurement unit data of the heavy-duty AMR, camera array image information of the environment around the heavy-duty AMR, and magnetic field strength sensor array monitoring results of the local magnetic field around the heavy-duty AMR. The relative pose calculation module is used to calculate the relative pose of the heavy-duty AMR based on the image information of the camera array. The attitude information fusion module is used to adjust the weight of the magnetometer data in the inertial measurement unit data in the fusion algorithm according to the monitoring results of the magnetic field strength sensor array, and fuse the inertial measurement unit data with the relative pose according to the adjusted fusion algorithm to obtain the attitude information of the heavy-duty AMR. The load physical characteristics sensing module is used to monitor the physical characteristics of the heavy load carried by the heavy-duty AMR, including the center of gravity position and moment of inertia. The attitude correction torque calculation module is used to calculate the attitude correction torque of the heavy-load AMR based on the attitude information of the heavy-load AMR and the physical characteristics of the heavy load. The thrust distribution and energy consumption optimization module is used to decompose the attitude correction torque into thrust adjustment commands for the eVTOL drive unit, and to perform energy consumption optimization thrust distribution for the eVTOL drive unit based on the thrust adjustment commands.

Images