An aerial work platform anti-collision obstacle avoidance automatic lifting real-time control system

Abstract

The application discloses a high-altitude operation platform anti-collision obstacle avoidance automatic lifting real-time control system and relates to the technical field of control systems. Three-dimensional point cloud data of the periphery of a high-altitude operation platform is acquired, a point cloud coordinate set of entity obstacles is output, the transient deflection displacement of the end of an arm support is superimposed with the rigid body space coordinates to output the absolute space coordinates of the platform, the height of the gravity center of the whole vehicle and the offset distance of the gravity center are calculated, the target Euclidean distance between the absolute space coordinates of the platform and the point cloud coordinate set is calculated, an obstacle avoidance safety boundary function is constructed, a real-time overturning moment function is constructed based on the mass distribution and the inertia parameters of the whole vehicle, in a prediction time domain, the forward invariance constraint of the obstacle avoidance safety boundary function and the anti-overturning constraint of the real-time overturning moment function are taken as hard boundary conditions, a quadratic programming cost function with a preset obstacle avoidance trajectory tracking as a target function is solved, an optimal control increment sequence is output and is converted into an analog electric signal, and an electro-hydraulic proportional actuator is driven to execute an obstacle avoidance action.

Description

Technical Field

[0001] This invention relates to the field of control system technology, and in particular to a real-time control system for automatic lifting and lowering of aerial work platforms to prevent collisions and obstacles. Background Technology

[0002] Aerial work platforms are widely used in construction, equipment maintenance, and bridge inspection. Due to the complexity and variability of the working environment at height, platforms not only face the risk of collisions with obstacles in the air, but also the danger of overturning due to uneven stress or excessive force.

[0003] Currently, Chinese invention with publication number CN120953386A discloses an intelligent control method and system for obstacle avoidance of aerial work platforms. It constructs a three-dimensional spatial map by collecting perception data through multimodal sensors, predicts the trajectory of dynamic obstacles using deep learning and digital twins, plans obstacle avoidance paths using random tree algorithms, and finally generates control commands for vertical lifting and horizontal displacement through fuzzy control algorithms to execute obstacle avoidance actions. The aforementioned technologies have the following shortcomings: Distorted safety boundaries: Traditional solutions assume the platform is an ideal rigid body and do not consider the dynamic deflection deformation of the long flexible boom under inertial forces and wind loads. This transient elastic deformation can negate the original physical safety buffer distance, leading to the possibility of physical collisions even along theoretically safe paths. Risk of instability and rollover: Existing systems sever the coupling between spatial geometric obstacle avoidance and vehicle dynamic anti-rollover. When performing obstacle avoidance actions such as acceleration, deceleration, or emergency detours, the nonlinear superposition of multiple rollover moments can easily lead to more severe vehicle rollover accidents. Lack of hard boundary guarantees: The underlying layer often uses PID or fuzzy control. These experience-based heuristic algorithms cannot transform the collision avoidance geometric distance and anti-rollover moment limit into mathematical constraints, making it difficult to provide forward-invariant absolute safety guarantees under extreme conditions. Summary of the Invention

[0004] The technical problem solved by this invention is that existing obstacle avoidance systems for aerial work platforms are unable to balance the positioning error caused by the deformation of the flexible boom with the dynamic anti-rollover safety of the entire vehicle. Traditional rigid body control does not take into account the transient deflection of the boom caused by wind load and acceleration, which is prone to collision due to end-positioning distortion. Furthermore, emergency obstacle avoidance actions can easily generate huge dynamic inertial moments, resulting in serious physical conflicts between distance collision avoidance constraints and moment anti-rollover constraints in control execution.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a real-time automatic lifting control system for collision avoidance and obstacle avoidance of aerial work platforms, comprising: The perception module is used to acquire three-dimensional point cloud data around the aerial work platform and cluster it into obstacle clusters. Discrete high-frequency noise is removed from the obstacle clusters, and the point cloud coordinate set of the physical obstacles is output. The observation module is used to solve the transient deflection displacement at the end of the boom, superimpose the transient deflection displacement with the rigid body spatial coordinates to output the absolute spatial coordinates of the platform, and calculate the height of the vehicle's center of gravity and the distance of the center of gravity offset. The constraint module is used to calculate the target Euclidean distance between the platform's absolute spatial coordinates and the point cloud coordinate set, construct the obstacle avoidance safety boundary function, and construct the anti-rollover constraint equation based on the vehicle's center of gravity height and center of gravity offset distance. The control module is used to solve the quadratic programming cost function with the objective function of tracking the preset obstacle avoidance trajectory, by taking the forward invariance constraint of the obstacle avoidance safety boundary function and the anti-rollover constraint equation as hard boundary conditions in the prediction time domain. It outputs the optimal control increment sequence and converts it into an analog electrical signal to drive the electro-hydraulic proportional actuator to perform obstacle avoidance actions.

[0006] Preferably, the sensing module includes: Acquire three-dimensional point cloud data around the aerial work platform, and calculate the first spatial Euclidean distance between each coordinate point in the three-dimensional point cloud data and its corresponding neighboring points. Cluster the coordinates of the first spatial Euclidean distance that are less than a preset distance threshold into obstacle clusters; Calculate the average distance from each coordinate point in the obstacle cluster to the remaining points in its local neighborhood and the standard deviation of the average distance; Remove coordinate points from the obstacle cluster whose average distance is greater than the upper limit of the preset information interval, and output the set of retained coordinate points as the point cloud coordinate set of the physical obstacle.

[0007] Preferably, the observation module includes: Obtain the first N natural frequencies of the boom of the aerial work platform and their corresponding modal damping ratios, and construct a reduced-order state-space model of the boom dynamics. The equivalent total mass of the boom, windward area, traveling acceleration, wind load dynamic pressure, and boom elevation angle of the aerial work platform are obtained. The traveling acceleration and the equivalent total mass of the boom are combined to convert into the inertial excitation force corresponding to the reduced-order state space model of boom dynamics. The wind load dynamic pressure and the windward area are combined to convert into the wind load excitation force corresponding to the reduced-order state space model of boom dynamics. The inertial excitation force and the wind load excitation force are input into the reduced-order state space model of boom dynamics as generalized excitation force input terms.

[0008] Preferably, the discrete time step of the current control cycle is obtained, and the implicit trapezoidal integral algorithm is used to iteratively solve the reduced-order state-space model of boom dynamics containing the generalized excitation force input term. The state vector at the current moment is updated, and the spatial displacement state quantity corresponding to the boom end is extracted from the state vector and output as the transient deflection displacement of the boom end. The absolute position coordinates of the chassis, the boom extension length, and the boom rotation angle of the aerial work platform are obtained. A spatial kinematics forward solution matrix is ​​constructed. The absolute position coordinates of the chassis are substituted into the spatial kinematics forward solution matrix for coordinate transformation. The rigid body spatial coordinates of the boom end are calculated. The rigid body spatial coordinates are then vector-superimposed with the transient deflection displacement to obtain the absolute spatial coordinates of the aerial work platform in the global absolute reference frame.

[0009] Preferably, the total mass of the aerial work platform, the chassis mass, the global center of gravity coordinates of the chassis, the boom mass, the local center of gravity coordinates of the boom, the work platform mass, and the local center of gravity coordinates of the work platform are obtained; Substitute the local center of gravity coordinates of the boom and the local center of gravity coordinates of the working platform into the spatial geometric rotation matrix to perform coordinate transformation, and obtain the global center of gravity coordinates of the boom and the global center of gravity coordinates of the working platform.

[0010] Preferably, the chassis mass is multiplied by the vertical component of the chassis global center of gravity coordinate to obtain the chassis boom vertical mass moment, the boom mass is multiplied by the vertical component of the boom global center of gravity coordinate to obtain the boom vertical mass moment, and the work platform mass is multiplied by the vertical component of the work platform global center of gravity coordinate to obtain the work platform vertical mass moment. The vertical mass moment of the chassis boom, the vertical mass moment of the boom, and the vertical mass moment of the working platform are added together to obtain the total vertical mass moment. The total vertical mass moment is then divided by the total mass of the aerial work platform to obtain the height of the vehicle's center of gravity.

[0011] Preferably, the chassis mass is multiplied by the horizontal component of the chassis global center of gravity coordinate to obtain the chassis horizontal mass moment, the boom mass is multiplied by the horizontal component of the boom global center of gravity coordinate to obtain the boom horizontal mass moment, and the work platform mass is multiplied by the horizontal component of the work platform global center of gravity coordinate to obtain the work platform horizontal mass moment. The horizontal mass moment of the chassis, the horizontal mass moment of the boom, and the horizontal mass moment of the work platform are added together to obtain the total horizontal mass moment. The total horizontal mass moment is then divided by the total mass of the aerial work platform to obtain the center of gravity offset distance.

[0012] Preferably, the constraint module is used to construct the obstacle avoidance safety boundary function and the real-time overturning moment function, specifically including: Traverse each coordinate point in the point cloud coordinate set, calculate the second spatial Euclidean distance between the platform's absolute spatial coordinates and the coordinate point, and extract the minimum value from the second spatial Euclidean distance as the target Euclidean distance; Calculate the Euclidean norm value of the transient deflection displacement vector at the end of the boom, and add the Euclidean norm value to the preset rigid physical tolerance radius to obtain the dynamic tolerance radius. Subtract the dynamic tolerance radius from the target Euclidean distance to obtain the obstacle avoidance safety boundary function; Obtain the total mass, local gravitational acceleration, windward area, and wind force coefficient of the aerial work platform; The horizontal distance from the tipping support edge of the aerial work platform to the geometric center of the chassis is obtained as the reference support half-width. The center of gravity offset distance is subtracted from the reference support half-width to obtain the effective anti-overturning lever arm from the center of gravity to the tipping support edge. The total mass of the aerial work platform, the local gravitational acceleration, and the effective anti-overturning lever arm are multiplied to obtain the anti-overturning stabilizing moment. Multiply the total mass of the aerial work platform, the traveling acceleration, and the height of the vehicle's center of gravity to obtain the dynamic inertial tipping term; The wind pressure center height on the windward side is calculated based on the boom elevation angle and boom extension length. The wind force coefficient, wind load dynamic pressure, windward area and wind pressure center height are multiplied to obtain the wind load overturning term. The dynamic inertial overturning term is added to the wind load overturning term to obtain the dynamic overturning moment, and an anti-overturning constraint equation is constructed with the condition that the dynamic overturning moment is less than the anti-overturning stabilizing moment.

[0013] Preferably, the control module includes: Obtain the preset obstacle avoidance trajectory and construct a quadratic programming cost function with the objective of minimizing the tracking error between the predicted trajectory of the aerial work platform and the preset obstacle avoidance trajectory; Obtain the preset prediction time domain, and within the preset prediction time domain, limit the value of the obstacle avoidance safety boundary function to be greater than or equal to zero to obtain the forward invariance constraint; Within the preset prediction time domain, the anti-rollover constraint equation is used as the anti-rollover constraint; The forward invariance constraint is combined with the anti-rollover constraint to generate joint hard boundary conditions.

[0014] Preferably, under the constraints of the joint hard boundary conditions, an extreme value solution algorithm is used to minimize the quadratic programming cost function and output the optimal control increment sequence; Extract the first control value from the optimal control increment sequence and convert the first control value into an analog electrical signal; The analog electrical signal is input into the electro-hydraulic proportional actuator of the aerial work platform to adjust the hydraulic flow and perform collision avoidance, obstacle avoidance, and anti-tipping actions.

[0015] The beneficial effects of this invention are as follows: By constructing a reduced-order state-space model of boom dynamics, the transient deflection displacement at the boom end is solved, and the transient deflection displacement is superimposed with the rigid body space coordinates to obtain the true absolute space coordinates of the platform. In order to eliminate the physical conflict between distance and torque in the underlying optimization algorithm, the absolute value of the transient deflection displacement is added to the preset rigid physical tolerance radius to form a dynamic tolerance radius, which is used to compensate for the spatial distance error caused by boom sway. Then, the forward invariance constraint of the obstacle avoidance safety boundary function containing the dynamic tolerance radius and the anti-rollover constraint of the real-time rollover torque function that integrates the static gravity rollover term, the dynamic inertial rollover term and the wind load rollover term are merged to generate a joint hard boundary condition. Finally, under the constraint of the joint hard boundary condition, the extreme value of the quadratic programming cost function is solved, the optimal control increment sequence is output and the electro-hydraulic proportional actuator is driven. This achieves that within the same control cycle, it can accurately supplement the anti-collision safety margin required for the flexible deformation of the long boom, and reduce the risk of secondary instability and rollover of the whole vehicle caused by emergency avoidance actions from the control source. Attached Figure Description

[0016] Figure 1 This is a basic flowchart of a real-time control system for automatic lifting and lowering of an aerial work platform to prevent collisions and obstacles, provided as an embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In this application, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0020] Example, refer to Figure 1A collision avoidance and obstacle avoidance automatic lifting real-time control system for aerial work platforms is provided, including: The perception module is used to acquire three-dimensional point cloud data around the aerial work platform and cluster it into obstacle clusters. Discrete high-frequency noise is removed from the obstacle clusters, and the point cloud coordinate set of the physical obstacles is output.

[0021] The perception module is used to acquire 3D point cloud data around the aerial work platform, and to perform obstacle clustering and noise removal, specifically including: Acquire 3D point cloud data around the aerial work platform and calculate the first spatial Euclidean distance between each coordinate point in the 3D point cloud data and its neighboring points.

[0022] The coordinates of points whose Euclidean distance in the first space is less than a preset distance threshold are clustered into the same obstacle cluster, thus completing the spatial segmentation of the point cloud data.

[0023] In one specific embodiment, obstacle clustering and segmentation based on radius search is used to obtain initial 3D point cloud data around the aerial work platform. For one coordinate point in the initial 3D point cloud data, a spatial neighborhood search is performed within a preset clustering search radius to find all neighboring points within the clustering search radius, and the first spatial Euclidean distance between the coordinate point and these neighboring points is calculated for each.

[0024] If the Euclidean distance in the first space is less than a preset distance threshold, the coordinate point and its corresponding neighboring points are determined to belong to the same entity. A region growing strategy is used to traverse all coordinate points, aggregating interconnected coordinate points into multiple independent obstacle clusters, thus completing the spatial segmentation of the point cloud data.

[0025] The clustering search radius is used to define the computational boundary of the spatial indexing algorithm and control the system's computing power consumption. Its value is preset based on the upper limit of the system's real-time computing power. The preset distance threshold is obtained offline based on the three-dimensional spatial angular resolution of the equipped LiDAR and the minimum physical size of the target obstacle. To avoid spatial search truncation leading to decision failure, the preset distance threshold is less than or equal to the clustering search radius. In a preferred embodiment, the preferred range for the preset clustering search radius and the preset distance threshold is 0.1 meters to 0.3 meters.

[0026] Calculate the average distance from each coordinate point in each obstacle cluster to the remaining points in the local neighborhood of each coordinate point, and solve for the standard deviation of the average distance.

[0027] In one specific embodiment, statistical feature calculation based on K-nearest neighbor search is used to perform statistical filtering for noise reduction on each obstacle cluster obtained from the segmentation. For a single coordinate point within the cluster... Set the number of local neighborhood search points , The value ranges from 20 to 50. The distance to this coordinate point is searched using the K-nearest neighbor algorithm. Recent A set of neighborhood points Calculate the coordinates of the points. To this set Inside The average distance between neighboring points , The mathematical expression is: ; in, For index The coordinates of the point.

[0028] After traversing all coordinate points within the obstacle cluster, calculate the global mean of the average distance between all coordinate points within the obstacle cluster. and the standard deviation of the average distance .

[0029] Points with an average distance greater than the upper limit of a preset signal interval are identified as discrete high-frequency noise and removed from the corresponding obstacle clusters.

[0030] The target obstacle clusters after noise removal are retained, and the set of all retained coordinate points is output as the point cloud coordinate set of the entity obstacle.

[0031] In one specific embodiment, a pre-set standard deviation factor is obtained. , The value range is from 1.0 to 2.0, and a pre-set upper limit for the confidence interval is constructed. , The mathematical expression is: ; The average distance corresponding to each coordinate point within the obstacle cluster Match each one with the preset upper limit of the information interval Perform a comparison. Determine if the average distance meets the requirements. The coordinates of the obstacles are discrete high-frequency noise generated by environmental interference, which is forcibly removed from the corresponding obstacle clusters. Finally, the target obstacle clusters after noise removal are retained, and the set of all retained coordinates is output as the point cloud coordinate set of the physical obstacles.

[0032] The observation module is used to solve the transient deflection displacement at the end of the boom, superimpose the transient deflection displacement with the rigid body spatial coordinates to output the absolute spatial coordinates of the platform, and calculate the height of the vehicle's center of gravity and the distance of the center of gravity offset.

[0033] The observation module includes: The first N natural frequencies and corresponding modal damping ratios of the boom of an aerial work platform, pre-calibrated offline, are obtained to construct a reduced-order state-space model of boom dynamics based on the modal truncation principle. By retaining the dominant low-frequency modes and truncating the high-frequency modes that contribute little to the overall displacement, a finite-dimensional reduced-order state-space model of boom dynamics can be obtained, which can be used to accurately describe the elastic vibration characteristics of a long flexible boom.

[0034] The equivalent total mass of the boom, windward area, traveling acceleration, wind load dynamic pressure, and boom elevation angle of the aerial work platform are obtained. The traveling acceleration and the equivalent total mass of the boom are combined to transform into the inertial excitation force corresponding to the reduced-order state space model of boom dynamics. The wind load dynamic pressure and the windward area are combined to transform into the wind load excitation force corresponding to the reduced-order state space model of boom dynamics. The inertial excitation force and the wind load excitation force are used together as generalized excitation force input terms and input into the reduced-order state space model of boom dynamics.

[0035] Furthermore, the inertial excitation force is obtained by multiplying the walking acceleration by the equivalent total mass of the boom. The wind load dynamic pressure, the windward area, and the preset shape coefficient are then multiplied to obtain the wind load excitation force acting vertically on the side of the boom. Based on the boom's structural type, the corresponding preset shape coefficient is selected by consulting engineering load design specifications. For the truss structure commonly found in aerial work platform booms, the preferred value range for the preset shape coefficient in this application is 1.0 to 1.4.

[0036] The discrete time step of the current control cycle is obtained, and the implicit trapezoidal integral algorithm is used to iteratively solve the reduced-order state-space model of boom dynamics containing the generalized excitation force input. The state vector at the current moment is updated, the spatial displacement state quantity corresponding to the physical node at the end of the boom is extracted from the state vector, and the spatial displacement state quantity is output as the transient deflection displacement at the end of the boom.

[0037] In one specific embodiment, the real-time control cycle of the aerial work platform controller is obtained as the discrete time step. The preferred range is 5ms to 10ms. The continuous-time state equation of the reduced-order state-space model of boom dynamics is expressed as: ; in, It is a state vector in the continuous time domain that includes modal coordinates and their first derivatives. The system state matrix, For the input matrix, It is a generalized excitation force input vector that includes inertial excitation force and wind load excitation force.

[0038] An implicit trapezoidal integral algorithm is used to discretize the continuous-time state equation. The discretization recursive formula of the implicit trapezoidal integral algorithm is expressed as: ; in, For the first The state vector at each discrete sampling time, where m is the iteration number of the current discrete control time. For the first The state vector at each discrete sampling time. The numerical damping parameter is preferably taken in the range of 0.5 to 0.8, which is used to balance the solution accuracy and algorithm stability. For the first The first derivative of the state vector at each discrete sampling time with respect to time. For the first The first derivative of the state vector at each discrete sampling time with respect to time. Obtained by solving implicit equations. The integral weighting parameter is preferably set between 0.5 and 1.0, and is used to balance the accuracy of numerical solution with the system's rigidity suppression capability.

[0039] The state vector is updated in real time through iterative solution. Spatial displacement state variables corresponding to the physical nodes at the boom end are extracted from the updated state vector and output as the transient deflection displacement of the boom end. The multibody dynamics equations of the flexible boom of an aerial work platform exhibit typical rigid characteristics. Conventional explicit integration algorithms, such as the fourth-order Runge-Kutta method, are prone to inducing high-frequency numerical oscillations or even divergence under the large discrete time steps commonly used in engineering. The implicit trapezoidal integration algorithm, by virtue of the unconditional stability of the rigid system, effectively filters out high-frequency truncation noise while ensuring strict convergence of the elastic deformation response, providing high-precision dynamic displacement data for subsequent solutions.

[0040] The absolute position coordinates of the chassis, the boom extension length, and the boom slewing angle of the aerial work platform are obtained. Based on the boom elevation angle, boom extension length, and boom slewing angle, a spatial kinematics forward solution matrix is ​​constructed. The absolute position coordinates of the chassis are substituted into the spatial kinematics forward solution matrix for coordinate transformation, and the rigid body spatial coordinates of the boom end are calculated. The rigid body spatial coordinates are then superimposed with the transient deflection displacement in three dimensions to obtain the actual three-dimensional coordinates, which are then output as the absolute spatial coordinates of the aerial work platform in the global absolute reference frame.

[0041] A two-dimensional rotation matrix in the luffing plane is constructed based on the boom elevation angle, and a rotation matrix in the slewing plane is constructed based on the boom rotation angle. Multiplying the two-dimensional rotation matrix and the rotation matrix in the slewing plane yields a three-dimensional rotation matrix describing the boom's spatial attitude. Based on the kinematic chain sequence of the base, rotation, and amplitude, a three-dimensional rotation matrix is ​​combined with the absolute position coordinates of the chassis to construct a... Spatial kinematics forward solution matrix The expression for the forward kinematics matrix is: ; in, A three-dimensional rotation matrix The corresponding element, This is the translation vector formed by the absolute position coordinates of the chassis. The boom extension length L is converted into a local homogeneous coordinate vector along the boom axis, set as... The local homogeneous coordinate vector is multiplied by the forward kinematics matrix to calculate the rigid body spatial coordinates at the boom end. These rigid body spatial coordinates are then superimposed with the transient deflection displacement in a three-dimensional vector operation to obtain the absolute spatial coordinates of the aerial work platform in the global absolute reference frame.

[0042] Obtain the total mass of the aerial work platform, chassis mass, global center of gravity coordinates of the chassis, boom mass, local center of gravity coordinates of the boom, work platform mass, and local center of gravity coordinates of the work platform.

[0043] A spatial geometric rotation matrix is ​​established based on the boom elevation angle. The local center of gravity coordinates of the boom are substituted into the spatial geometric rotation matrix for multiplication and mapping operations, and the output is the global center of gravity coordinates of the boom corresponding to the global absolute reference frame. The local center of gravity coordinates of the working platform are substituted into the spatial geometric rotation matrix for multiplication and mapping operations, and the output is the global center of gravity coordinates of the working platform corresponding to the global absolute reference frame.

[0044] The vertical mass moment of the chassis boom is obtained by multiplying the chassis mass by the vertical component of the chassis's global center of gravity coordinates. The vertical mass moment of the boom is obtained by multiplying the boom mass by the vertical component of the boom's global center of gravity coordinates. The vertical mass moment of the work platform is obtained by multiplying the work platform mass by the vertical component of the work platform's global center of gravity coordinates.

[0045] The vertical mass moment of the chassis boom, the vertical mass moment of the boom, and the vertical mass moment of the work platform are added together to obtain the total vertical mass moment. The total vertical mass moment is divided by the total mass of the aerial work platform to obtain the weighted average value of the vertical position. The weighted average value of the vertical position is used as the height of the vehicle's center of gravity.

[0046] The chassis horizontal mass moment is obtained by multiplying the chassis mass by the horizontal component of the chassis's global center of gravity coordinates. The boom horizontal mass moment is obtained by multiplying the boom mass by the horizontal component of the boom's global center of gravity coordinates. The work platform horizontal mass moment is obtained by multiplying the work platform mass by the horizontal component of the work platform's global center of gravity coordinates.

[0047] The horizontal mass moment of the chassis, the horizontal mass moment of the boom, and the horizontal mass moment of the work platform are added together to obtain the total horizontal mass moment. The total horizontal mass moment is then divided by the total mass of the aerial work platform to obtain the weighted average value of the horizontal position. This weighted average value of the horizontal position is used as the center of gravity offset distance. Specifically: Along the X-axis and Y-axis of the global coordinate system, the horizontal mass moments of the chassis, boom, and work platform are added together to obtain the total horizontal mass moment in the X-axis and Y-axis directions, respectively. These are then divided by the total mass of the aerial work platform to obtain the X-axis and Y-axis coordinate components of the vehicle's center of gravity. Determine the normal direction of the most unfavorable overturning support edge under the current working condition, project the horizontal coordinate of the vehicle's center of gravity onto this normal direction, and obtain the vertical offset distance of the center of gravity relative to the overturning support edge. Use this vertical offset distance as the center of gravity offset distance.

[0048] The constraint module is used to calculate the target Euclidean distance between the platform's absolute spatial coordinates and the point cloud coordinate set, construct the obstacle avoidance safety boundary function, and construct the anti-rollover constraint equation based on the vehicle's center of gravity height and center of gravity offset distance.

[0049] The control module, within the prediction time domain, uses the forward invariance constraints of the obstacle avoidance safety boundary function and the anti-tipping constraint equations as hard boundary conditions to solve a quadratic programming cost function with the objective function of tracking a preset obstacle avoidance trajectory. It outputs the optimal control increment sequence and converts it into analog electrical signals to drive the electro-hydraulic proportional actuator to perform obstacle avoidance actions. The constraint module is used to construct the obstacle avoidance safety boundary function and the real-time tipping moment function, specifically including: Iterate through each coordinate point in the point cloud coordinate set, calculate the second space Euclidean distance between the platform's absolute space coordinates and each coordinate point, extract the minimum value from the second space Euclidean distance, and use it as the target Euclidean distance.

[0050] Obtain the preset rigid physical tolerance radius representing the physical safety margin of the aerial work platform foundation, calculate the Euclidean norm value of the transient deflection displacement vector at the end of the boom, and add the Euclidean norm value to the preset rigid physical tolerance radius to obtain the dynamic tolerance radius.

[0051] The preset rigid physical tolerance radius is determined based on the external dimensions of the aerial work platform and the measurement error of the sensors.

[0052] Subtracting the dynamic tolerance radius from the target Euclidean distance yields the obstacle avoidance safety boundary function.

[0053] When the obstacle avoidance safety boundary function value > 0, it indicates that the platform is in a safe obstacle avoidance zone; when the obstacle avoidance safety boundary function value > 0, it indicates that the platform is in a safe obstacle avoidance zone. When the value is zero, it indicates a collision risk. The obstacle avoidance safety boundary function dynamically scales the safety buffer by introducing the absolute value of transient deflection displacement, thus ensuring the absolute physical safety distance for obstacle avoidance calculations even when the boom experiences elastic vibration.

[0054] Obtain the total mass of the aerial work platform, local gravitational acceleration, windward area, and wind force coefficient.

[0055] The horizontal distance from the tipping support edge of the aerial work platform to the geometric center of the chassis is obtained as the reference support half-width. The center of gravity offset distance is subtracted from the reference support half-width to obtain the effective anti-tipping lever arm from the center of gravity to the tipping support edge. The total mass of the aerial work platform, the local gravitational acceleration, and the effective anti-tipping lever arm are multiplied to obtain the anti-tipping stabilizing moment.

[0056] Multiplying the total mass of the aerial work platform, its traveling acceleration, and the height of its center of gravity yields the dynamic inertial tilting term. The inertial force acts on the center of gravity, with the lever arm being the height of the center of gravity. The dynamic inertial tilting term quantifies the dynamic instability torque caused by the start and stop of travel.

[0057] The wind pressure center height on the windward side is calculated based on the boom elevation angle and boom extension length. The wind load overturning term is obtained by multiplying the wind force coefficient, wind load dynamic pressure, windward area, and wind pressure center height. The wind pressure center height is taken as the height of the spatial geometric center of the boom and working platform in the deployed state.

[0058] The dynamic inertial overturning term is added to the wind load overturning term to obtain the dynamic overturning moment, and an anti-overturning constraint equation is constructed with the condition that the dynamic overturning moment is less than the anti-overturning stabilizing moment.

[0059] The anti-tipping constraint equations, in terms of physical logic, explicitly classify gravity as a beneficial force that provides stability margin, while classifying inertia and wind load as harmful forces that deplete stability margin. Compared to simple moment superposition, the anti-tipping constraint equations are more rigorous in the engineering mechanics paradigm and introduce the wind pressure center height as the actual lever arm, which can accurately reflect the dynamic anti-tipping safety boundary under the working conditions of high-altitude long boom scaffolding.

[0060] The control module is used to solve the quadratic programming cost function with the objective function of tracking the preset obstacle avoidance trajectory, by taking the forward invariance constraint of the obstacle avoidance safety boundary function and the anti-rollover constraint equation as hard boundary conditions in the prediction time domain. It outputs the optimal control increment sequence and converts it into an analog electrical signal to drive the electro-hydraulic proportional actuator to perform obstacle avoidance actions.

[0061] The control module includes: Obtain the preset obstacle avoidance trajectory and construct a quadratic programming cost function with the objective of minimizing the tracking error between the predicted trajectory of the aerial work platform and the preset obstacle avoidance trajectory.

[0062] A preset prediction time domain is obtained. Within this domain, the value of the obstacle avoidance safety boundary function is limited to be greater than or equal to zero, resulting in forward invariance constraints. These forward invariance constraints utilize control obstacle function theory to transform the spatial geometric collision avoidance distance into a hard boundary in mathematical programming, ensuring that the system state does not exceed the dynamic safety envelope boundary during its forward rolling evolution.

[0063] Within the preset prediction time domain, the anti-rollover constraint equation is used as the anti-rollover constraint.

[0064] By combining the forward invariance constraint and the anti-rollover constraint, joint hard boundary conditions are generated.

[0065] Based on the current discrete-time state, the trajectory of the aerial work platform within a future time window is predicted. The goal is to minimize the tracking error between the predicted trajectory and the preset obstacle avoidance trajectory, and to smooth the control actions. A quadratic programming cost function is constructed, and its mathematical expression is: ; in, To predict the number of forward prediction steps in the time domain, For the first Predicted spatial trajectory of aerial work platform For the first The preset obstacle avoidance trajectory of the step, For the first The control increment sequence of the step. For tracking error positive definite weight matrix, To control the incremental positive definite weight matrix.

[0066] Under the constraints of joint hard boundary conditions, an extreme value solution algorithm is used to minimize the quadratic programming cost function, outputting a control sequence that minimizes the quadratic programming cost function. This control sequence is then used as the optimal control increment sequence. The solution process involves finding a set of values ​​that minimizes the quadratic programming cost function while satisfying all space collision avoidance and anti-rollover safety constraints. The smallest control sequence is used as the optimal control increment sequence.

[0067] Extract the first control value from the optimal control increment sequence and convert the first control value into an analog electrical signal.

[0068] The rolling execution mechanism based on model predictive control extracts only the first control value from the optimal control increment sequence and superimposes it onto the physical control quantity of the previous moment to obtain the final control command for the current moment. The final control command is then converted into an analog electrical signal, such as a 0-10V voltage signal or a 4-20mA industrial standard current signal, through the built-in digital-to-analog converter circuit of the aerial work platform controller.

[0069] The analog electrical signal is input to the electro-hydraulic proportional actuator of the aerial work platform, which drives the electro-hydraulic proportional valve to adjust the opening and flow of the hydraulic oil circuit, and performs anti-collision and obstacle avoidance and anti-tipping actions.

[0070] To address the technical challenges of existing aerial work platform obstacle avoidance systems, which fail to consider end-effector positioning distortion due to flexible boom deformation and are prone to generating massive dynamic inertial moments during emergency obstacle avoidance, potentially leading to vehicle instability and rollover, this application constructs a reduced-order state-space model of boom dynamics to solve for the transient deflection displacement at the boom end. This transient deflection displacement is then superimposed with rigid body spatial coordinates to obtain the true absolute spatial coordinates of the platform. To eliminate the physical conflict between distance and torque in the underlying optimization algorithm, this application adds the absolute value of the transient deflection displacement to a preset rigid physical tolerance radius to construct a dynamic tolerance radius, which compensates for spatial distance errors caused by boom swaying. Furthermore, the forward invariance constraint of the obstacle avoidance safety boundary function, which includes the dynamic tolerance radius, and the anti-rollover constraint of the real-time rollover moment function, which integrates the static gravity rollover term, the dynamic inertial rollover term, and the wind load rollover term, are combined to generate a joint hard boundary condition. Finally, under the constraint of the joint hard boundary condition, the extreme value of the quadratic programming cost function is solved, the optimal control increment sequence is output, and the electro-hydraulic proportional actuator is driven. This achieves the ability to accurately supplement the anti-collision safety margin required for the flexible deformation of the long boom within the same control cycle, and to reduce the risk of secondary instability and rollover of the whole vehicle caused by emergency avoidance actions from the control source.

[0071] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0072] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the principles of this invention should be included within the scope of protection of this invention.

Claims

1. A real-time automatic lifting control system for collision avoidance and obstacle avoidance of aerial work platforms, characterized in that, include: The perception module is used to acquire three-dimensional point cloud data around the aerial work platform and cluster it into obstacle clusters. Discrete high-frequency noise is removed from the obstacle clusters, and the point cloud coordinate set of the physical obstacles is output. The observation module is used to solve the transient deflection displacement at the end of the boom, superimpose the transient deflection displacement with the rigid body spatial coordinates to output the absolute spatial coordinates of the platform, and calculate the height of the vehicle's center of gravity and the distance of the center of gravity offset. The constraint module is used to calculate the target Euclidean distance between the platform's absolute spatial coordinates and the point cloud coordinate set, construct the obstacle avoidance safety boundary function, and construct the anti-rollover constraint equation based on the vehicle's center of gravity height and center of gravity offset distance. The control module is used to solve the quadratic programming cost function with the objective function of tracking the preset obstacle avoidance trajectory, by taking the forward invariance constraint of the obstacle avoidance safety boundary function and the anti-rollover constraint equation as hard boundary conditions in the prediction time domain. It outputs the optimal control increment sequence and converts it into an analog electrical signal to drive the electro-hydraulic proportional actuator to perform obstacle avoidance actions.

2. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 1, characterized in that, The sensing module includes: Acquire three-dimensional point cloud data around the aerial work platform, and calculate the first spatial Euclidean distance between each coordinate point in the three-dimensional point cloud data and its corresponding neighboring points. Cluster the coordinates of the first spatial Euclidean distance that are less than a preset distance threshold into obstacle clusters; Calculate the average distance from each coordinate point in the obstacle cluster to the remaining points in its local neighborhood and the standard deviation of the average distance; Remove coordinate points from the obstacle cluster whose average distance is greater than the upper limit of the preset information interval, and output the set of retained coordinate points as the point cloud coordinate set of the physical obstacle.

3. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 2, characterized in that, The observation module includes: Obtain the first N natural frequencies of the boom of the aerial work platform and their corresponding modal damping ratios, and construct a reduced-order state-space model of the boom dynamics. The equivalent total mass of the boom, windward area, traveling acceleration, wind load dynamic pressure, and boom elevation angle of the aerial work platform are obtained. The traveling acceleration and the equivalent total mass of the boom are combined to convert into the inertial excitation force corresponding to the reduced-order state space model of boom dynamics. The wind load dynamic pressure and the windward area are combined to convert into the wind load excitation force corresponding to the reduced-order state space model of boom dynamics. The inertial excitation force and the wind load excitation force are input into the reduced-order state space model of boom dynamics as generalized excitation force input terms.

4. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 3, characterized in that, The discrete time step of the current control cycle is obtained, and the implicit trapezoidal integral algorithm is used to iteratively solve the reduced-order state-space model of boom dynamics containing the generalized excitation force input term. The state vector at the current moment is updated, and the spatial displacement state quantity corresponding to the boom end is extracted from the state vector and output as the transient deflection displacement of the boom end. The absolute position coordinates of the chassis, the boom extension length, and the boom rotation angle of the aerial work platform are obtained. A spatial kinematics forward solution matrix is ​​constructed. The absolute position coordinates of the chassis are substituted into the spatial kinematics forward solution matrix for coordinate transformation. The rigid body spatial coordinates of the boom end are calculated. The rigid body spatial coordinates are then vector-superimposed with the transient deflection displacement to obtain the absolute spatial coordinates of the aerial work platform in the global absolute reference frame.

5. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 4, characterized in that, Obtain the total mass, chassis mass, global center of gravity coordinates of the chassis, boom mass, local center of gravity coordinates of the boom, work platform mass, and local center of gravity coordinates of the work platform of the aerial work platform. Substitute the local center of gravity coordinates of the boom and the local center of gravity coordinates of the working platform into the spatial geometric rotation matrix to perform coordinate transformation, and obtain the global center of gravity coordinates of the boom and the global center of gravity coordinates of the working platform.

6. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 5, characterized in that, The chassis mass is multiplied by the vertical component of the chassis global center of gravity coordinate to obtain the chassis boom vertical mass moment; the boom mass is multiplied by the vertical component of the boom global center of gravity coordinate to obtain the boom vertical mass moment; and the work platform mass is multiplied by the vertical component of the work platform global center of gravity coordinate to obtain the work platform vertical mass moment. The vertical mass moment of the chassis boom, the vertical mass moment of the boom, and the vertical mass moment of the working platform are added together to obtain the total vertical mass moment. The total vertical mass moment is then divided by the total mass of the aerial work platform to obtain the height of the vehicle's center of gravity.

7. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 6, characterized in that, The chassis mass is multiplied by the horizontal component of the chassis global center of gravity coordinate to obtain the chassis horizontal mass moment; the boom mass is multiplied by the horizontal component of the boom global center of gravity coordinate to obtain the boom horizontal mass moment; and the work platform mass is multiplied by the horizontal component of the work platform global center of gravity coordinate to obtain the work platform horizontal mass moment. The horizontal mass moment of the chassis, the horizontal mass moment of the boom, and the horizontal mass moment of the work platform are added together to obtain the total horizontal mass moment. The total horizontal mass moment is then divided by the total mass of the aerial work platform to obtain the center of gravity offset distance.

8. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 7, characterized in that, The constraint module is used to construct the obstacle avoidance safety boundary function and the real-time overturning moment function, specifically including: Traverse each coordinate point in the point cloud coordinate set, calculate the second spatial Euclidean distance between the platform's absolute spatial coordinates and the coordinate point, and extract the minimum value from the second spatial Euclidean distance as the target Euclidean distance; Calculate the Euclidean norm value of the transient deflection displacement vector at the end of the boom, and add the Euclidean norm value to the preset rigid physical tolerance radius to obtain the dynamic tolerance radius. Subtract the dynamic tolerance radius from the target Euclidean distance to obtain the obstacle avoidance safety boundary function; Obtain the total mass, local gravitational acceleration, windward area, and wind force coefficient of the aerial work platform; The horizontal distance from the tipping support edge of the aerial work platform to the geometric center of the chassis is obtained as the reference support half-width. The center of gravity offset distance is subtracted from the reference support half-width to obtain the effective anti-overturning lever arm from the center of gravity to the tipping support edge. The total mass of the aerial work platform, the local gravitational acceleration, and the effective anti-overturning lever arm are multiplied to obtain the anti-overturning stabilizing moment. Multiply the total mass of the aerial work platform, the traveling acceleration, and the height of the vehicle's center of gravity to obtain the dynamic inertial tipping term; The wind pressure center height on the windward side is calculated based on the boom elevation angle and boom extension length. The wind force coefficient, wind load dynamic pressure, windward area and wind pressure center height are multiplied to obtain the wind load overturning term. The dynamic inertial overturning term is added to the wind load overturning term to obtain the dynamic overturning moment, and an anti-overturning constraint equation is constructed with the condition that the dynamic overturning moment is less than the anti-overturning stabilizing moment.

9. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 8, characterized in that, The control module includes: Obtain the preset obstacle avoidance trajectory and construct a quadratic programming cost function with the objective of minimizing the tracking error between the predicted trajectory of the aerial work platform and the preset obstacle avoidance trajectory; Obtain the preset prediction time domain, and within the preset prediction time domain, limit the value of the obstacle avoidance safety boundary function to be greater than or equal to zero to obtain the forward invariance constraint; Within the preset prediction time domain, the anti-rollover constraint equation is used as the anti-rollover constraint; The forward invariance constraint is combined with the anti-rollover constraint to generate joint hard boundary conditions.

10. The real-time automatic lifting control system for collision avoidance and obstacle avoidance of an aerial work platform as described in claim 9, characterized in that, Under the constraints of the joint hard boundary conditions, the quadratic programming cost function is minimized using an extreme value solution algorithm, and the optimal control increment sequence is output. Extract the first control value from the optimal control increment sequence and convert the first control value into an analog electrical signal; The analog electrical signal is input into the electro-hydraulic proportional actuator of the aerial work platform to adjust the hydraulic flow and perform collision avoidance, obstacle avoidance, and anti-tipping actions.

Images