Mechanical arm control system and control method of intelligent photovoltaic panel mounting robot

Abstract

The invention provides a mechanical arm control system and control method of a photovoltaic panel intelligent installation robot, and relates to the technical field of photovoltaic equipment installation. According to the method, a three-dimensional point cloud map is constructed, the curvature characteristics of the surface of an obstacle are recognized, meanwhile, a joint angle is fed back through a magnetic encoder, and normal pressure and tangential friction force are monitored through a sensor; when the normal pressure exceeds a threshold value, a reverse displacement compensation amount is calculated based on reverse dynamics, a compensation instruction is generated, and a safe approaching direction is determined in combination with curvature characteristics; and the joint angular velocity and the torque threshold are adjusted in real time through a nonlinear sliding mode control algorithm, a smooth joint target torque curve is generated, and it is ensured that the mechanical arm stably tracks the path and the tail end friction force is controlled. And in an abnormal condition, the system triggers an alarm and freezes the mechanical arm until the operator confirms the compensation strategy. The control precision and adaptability of the mechanical arm are improved, the photovoltaic panel is effectively prevented from being damaged, and the installation efficiency and reliability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic equipment installation, and specifically to a robotic arm control system and control method for an intelligent photovoltaic panel installation robot. Background Art

[0002] During the intelligent installation of photovoltaic panels, the precise control and reliable operation ability of the robotic arm are the core elements determining the installation efficiency and equipment safety. However, the existing technologies face multiple challenges in practical applications; it is difficult for the robotic arm to quickly sense and analyze features in special installation environments, resulting in the robotic arm being unable to dynamically adapt to the continuous changes in the environment, which easily causes trajectory deviation or mechanical collision; there are significant shortcomings in the interactive control when the end effector contacts the photovoltaic panel, and traditional methods lack the ability of real-time closed-loop regulation of the contact force, which easily causes damage to the surface of the photovoltaic panel or installation failure. Summary of the Invention

[0003] To achieve the above objectives, the present invention is realized through the following technical solutions: A robotic arm control system for an intelligent photovoltaic panel installation robot, comprising:

[0004] Sensor module: A three-dimensional force sensor detects the normal pressure and tangential friction force when the end effector contacts the photovoltaic panel; a magnetic encoder real-time feeds back joint angle data; a binocular vision sensor constructs a three-dimensional point cloud map of the photovoltaic panel installation area and identifies the surface curvature of obstacles.

[0005] Central processing module: Processes the data collected by the sensor module; calculates the safe approaching direction of the end effector of the robotic arm according to the identified surface curvature of the obstacle; compares the normal pressure with a preset compressive threshold of the photovoltaic panel, and generates a reverse pressure compensation instruction when the normal pressure exceeds the threshold; based on the joint angle data, combines the safe approaching direction and the reverse pressure compensation instruction, and generates anti-slip movement parameters of the robotic arm joints through the pseudo-inverse method of the Jacobian matrix and the dynamic damping coefficient.

[0006] Path planning module: Calculates the target movement parameters of each joint based on the anti-slip movement parameters of the robotic arm joints, and corrects the anti-jitter amount of the joint angular velocity and the dynamic suppression coefficient of the tangential friction force of the end effector according to the joint angle feedback, and generates a path planning instruction.

[0007] Drive module: A harmonic reducer directly connected to the robotic arm joint and a torque feedback servo motor. The harmonic reducer adjusts the output torque according to the path planning instruction to track the planned path. The torque feedback servo motor limits the tangential friction force based on the dynamic suppression coefficient, and at the same time adjusts the motor speed according to the trajectory of the path. Among them, a hysteresis interval of the servo motor torque threshold is constructed according to the dynamic suppression coefficient, and the threshold is tightened in the rising stage of the friction force and relaxed in the falling stage.

[0008] Human - machine interaction module: Trigger an alarm when the path - planning instruction of the robotic arm is abnormal, and receive the emergency reverse compensation amount input by the operator to overwrite the path - planning instruction.

[0009] Furthermore, the central processing module constructs a three - dimensional point - cloud map of the photovoltaic panel installation area based on the binocular vision sensor, performs spatial discretization on the point - cloud data, and divides the continuous surface into a dense set of discrete points. For each discrete point, all the point - cloud data within a neighborhood radius of 5 cm around it is extracted to construct a local neighborhood covariance matrix. Through eigenvalue decomposition, three orthogonal eigenvectors and their corresponding eigenvalues are obtained. The principal curvature is determined by the direction of the eigenvector corresponding to the minimum eigenvalue, which represents the maximum bending degree of the local surface, while the curvature direction is defined by the first - principal - component eigenvector of the covariance matrix, reflecting the main extension trend of the surface. The curvature change rate between adjacent discrete points is obtained by calculating the ratio of the difference in principal curvatures of two points to the Euclidean distance. Then, a curvature gradient field is constructed along the surface to quantify the continuous change characteristics of curvature in space. The boundary of the curvature - gradient mutation region is identified by detecting clusters of discrete points with a change rate exceeding a preset threshold in the gradient field, forming the contour of the curvature - discontinuous region. After establishing a local coordinate system with the current position of the end - effector as the origin, the global curvature gradient field is projected onto this coordinate system, and the initial approximation direction vector is decomposed into a normal component in the curvature - gradient direction and a tangential component in the tangent - plane direction. By calculating the curvature - gradient change rate in each direction within the tangent plane, the tangential component with a change rate lower than 0.1 mm -2 is selected as the candidate safe direction. Finally, the tangential path with the smallest curvature - gradient direction component and the lowest change rate is selected as the safe approach direction, enabling the end - effector to approach the photovoltaic panel along a trajectory with a gentle change in surface curvature.

[0010] Furthermore, after the central processing module receives the normal - pressure data of the end - effector detected by the three - dimensional force sensor, it calculates the deviation amount ΔP between the current pressure and the preset anti - compression threshold of the photovoltaic panel. When ΔP exceeds the safety threshold, based on inverse dynamics, the pressure difference is converted into an inverse - displacement compensation amount ΔX of the end - effector along the normal direction, generating an instruction containing the compensation direction and amplitude. To achieve the precise mapping of the end - compensation amount to the joint space, inverse kinematic solution is performed through the pseudo - inverse method of the Jacobian matrix: Based on the current joint angles θ real - time feedback by the magnetic encoder, the Jacobian matrix J of the robotic arm is constructed. J is calculated through the geometric parameters and kinematics of each joint, representing the differential sensitivity of the end - effector pose change to the joint angles. Subsequently, the pseudo - inverse matrix J + damped of the Jacobian matrix is calculated through singular - value decomposition, and a damping pseudo - inverse is constructed by introducing a dynamic damping coefficient λ to suppress numerical divergence in singular configurations, where J T is the transpose of the Jacobian matrix; through the formula , the end - reverse displacement compensation amount ΔX is mapped to the joint angle increments ΔΨ, thereby establishing the dynamic correlation between the joint angles and the end - pressure.

[0011] Furthermore, when dynamically adjusting the upper limit of the joint angular velocity, a pressure - angular velocity proportional attenuation relationship is proposed. The normal pressure change rate ΔP / Δt is calculated in real - time, and its value is the mean change amount within the current pressure deviation ΔP and the time window Δt. Subsequently, based on the initial upper limit of the angular velocity ω0 and the attenuation coefficient k calibrated through experiments, the upper limit of the joint angular velocity is dynamically corrected through ; where is the corrected upper limit of the joint angular velocity, k is determined through calibration experiments, and the value of k is adjusted in typical pressure mutation scenarios so that the joint motion can still stably track the path after deceleration and the end - friction does not exceed the limit. At the same time, combining the deviation between the actual joint angle and the target angle fed back by the magnetic encoder, an anti - jitter amount of the angular velocity is generated through sliding - mode control to suppress high - frequency oscillations. And based on the end - tangential friction data fed back by the three - dimensional force sensor, a mapping relationship between friction and torque is constructed to adjust the torque threshold of the servo motor in real - time. Finally, smooth anti - slip motion parameters are generated through multi - parameter collaborative optimization, so that the end - friction is controlled and the motion trajectory is stable during the reverse pressure compensation process of the robotic arm.

[0012] Furthermore, the path - planning module, based on the target joint angles and angular velocities in the anti - slip motion parameters of the robotic - arm joints generated by the central - processing module, calculates the target motion trajectories of each joint through the inverse kinematics of the robotic arm, generating the initial joint target angle sequence and the corresponding angular - velocity sequence. The generation of the dynamic suppression coefficient is designed with a clear friction - trend relationship. First, the end - effector tangential friction data fed back by the three - dimensional force sensor is read in real - time, and the friction change rate Δf / Δt is calculated, and its value is the difference between the current friction f t and the friction f t-1 at the previous time - stamp divided by the sampling time interval Δt. If it is detected that Δf / Δt exceeds the preset threshold three times in a row, it is determined that the friction is in an abnormal growth trend, and the dynamic suppression coefficient is triggered to increase. The update formula of the dynamic suppression coefficient C is , where C0 is the initial suppression coefficient and α is the growth factor determined through calibration experiments. The calibration method is to adjust the value of α in typical friction - mutation scenarios so that the suppression coefficient can not only quickly respond to friction over - limit but also avoid the robotic - arm motion lag caused by excessive suppression.

[0013] Furthermore, in the trajectory interpolation, the piece - wise cubic Hermite interpolation method is used to smoothly generate the joint - angle sequence. The anti - jittered joint - angular - velocity sequence and the target - angle sequence are aligned according to the time - stamp, and discrete data points (t i ,θ i ,ω i ) are extracted, where t i is the time - stamp, θi is the target joint angle, ω i is the angular velocity after anti-shake; A cubic polynomial function is constructed between every two adjacent data points, satisfying the continuity of the function value and the first derivative (angular velocity) at the interpolation points, and maintaining data monotonicity to avoid overshoot; The polynomial coefficients are determined by solving a system of boundary condition equations. For example, in the time interval [t i , t i+1 , the interpolation function H(t) needs to satisfy H(t i ) = θ i , representing the target joint angle at time t i , H(t i+1 ) = θ i+1 , representing the target joint angle at time t i+1 , H ’ (t i ) = ω i , representing the anti-shake angular velocity at time t i , H ’ (t i+1 ) = ω i+1 , representing the anti-shake angular velocity at time t i+1 ; Synchronously map the dynamic suppression coefficient C to the interpolation trajectory according to the time stamp to form a continuous path command including joint angle, angular velocity and suppression coefficient, that is, the path planning command; In the finally generated command, each time stamp is associated with a specific joint angle value H(t), an angular velocity limit H ’ (t) and a torque suppression threshold C(t), which are output to the harmonic reducer and the torque feedback servo motor of the drive module, controlling the harmonic reducer to adjust the output torque according to the interpolation trajectory, and at the same time the servo motor based on C(t) limits the tangential friction force in real time to ensure motion smoothness and end stability.

[0014] Furthermore, the harmonic reducer in the drive module receives the joint angle sequence and the corresponding time stamp in the path planning command, parses the target joint angle value of the current time stamp in real time, combines the actual joint angle fed back by the magnetic encoder, calculates the angle deviation and determines the adjustment amount of the output torque corresponding to the preset torque and angle deviation, driving the harmonic reducer to output a torque matching the target angle to track the planned path; The torque feedback servo motor reads the dynamic suppression coefficient and the angular velocity limit in the path planning command, and dynamically adjusts the motor speed according to the angular velocity limit of the current time stamp to synchronize it with the angular velocity demand of the path trajectory; During the adjustment process, based on the dynamic suppression coefficient and the end tangential friction force data detected in real time, the output torque of the servo motor is limited within the threshold range corresponding to the suppression coefficient through a torque control algorithm. When the friction force exceeds the threshold, the output torque is reduced according to the suppression coefficient ratio to limit the tangential friction force.

[0015] Further, after the harmonic reducer in the drive module receives the joint angle sequence and the corresponding timestamps in the path planning instruction, it parses the target joint angle value θ at the current timestamp target , combines it with the actual joint angle θ fed back by the magnetic encoder actual , and calculates the angle deviation ; Based on the preset torque and angle deviation, through , where Δτ is the output torque adjustment amount, K p , K i , K d are control parameters calibrated based on torque and angle by the harmonic reducer, ∫Δθdt is the integral term of the angle deviation, used to eliminate the steady-state error; dΔθ / dt is the differential term of the angle deviation, used to predict the change trend; the calibration method is that under no-load and load conditions, a stepped torque is applied to the harmonic reducer and the angle response curve is recorded, and the proportional coefficient K p , integral coefficient K i , and differential coefficient K d are obtained by least squares fitting to make the linear response relationship between Δτ and Δθ; drive the harmonic reducer to output the adjusted torque τ = τ prev + Δτ, so that the actual angle tracks the target angle, where τ prev represents the output torque at the previous moment.

[0016] Further, the torque feedback servo motor synchronously reads the dynamic suppression coefficient C and the angular velocity limit value in the path planning instruction, and adjusts the motor speed according to ω at the current timestamp max to synchronize it with the trajectory angular velocity requirement; during the torque control process, the dynamic suppression coefficient C is used for constraint. When C increases, the proportional gain K p new = K p / (1 + β·C), where K p new represents the corrected proportional gain, and β is the attenuation factor calibrated through the friction overrun experiment, used to balance the torque response speed and stability; at the same time, based on the end tangential friction force f fed back by the three-dimensional force sensor t , the current torque threshold τ limit is calculated through the mapping relationship, specifically , where κ is the torque-friction conversion coefficient, f max is the preset safety threshold, and C is the dynamic suppression coefficient; if the friction force f t exceeds the threshold, then the output torque is limited according to to avoid the out-of-control of the tangential friction force, where τ demand represents the required torque calculated according to the motion requirements of the robotic arm, and τ limitIt represents the torque threshold calculated based on the dynamic suppression coefficient and the tangential frictional force data detected in real time; the harmonic reducer adjusts the torque to drive the joint movement, and the servo motor dynamically adjusts the output according to the torque threshold, so that while the robotic arm tracks the path, the tangential frictional force at the end is strictly controlled by the dynamic suppression coefficient; among them, according to the real-time value of the dynamic suppression coefficient C, a hysteresis interval of the servo motor torque threshold is constructed. During the rising stage of the frictional force, the threshold is tightened, and during the falling stage, the threshold is relaxed, avoiding frequent torque jumps caused by traditional fixed thresholds.

[0017] A control method for the robotic arm of a photovoltaic panel intelligent installation robot, including:

[0018] S1 Scanning the installation area of the photovoltaic panel through a binocular vision sensor to extract the surface curvature characteristics of the obstacle, and at the same time, the magnetic encoder real-time feedbacks the joint rotation angle of the robotic arm, and the normal pressure and tangential frictional force of the end effector are obtained in real time through a three-dimensional force sensor;

[0019] S2 According to the surface curvature characteristics of the obstacle, calculating the approaching path of the end effector of the robotic arm along the normal direction of the photovoltaic panel surface, and generating a safe approaching angle;

[0020] S3 Generating a path planning instruction through dynamic pressure compensation and anti-slip movement parameters, and the specific steps include:

[0021] S31 When the normal pressure exceeds the preset anti-compression threshold of the photovoltaic panel, calculating the reverse compensation pressure according to the over-limit ratio, and generating a reverse pressure compensation instruction;

[0022] S32 Based on the safe approaching angle and the reverse pressure compensation instruction, combining the joint angle data of the magnetic encoder to generate the anti-slip movement parameters of the robotic arm joints;

[0023] S33 Using a non-linear sliding mode control algorithm to calculate the target movement parameters of the robotic arm joints, including the anti-vibration amount and the dynamic suppression coefficient of the tangential frictional force, and generating a path planning instruction;

[0024] S4 Driving the servo motor to rotate according to the path planning instruction torque feedback, controlling the robotic arm to move along the planned path, and limiting the tangential frictional force within the preset safe range;

[0025] S5 When the normal pressure continuously exceeds the set number of times, the tangential frictional force exceeds the preset safe range, or the pose deviation of the end effector exceeds the allowable tolerance, and the deviation between the actual torque and the target torque of the robotic arm joint continuously exceeds the dynamic threshold, triggering an alarm and freezing the movement of the robotic arm until the operator confirms the compensation strategy through the human-machine safety interaction module.

[0026] Furthermore, in the design of the non-linear sliding mode control algorithm, based on the actual joint angle feedback from the magnetic encoder and the target angle generated by the path planning module, the system defines the tracking error of the joint angle as the difference between the actual angle and the target angle, and further calculates its time derivative as the angular velocity error. The sliding mode surface function is constructed by linearly combining the angle error and the angular velocity error according to a specific weight ratio, where the weight coefficient of the angular velocity error is dynamically determined by the Lyapunov stability criterion, enabling the motion trajectory of the system on the sliding mode surface to quickly converge to the equilibrium point. To suppress the high-frequency chattering caused by the sign function in traditional sliding mode control, the sign function is replaced by a saturation function, and its boundary value is dynamically adjusted by the anti-chattering amount provided by the path planning module. When the tangential friction force of the end effector is real-time feedback by the three-dimensional force sensor and a mutation trend is detected, the anti-chattering amount adaptively scales according to the friction force change rate, thereby dynamically widening the boundary of the saturation function and effectively smoothing the torque output while maintaining the tracking accuracy.

[0027] Furthermore, by real-time collecting the tangential friction force data of the end and calculating its change rate, a non-linear adjustment factor is generated after comparing the change rate with a preset threshold. The adjustment factor is inversely related to the sliding mode switching gain. When the friction force increases, the adjustment factor automatically attenuates the switching gain, reducing the mutation amplitude of the control quantity and avoiding torque oscillation caused by external disturbances. The sliding mode surface function, reaching law, and dynamic suppression coefficient are fused, and a smooth joint target torque curve is generated through multi-variable collaborative calculation. This curve is further mapped to the torque suppression threshold of the path planning instruction, driving the harmonic reducer and servo motor to strictly limit the tangential friction force of the end within the preset safety range while tracking the target angle.

[0028] The present invention provides a manipulator control system and a control method for a photovoltaic panel intelligent installation robot, having the following beneficial effects:

[0029] 1. By setting up a sensor module composed of multiple sensors, the present invention can comprehensively and real-time obtain the state and environmental information of the manipulator, providing rich data support for precise control and improving the control accuracy of the manipulator.

[0030] 2. Through the optimized control algorithm, the present invention can quickly and accurately process data and plan the motion path of the manipulator, realizing precise control of the manipulator and improving the accuracy and stability of installation.

[0031] 3. The real-time monitoring and adjustment function of the present invention enables the manipulator to promptly respond to various abnormal situations during the operation process, avoiding damage to the photovoltaic panel and installation failure, and improving the reliability and success rate of installation. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic flow chart of the present invention. Detailed implementation manners

[0033] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0034] Embodiment 1: Refer to Figure 1 , the embodiment of the present invention provides a manipulator control system for a photovoltaic panel intelligent installation robot, and the specific implementation manner is as follows:

[0035] The system consists of a sensor module, a central processing module, a path planning module, a driving module and a human-machine interaction module. Each module works together to achieve high-precision installation of photovoltaic panels. The three-dimensional force sensor in the sensor module is installed at the contact point between the end effector of the manipulator and the photovoltaic panel to detect the normal pressure and tangential friction force in real time. Its sampling frequency is 1000Hz, and the measurement accuracy is ±0.5N. The magnetic encoder is integrated at each joint of the manipulator, with a resolution of 0.001 degrees, and it can feedback the joint rotation angle data in real time. The binocular vision sensor is deployed at the base of the manipulator to scan the installation area of the photovoltaic panel and generate a three-dimensional point cloud map. Its scanning accuracy is ±0.1mm, and it can identify the surface curvature characteristics of obstacles. For example, in the installation scenario, the binocular vision sensor collects point cloud data at a rate of 30 frames per second. By spatially discretizing the continuous surface, it is divided into a discrete point set with a spacing of 2mm. For each discrete point, the point cloud data within a neighborhood with a radius of 5cm is extracted, a local covariance matrix is constructed and eigenvalue decomposition is performed to extract the principal curvature and the curvature direction. The principal curvature is determined by the direction of the eigenvector corresponding to the minimum eigenvalue, and the curvature direction is defined by the first principal component eigenvector. The curvature change rate between adjacent discrete points is calculated by the ratio of the difference in principal curvature between two points and the Euclidean distance. For example, when the distance between two points is 5mm and the principal curvature difference is 0.02mm -2 , the curvature change rate is 0.004mm -2 .

[0036] After receiving the sensor data, the central processing module constructs a curvature gradient field based on the three-dimensional point cloud map and identifies the regions with sudden changes in the curvature gradient. A local coordinate system is established with the current position of the end effector as the origin, and the global curvature gradient field is projected onto this coordinate system. The initial approximation direction vector is decomposed into components in the curvature gradient direction and the tangent plane direction. By screening the tangent plane directions with a curvature gradient change rate lower than 0.1mm -2 , the optimal safe approximation direction is selected. For example, when the curvature gradient change rate of a certain tangent plane direction is detected to be 0.05mm -2When it is determined to be a safe path; the central processing module continuously compares the normal pressure feedback by the three-dimensional force sensor with the preset anti-compression threshold of the photovoltaic panel, such as 300N. When the pressure exceeds the limit, the reverse displacement compensation amount of the end effector is calculated based on inverse dynamics; the end compensation amount is mapped to the joint space through the pseudo-inverse method of the Jacobian matrix. For example, when the normal pressure deviation ΔP = 50N, the reverse displacement compensation amount ΔX = 0.2mm, and the dynamic damping coefficient λ in the pseudo-inverse calculation of the Jacobian matrix is set to 0.1 to ensure numerical stability under singular configurations; the joint angle increment ΔΨ is calculated by the formula where J + damped is the damped pseudo-inverse of the Jacobian matrix; the central processing module synchronously calculates the upper limit attenuation amount of the joint angular velocity. For example, the initial upper limit of the angular velocity ω0 = 1.5rad / s. When the pressure change rate ΔP / Δt = 10N / s, the attenuation coefficient k = 0.1, and the corrected upper limit of the angular velocity is 0.75rad / s to suppress the over-limit of the end friction force.

[0037] The path planning module generates a target joint angle sequence and an angular velocity sequence based on the anti-slip movement parameters; the piecewise cubic Hermite interpolation method is used to smooth the trajectory. For example, within the time interval [0.5s, 1.0s], the initial angle θ0 = 30°, the target angle θ1 = 45°, the anti-shake angular velocity ω0 = 0.8rad / s, and the interpolation function H(t) satisfies H(0.5) = 30°, H(1.0) = 45°, H ’ (0.5) = 0.8rad / s, H ’ (1.0) = 0.6rad / s, generating a continuous and smooth angle curve; the dynamic suppression coefficient C is adjusted according to the change rate of the end tangential friction force. For example, the initial value C0 = 0.5. When it is continuously detected three times that Δf / Δt > 10N / s, the growth factor α = 0.2 triggers an increase, and the update formula is C = 0.7; each time stamp in the path planning instruction is associated with an angle value, an angular velocity limit value, and a torque suppression threshold. For example, at t = 1.2s, θ = 40°, and C = 0.8.

[0038] After receiving the path planning instruction, the harmonic reducer in the drive module analyzes the target joint angle θ target , combines it with the actual angle θ actual fed back by the magnetic encoder to calculate the deviation , and adjusts the output torque through a control algorithm; for example, when Δθ = 0.5°, the preset control parameters K p = 10Nm / deg, K i = 0.5Nm / (deg·s), K d = 2Nm·s / deg, calculate the torque increment , driving the adjusted torque output of the harmonic reducer; the torque feedback servo motor adjusts the rotational speed according to the dynamic suppression coefficient C = 0.8 and the angular velocity limit ω max = 1.0 rad / s, and at the same time, based on the torque threshold limits the output torque; for example, when the real-time frictional force f t = 45 N, τ limit = 0.8 × 0.8 × 50 = 32 Nm. If the required torque τ demand = 40 Nm, then the actual output torque τ output = min(40, 32) = 32 Nm, ensuring that the tangential frictional force is controlled; among them, according to the real-time value of the dynamic suppression coefficient C, a hysteresis interval of the servo motor torque threshold is constructed , tightening the threshold during the rising stage of the frictional force and relaxing the threshold during the falling stage, avoiding frequent torque jumps caused by traditional fixed thresholds. Through the non-linear coupling of the dynamic suppression coefficient C and the hysteresis interval, problems such as torque jumps, response lags, poor adaptability to multiple materials, and high energy consumption in traditional control are solved, realizing high-precision and low-power robust control in the photovoltaic panel installation scenario.

[0039] The human-machine interaction module triggers an alarm in case of an abnormality. For example, when the normal pressure exceeds the limit continuously for 3 times or the tangential frictional force exceeds 50 N, the movement of the robotic arm is frozen and the operator is prompted to input an emergency compensation amount; the operator can input the compensation amount through the interface, such as a reverse displacement amount of ±10%, covering the original path planning instruction; for example, when it is detected that the end pose deviation exceeds 5 mm, the operator inputs a +8% compensation amount, and the system adjusts the reverse displacement amount from 0.2 mm to 0.216 mm and re-plans the path.

[0040] Embodiment 2: A method for controlling the robotic arm of a photovoltaic panel intelligent installation robot, and the specific implementation manner is as follows:

[0041] When implementing the method for controlling the robotic arm of the photovoltaic panel intelligent installation robot, first, a binocular vision sensor is used to comprehensively scan the photovoltaic panel installation area to construct a three-dimensional point cloud map of the installation area; at the same time, the magnetic encoder real-time feedbacks the rotation angle of the robotic arm joint, and the three-dimensional force sensor continuously monitors the normal pressure and tangential frictional force when the end effector contacts the photovoltaic panel; assume that during a certain installation operation, the end effector of the robotic arm approaches the surface of the photovoltaic panel, and at this time, the three-dimensional force sensor detects that the normal pressure when the end effector contacts the photovoltaic panel gradually increases. When the normal pressure reaches 120 N, it exceeds the preset photovoltaic panel compressive threshold of 100 N.

[0042] In the face of a situation beyond the threshold, the central processing unit quickly intervenes to calculate the deviation between the normal pressure and the preset threshold, and the deviation is 20 N. Based on the principle of inverse dynamics, the pressure difference is converted into the inverse displacement compensation amount of the end effector along the normal direction, and then an inverse pressure compensation instruction including the compensation direction and amplitude is generated. At the same time, combined with the joint angle data fed back by the magnetic encoder, through inverse kinematics calculation, the inverse displacement compensation amount of the end effector is mapped to the joint space to establish the dynamic correlation between the joint angle increment and the change of the end normal pressure.

[0043] According to the surface curvature characteristics of the obstacle extracted by the binocular vision sensor, calculate the approaching path of the end effector of the robotic arm along the normal direction of the photovoltaic panel surface, and generate a safe approaching angle. For example, when there is a roof with a certain inclination angle in the installation area, the safe approaching angle will be calculated according to the curvature change of the roof surface, so that the end effector can smoothly approach the photovoltaic panel along this angle.

[0044] On the basis of generating the inverse pressure compensation instruction and the safe approaching angle, further adopt the non-linear sliding mode control algorithm to calculate the target motion parameters of the robotic arm joints, including the anti-vibration amount and the dynamic suppression coefficient of the tangential friction force. Assume that the anti-vibration amount is set to 0.05 rad / s, and the initial value of the dynamic suppression coefficient is 0.8. After algorithm calculation, the updated dynamic suppression coefficient is 0.85, and finally a path planning instruction is generated.

[0045] According to the path planning instruction, the driving torque feedback servo motor rotates to control the robotic arm to move along the planned path. In this process, the tangential friction force is strictly limited within the preset safe range. For example, the upper limit of the preset tangential friction force safe range is 80 N. If it is detected that the tangential friction force approaches or reaches this upper limit value during the movement, the robotic arm control system will adjust in real time to ensure that the tangential friction force is always within the safe range.

[0046] However, various abnormal situations will be encountered in actual operation. For example, when the normal pressure exceeds the preset threshold three times in a row, or the tangential friction force exceeds the preset safe range, or the pose deviation of the end effector exceeds the allowable tolerance. For example, the tolerance is set to 0.1 mm, and the actual deviation reaches 0.15 mm, and the deviation between the actual torque and the target torque of the robotic arm joint continuously exceeds the dynamic threshold. For example, the dynamic threshold is set to 5 N·m, and the actual deviation reaches 6 N·m, the system will trigger an alarm and freeze the movement of the robotic arm. At this time, the operator needs to confirm the compensation strategy through the human-machine safety interaction module, such as inputting an emergency inverse compensation amount to overwrite the original path planning instruction, so that the robotic arm can continue to perform the photovoltaic panel installation operation safely and stably according to the new compensation strategy.

[0047] Through this control method, the robotic arm of the intelligent PV panel installation robot can achieve high-precision and high-stability installation operations in complex environments, effectively avoiding damage to PV panels and installation failures, and improving the reliability and success rate of installation.

[0048] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A manipulator control system for a photovoltaic panel intelligent installation robot, characterized in that, Including: Sensor module: The three-dimensional force sensor detects the normal pressure and tangential friction force when the end effector contacts the photovoltaic panel; The magnetic encoder real-time feedbacks joint angle data; The binocular vision sensor constructs a three-dimensional point cloud map of the photovoltaic panel installation area and identifies the surface curvature of obstacles; Central processing module: Processes the data collected by the sensor module; Calculates the safe approaching direction of the end effector of the robotic arm according to the identified surface curvature of the obstacle; Compares the normal pressure with the preset compressive threshold of the photovoltaic panel, and generates a reverse pressure compensation command when the normal pressure exceeds the threshold; Based on the joint angle data, combines the safe approaching direction and the reverse pressure compensation command, and generates the anti-slip movement parameters of the robotic arm joints through the pseudo-inverse method of the Jacobian matrix and the dynamic damping coefficient; Path planning module: Calculates the target movement parameters of each joint based on the anti-slip movement parameters of the robotic arm joints, and corrects the anti-jitter amount of the joint angular velocity and the dynamic suppression coefficient of the tangential friction force of the end effector according to the joint angle feedback, and generates a path planning command; Drive module: The harmonic reducer and torque feedback servo motor directly connected to the robotic arm joints. The harmonic reducer adjusts the output torque according to the path planning command to track the planned path. The torque feedback servo motor limits the tangential friction force based on the dynamic suppression coefficient, and at the same time adjusts the motor speed according to the trajectory of the path. Among them, a hysteresis interval of the servo motor torque threshold is constructed according to the dynamic suppression coefficient, and the threshold is tightened in the rising stage of the friction force and relaxed in the falling stage; Human-machine interaction module: Triggers an alarm when the path planning command of the robotic arm is abnormal, and receives the emergency reverse compensation amount input by the operator to overwrite the path planning command.

2. The robotic arm control system of a photovoltaic panel intelligent installation robot according to claim 1, wherein: The central processing module performs spatial discretization processing on the point cloud data in the binocular vision sensor, extracts the principal curvature and curvature direction of each discrete point on the obstacle surface, and calculates the curvature change rate between adjacent discrete points through the curvature gradient; Constructs an obstacle surface curvature gradient field according to the curvature change rate, and identifies the boundary of the curvature gradient mutation region; Establishes a local coordinate system with the current position of the end effector as the origin, projects the curvature gradient field onto this local coordinate system, decomposes the initial approaching direction vector into the components of the curvature gradient direction and the tangent plane direction, and selects the tangent plane direction with the smallest component of the curvature gradient direction as the safe approaching direction.

3. The robotic arm control system of an intelligent photovoltaic panel installation robot according to claim 1, wherein: The central processing module calculates the dynamic difference between it and the preset compressive threshold of the photovoltaic panel based on the normal pressure data of the end effector. When the difference exceeds the threshold, converts the pressure difference into the reverse displacement compensation amount of the end effector in the normal direction based on inverse dynamics, and generates a reverse pressure compensation command including the compensation direction and amplitude; Based on the joint angle data, maps the normal displacement of the end of the reverse pressure compensation command to the joint angle increments through the pseudo-inverse method of the Jacobian matrix and the dynamic damping coefficient; Combines the tangent plane vector component of the safe approaching direction, constructs the motion constraint conditions in the joint space, and adjusts the upper limit of the angular velocity and the torque threshold through the real-time ratio of the joint angle change rate and the normal pressure change rate, and generates the anti-slip movement parameters that suppress the tangential friction force of the end, so that the joint trajectory is smooth and the end contact force is strictly controlled by the compressive threshold of the photovoltaic panel.

4. The robotic arm control system of an intelligent photovoltaic panel installation robot according to claim 3, characterized in that: The path planning module generates the target joint angles and angular velocities in the anti-slip movement parameters of the robotic arm joints generated by the central processing module, and calculates the target movement trajectories of each joint through inverse kinematics of the robotic arm to generate an initial joint target angle sequence and the corresponding angular velocity sequence; it receives the joint angle data fed back by the magnetic encoder in real time, calculates the real-time deviation between the current joint angle and the target angle, dynamically adjusts the anti-shake amount of the joint angular velocity based on the time derivative of the deviation value, and limits the angular velocity change rate within a preset smoothing threshold to form an anti-shake joint angular velocity sequence. It synchronously reads the tangential friction force data of the end effector fed back by the three-dimensional force sensor, and adjusts the dynamic suppression coefficient of the end tangential friction force with the change rate of the friction force as the input according to the mapping relationship between the friction force and the joint torque, so that the suppression coefficient increases with the increase of the friction force; it synchronizes the anti-shake joint angular velocity sequence, the dynamic suppression coefficient and the target angle sequence in time to generate a path planning instruction, and outputs it to the harmonic reducer and the torque feedback servo motor of the drive module to control the harmonic reducer to adjust the output torque according to the angle sequence.

5. The robotic arm control system of an intelligent photovoltaic panel installation robot according to claim 4, characterized in that: In the drive module, the harmonic reducer receives the joint angle sequence and the corresponding time stamp in the path planning instruction, analyzes the target joint angle value of the current time stamp, combines the actual joint angle fed back by the magnetic encoder, calculates the angle deviation, and determines the adjustment amount of the output torque based on the preset torque and the angle deviation, and drives the harmonic reducer to output a torque matching the target angle to track the planned path. The torque feedback servo motor reads the dynamic suppression coefficient and the angular velocity limit value in the path planning instruction, dynamically adjusts the motor speed according to the angular velocity limit value of the current time stamp to synchronize it with the angular velocity requirement of the path trajectory; during the adjustment process, through the dynamic suppression coefficient and the real-time detected end tangential friction force data, the output torque of the servo motor is limited within the threshold range corresponding to the suppression coefficient by the torque control algorithm. When the friction force exceeds the threshold, the output torque is reduced according to the suppression coefficient ratio to limit the tangential friction force.

6. A control method for a robotic arm control system of a photovoltaic panel intelligent installation robot as described in any one of claims 1-5, characterized in that, It includes the following steps: S1. Scan the installation area of the photovoltaic panel through the binocular vision sensor to extract the surface curvature characteristics of the obstacle. At the same time, the magnetic encoder feeds back the joint rotation angles of the robotic arm in real time, and the three-dimensional force sensor obtains the normal pressure and tangential friction force of the end effector in real time. S2. According to the surface curvature characteristics of the obstacle, calculate the approaching path of the end effector of the robotic arm along the normal direction of the photovoltaic panel surface, and generate a safe approaching angle. S3. Generate a path planning instruction through dynamic pressure compensation and anti-slip movement parameters. The specific steps include: S31. When the normal pressure exceeds the preset anti-compression threshold of the photovoltaic panel, calculate the reverse compensation pressure according to the over-limit ratio, and generate a reverse pressure compensation instruction. S32. Based on the safe approaching angle and the reverse pressure compensation instruction, combine the joint angle data of the magnetic encoder to generate the anti-slip movement parameters of the robotic arm joints. S33. Use the non-linear sliding mode control algorithm to calculate the target movement parameters of the robotic arm joints, including the anti-shake amount and the dynamic suppression coefficient of the tangential friction force, and generate a path planning instruction. S4. Drive the servo motor to rotate according to the path planning instruction, control the robotic arm to move along the planned path, and limit the tangential friction force within the preset safe range; S5. When the normal pressure continuously exceeds the set number of times, the tangential friction force exceeds the preset safe range, or the pose deviation of the end effector exceeds the allowable tolerance, and the deviation between the actual torque and the target torque of the robotic arm joint continuously exceeds the dynamic threshold, trigger an alarm and freeze the movement of the robotic arm until the operator confirms the compensation strategy through the human-machine safety interaction module.

7. The control method of the robotic arm control system of a photovoltaic panel intelligent installation robot according to claim 6, characterized in that: The non-linear sliding mode control algorithm is based on the target joint angle sequence in the anti-slip movement parameters generated by the path planning module and the actual joint angle data feedback by the magnetic encoder. Define the joint angle tracking error and its time derivative as state variables, and construct a non-linear sliding mode surface function containing the angle error and the angular velocity error; According to the sliding mode surface function and the dynamic suppression coefficient threshold of the end tangential friction force, design a non-linear reaching law, and derive the control quantity expression of the joint torque through the Lyapunov stability criterion; Introduce a saturation function of the joint angle deviation to replace the sign function in the control quantity, and use the anti-chattering amount as the saturation boundary value to suppress high-frequency chattering; According to the real-time data of the end tangential friction force feedback by the three-dimensional force sensor, non-linearly couple the dynamic suppression coefficient with the friction force change rate to generate an adjustment factor for the sliding mode switching gain, so that the switching gain decays with the increase of the friction force, reducing the risk of torque mutation; Integrate the calculated joint torque control quantity, anti-chattering amount and dynamic suppression coefficient to generate a smooth joint target torque curve, and map it to the torque suppression threshold of the path planning instruction, driving the robotic arm joint to limit the tangential friction force within the preset safe range while tracking the target angle.

Images