An end effector for adaptive gripping of photovoltaic modules

Abstract

The application discloses an end gripper for adaptive gripping of photovoltaic modules, comprising an end gripper frame, a visual relative pose measurement module, a multi-axis linkage pose compensation mechanism, a vacuum suction execution module and a main controller. The visual relative pose measurement module solves the spatial relative pose parameters of the photovoltaic module relative to the end gripper frame in real time. The main controller extracts the rotational deviation amount in the parameters, generates a servo adjustment instruction to drive the multi-axis linkage pose compensation mechanism to perform iterative approximation type pose correction, forming a visual servo closed-loop control. When the rotational deviation amount converges within the preset tolerance range, it is determined that the pose matching state is reached and the vacuum suction execution module is triggered to work. The application realizes real-time adaptive compensation for dynamic angle deviation of the photovoltaic module during conveying, solves the problems that the existing end gripper lacks integrated multi-dimensional pose adjustment capability and deep collaborative closed-loop control, and significantly improves the gripping precision and automation level.

Claims

1. An end effector for adaptive grasping of photovoltaic modules, characterized in that, include: End effector frame; A visual relative pose measurement module is fixedly installed on the front operating surface of the end effector frame. The visual relative pose measurement module includes an image acquisition unit and a vision processor. The vision processor is configured to: during the entire approach process of the end effector moving towards the photovoltaic module, based on the continuous image frames acquired by the image acquisition unit, calculate in real time the spatial relative pose parameters of the surface features of the photovoltaic module relative to a preset physical grasping reference frame on the end effector frame through feature point matching and 3D reconstruction algorithms. The spatial relative pose parameters are used to characterize the amount of displacement deviation and rotation deviation that the photovoltaic module needs to be compensated for. A multi-axis linkage attitude compensation mechanism is connected to the end effector frame and includes three independently controlled servo motor adjustment units. Each servo motor adjustment unit compensates for one rotational degree of freedom. The multi-axis linkage attitude compensation mechanism is used to drive the end effector frame to rotate around the corresponding axis according to the received adjustment command, so as to eliminate the rotational deviation. A vacuum adsorption execution module is installed at the bottom of the end effector frame and includes at least one vacuum adsorption circuit for fixing the photovoltaic module by negative pressure adsorption after the end effector frame and the photovoltaic module surface reach a pose matching state. The main controller, which is connected to the visual relative pose measurement module, the multi-axis linkage attitude compensation mechanism, and the vacuum adsorption execution module respectively, is configured as follows: The spatial relative pose parameters output by the visual relative pose measurement module are acquired in real time. The rotational deviation in the spatial relative pose parameters is extracted, and a servo adjustment sequence command is generated based on the extracted rotational deviation to drive the multi-axis linkage attitude compensation mechanism to perform iterative approximation attitude correction according to the servo adjustment sequence command. During the iterative approximation process, the spatial relative pose parameters are continuously updated. When all components of the rotational deviation converge to the preset zero value neighborhood, the visual servo closed-loop correction is determined to be complete, and the pose matching state is achieved. In the pose matching state, the vacuum adsorption execution module is triggered to start working.

2. The end effector for adaptive grasping of photovoltaic modules as described in claim 1, characterized in that, The vacuum adsorption execution module includes a first vacuum circuit and a second vacuum circuit. The two vacuum circuits are independent of each other in the gas path and are symmetrically distributed on both sides of the center of gravity of the photovoltaic module on the end effector frame. Each vacuum circuit includes a vacuum generator, a vacuum filter, a vacuum solenoid valve, and a vacuum suction cup assembly. The air inlet of the vacuum generator is connected to a gas source through a pipeline. The vacuum port of the vacuum generator is connected in series with the vacuum filter and the vacuum solenoid valve through a pipeline. The output end of the vacuum solenoid valve is connected to a corresponding set of vacuum suction cup assemblies through a diversion pipeline. The vacuum generator is used to generate negative pressure, the vacuum filter is used to purify the gas entering the vacuum circuit, the vacuum solenoid valve is used to open and close the gas path of the corresponding vacuum circuit under the control of the main controller, and the vacuum suction cup assembly is used to adsorb photovoltaic modules through negative pressure when the gas path is connected.

3. The end effector for adaptive grasping of photovoltaic modules as described in claim 2, characterized in that, A vacuum pressure sensor is installed on the main pipe of each vacuum circuit. The vacuum pressure sensor is connected to the main controller through an analog input channel to convert the collected vacuum pressure value into a 4-20mA standard analog signal and transmit it to the main controller. The main controller is also configured to: after triggering the vacuum adsorption execution module to start working, independently perform vacuum pressure closed-loop control for each vacuum circuit based on the vacuum pressure value fed back by the vacuum pressure sensor.

4. The end effector for adaptive grasping of photovoltaic modules according to claim 3, characterized in that, The vacuum pressure closed-loop control includes: Set pressure operating range [P] min ,P max ], where P min To preset the lower threshold, P max The preset upper limit threshold; Real-time monitoring of vacuum pressure value P current ; If P current <P min If so, the vacuum generator in the corresponding vacuum circuit will stop working; If P current >P max If so, the vacuum generator of the corresponding vacuum circuit will be started to increase the vacuum pressure; If P current ∈[P min ,P max If ], the current working state of the corresponding vacuum circuit will be maintained.

5. The end effector for adaptive grasping of photovoltaic modules as described in claim 2, characterized in that, Also includes: The signal detection system includes a proximity switch, which is mounted on the end effector frame and is at a distance less than a preset distance from the vacuum suction cup assembly. The output terminal of the proximity switch is electrically connected to the main controller and is used to output a position signal to the main controller when the suction cup of the vacuum adsorption execution module is attached to the surface of the photovoltaic module. The main controller is configured to trigger the vacuum adsorption execution module to start working only when it simultaneously determines that the pose matching state has been reached and receives the position signal output by the proximity switch.

6. The end effector for adaptive grasping of photovoltaic modules according to claim 2, characterized in that, Each suction cup in the vacuum suction cup group integrates a micro pressure sensor array. The micro pressure sensor array is arranged in a grid on the suction cup adsorption surface to measure the pressure value of multiple local points in the contact area between the suction cup and the photovoltaic module in real time, and generate local pressure distribution data. The main controller is also configured to perform the following steps: Receive local pressure distribution data of all suction cups, and receive real-time identification of photovoltaic module surface structure features by the visual relative pose measurement module. The surface structure features include at least the module frame position, junction box position, and known defect areas. The surface structure features are mapped to the coordinate system of the end effector frame. Combined with the mechanical properties of photovoltaic module materials, a safe area that allows high-pressure adsorption and a restricted area that requires avoiding high pressure or contact are delineated. Based on the mapped safe and restricted areas map and the current local pressure distribution data, the optimal suction cup activation combination at the current moment is dynamically calculated, and the corresponding target pressure value is calculated for each activated suction cup. The calculation aims to ensure that the total suction force meets the requirements and the pressure distribution avoids the restricted areas. The vacuum adsorption execution module is controlled to activate only the suction cups in the optimal suction cup activation combination, and the pressure distribution of each activated suction cup is made close to its corresponding target pressure value by adjusting the working parameters of the vacuum generator of each vacuum circuit.

7. The end effector for adaptive grasping of photovoltaic modules as described in claim 1, characterized in that, The vision processor is configured to execute the following algorithm steps: The image acquisition unit acquires a raw image containing features of the photovoltaic module edge or junction box location. The original image is preprocessed to enhance feature contrast and reduce noise interference; The contour feature points of the photovoltaic module are extracted using an edge detection algorithm; The extracted contour feature points are matched with the preset standard photovoltaic module template features. Based on the successfully matched feature point pairs, the three-dimensional pose of the photovoltaic module relative to the coordinate system of the image acquisition unit is calculated using the PnP algorithm. By combining the fixed calibration relationship between the image acquisition unit and the end effector frame, the three-dimensional pose is transformed to the physical grasping reference frame to obtain the spatial relative pose parameters.

8. The end effector for adaptive grasping of photovoltaic modules as described in claim 1, characterized in that, The rotational degrees of freedom correspond to X-axis rotation, Y-axis tilt, and Z-axis deflection, respectively. The main controller is configured to execute the following iterative approximation attitude correction algorithm: Set the preset tolerance threshold for each rotational degree of freedom. , , Set the maximum number of iterations N; For the k-th iteration, where k ranges from [1, N], the following steps are performed: Obtain the current spatial relative pose parameters and extract the rotational deviation θ. k =[θ x ,θ y ,θ z ]; Calculate the servo adjustment amount Δθ k = K p ·θ k +K i ·Σθ+K d ·(θ k -θ {k-1} ), where K p , K i ,K d For the preset control parameters, Σθ is the sum of all historical rotation deviations from the 1st to the kth iteration, θ {k-1} This represents the rotational deviation during the previous iteration. Generate and output the corresponding Δθ k The servo adjustment sequence command is executed, and the servo motor adjustment unit of the multi-axis linkage attitude compensation mechanism is waited to complete the execution and obtain the updated spatial relative pose parameters. Repeat the above iterative steps until ||θ {k+1} ||<[ , , ] T If the iteration loop is broken, the visual servo closed-loop correction is considered complete. If the condition is not met even after k reaches N, the iteration process will exit and an error message will be output.

9. The end effector for adaptive grasping of photovoltaic modules according to claim 1, characterized in that, It also includes a digital twin prediction module, which is connected to the main controller and configured to perform the following steps: Based on the mass distribution and joint stiffness parameters of the end effector frame, a digital twin model of end effector dynamics is constructed. Based on the historical operating data of the photovoltaic module conveyor line, the typical vibration spectrum and attitude fluctuation mode of the conveyor line at the preset speed are analyzed and extracted. The two are then integrated to establish an attitude change prediction model. During the iterative approximation attitude correction process performed by the main controller, multiple sets of historical spatial relative pose parameters continuously output by the visual relative pose measurement module are received in real time, and a historical parameter sequence is constructed in chronological order. The historical parameter sequence is input into the attitude change prediction model to calculate the predicted attitude change of the photovoltaic module relative to the end effector frame within a future preset time window. The predicted attitude change includes the predicted displacement deviation and the predicted rotation deviation. The main controller is further configured to: when generating servo adjustment sequence instructions, generate a composite adjustment instruction containing a feedforward compensation term based on the currently extracted rotational deviation and the predicted rotational deviation output by the digital twin prediction module, and drive the multi-axis linkage attitude compensation mechanism to perform advanced attitude adjustment.

10. The end effector for adaptive grasping of photovoltaic modules according to claim 1, characterized in that, The image acquisition unit is an event camera, used to output event data containing pixel position, timestamp and polarity when a pixel brightness change is detected to exceed a preset brightness threshold, forming an asynchronous event stream; The vision processor is configured to perform the following steps: Receive the asynchronous event stream output by the event camera, and perform noise filtering and time consistency verification on the event stream; In the event domain, by identifying the spatiotemporal clustering patterns of events, event clusters corresponding to the edge or corner features of photovoltaic modules are extracted, and the motion trajectories of these event clusters are tracked within a continuous time window. Based on the tracked event cluster motion trajectories, the global motion vector on the surface of the photovoltaic module is estimated, motion compensation is performed on the event flow, and the events within a specific time window are accumulated and reconstructed into a grayscale image; Based on the compensated event cluster 3D motion trajectory or reconstructed grayscale image, the spatial relative pose parameters of the photovoltaic module surface features relative to the preset physical grasping reference frame on the end effector frame are calculated in real time through feature point matching and 3D reconstruction algorithms.

Images