Reimbursement robot equipment based on AI large model bill discrimination

Abstract

The invention discloses reimbursement robot equipment based on AI large model bill discrimination, and belongs to the technical field of financial reimbursement automation equipment. The equipment integrates deep linkage of a printing execution module, a reimbursement robot hardware system and an AI recognition auditing system, and realizes integrated operation of paper bill scanning recognition, electronic bill generation, auditing result associated printing and filing through cooperative control of a mechanical arm and closed-loop connection of a printing process and a reimbursement full link. Closed-loop circulation of bills in the bill delivery bin, the scanning position, the temporary storage bin and the filing bin is achieved through the multi-axis mechanical arm; adopting a ViT + Qwen multi-mode large model to realize end-to-end structured extraction of the bill; and in combination with a background AI auditing engine and a multi-dimensional analysis module, full-chain automation from bill physical storage, intelligent identification and compliance auditing to prediction and archiving is completed.

Description

Technical Field

[0001] This invention relates to the field of invoice printing, and more specifically, to a reimbursement robot device based on AI large-scale model for invoice discrimination. Background Technology

[0002] In existing financial reimbursement systems, the printing function is mostly a separate module, disconnected from the processes of scanning, recognizing, reviewing, and archiving invoices. Traditional solutions require manual intervention to trigger the printing of electronic invoices after approval, followed by merging and archiving the printed copies with the original paper invoices. This approach has the following drawbacks: significant disconnects between the printing and reimbursement processes; low processing efficiency due to manual operation; frequent errors in matching original invoices with electronic printouts, affecting audit traceability; and a lack of intelligent control over printing triggering, potentially leading to printing without prior review or duplicate printing. Furthermore, the printing module of existing reimbursement robots is not deeply integrated with AI recognition systems and robotic arm transfer systems, failing to achieve a fully automated closed-loop process.

[0003] Among the existing technologies, CN211415179U is a basic reimbursement robot structure (scanner + storage box) that lacks an AI recognition module; in addition, CN119740969A is a pure software RPA reimbursement solution that has no hardware automation capabilities.

[0004] Therefore, we have made improvements to this by proposing a reimbursement robot device based on AI large-scale model for invoice discrimination. Summary of the Invention

[0005] To achieve the above-mentioned objectives, this invention provides a reimbursement robot device based on AI large-scale model for invoice discrimination, in order to solve the aforementioned problems.

[0006] The present invention is as follows: It includes a printing unit, a robotic arm gripper assembly, a printing compartment, and a background linkage control module. The printing unit is connected to the AI ​​processing unit and approval system of the reimbursement robot. The robotic arm gripper assembly is adapted to the printing compartment, temporary storage compartment, and archiving compartment to realize the grabbing and transfer of documents.

[0007] Preferably, the printing unit is equipped with an encrypted data receiving interface, which can receive structured JSON data output by the AI ​​processing unit and approval instructions from the approval system, and automatically generate electronic invoices and reimbursement summary sheets.

[0008] Preferably, the printing compartment is equipped with an arrival detection sensor. When the printed part is output to the printing compartment, the sensor sends a transfer signal to the robotic arm gripper assembly, triggering the document merging operation.

[0009] Preferably, the robotic arm gripper assembly includes a horizontal drive motor, a vertical drive motor, a horizontal guide rail, a vertical guide rail, grippers, and a gripper drive motor, which can accurately grasp and merge the original paper documents and the output of the printing unit in the temporary storage compartment.

[0010] Preferably, the background linkage control module is connected to the AI ​​trend prediction and analysis system, which can receive multi-dimensional analysis data and support printing structured management reports containing trend curves, pie charts, and prediction tables.

[0011] Preferably, the printing unit's printing trigger logic is linked to the reimbursement approval process, and the printing program is automatically started only after the AI ​​intelligent review is passed and compliance verification and duplicate reimbursement detection are completed.

[0012] Preferably, the robotic arm gripper assembly is adapted to the partition grid design of the filing compartment, and can accurately place the merged paper documents and printouts into the corresponding filing compartments according to department / month classification rules, with a positioning error of ≤1mm.

[0013] The usage process of the expense reimbursement robot device based on AI large model for invoice discrimination provided by this invention is as follows: The person seeking reimbursement clicks the "Submit" button on touchscreen 11, unlocks the electronic lock 2 in the submission compartment 1, and closes the compartment door after the paper invoice is placed in. The robotic arm gripper 4 uses infrared positioning to grab the invoice and transfers it to the fully automatic scanning station 13 to complete image acquisition. The image is uploaded to the AI ​​processing unit for structured extraction. The person seeking reimbursement submits an electronic reimbursement application through the backend system. The system automatically associates the structured data to generate a reimbursement form. After AI intelligent review and approval, the backend linkage control module sends a printing command to the printing unit. The printing unit prints the electronic invoice and summary sheet, outputs it to the printing compartment 20, and triggers sensor signals. The robotic arm gripper 4 takes out the original paper invoice from the temporary storage compartment 12, merges it with the printed documents in the printing compartment 20, and transfers it to the corresponding grid in the archiving compartment 17 according to the department / month classification rules. The backend AI trend prediction and analysis system can trigger the printing of management reports as needed, and the printing unit outputs paper reports.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Strong integration: The printing unit is deeply integrated with the entire reimbursement process, achieving an automated closed loop of "identification-approval-printing-merging-archiving" without manual intervention; 2. Precise and efficient: The robotic arm's positioning error is ≤1mm, ensuring that the original invoice and the printed copy match correctly, and the time for printing and merging a single invoice is ≤8 seconds; 3. Convenient auditing: Printed copies are strongly linked to original documents, and logs are kept throughout the entire process, improving audit traceability efficiency by 300%; 4. Wide adaptability: Supports batch printing of ≤50 pages after continuous scanning, adapting to the electronic printing needs of various complex documents such as blurry, handwritten, and multilingual documents; 5. Flexible deployment: No need to configure print templates; print output format can be adjusted via natural language commands, making it ready to use immediately. Attached Figure Description

[0015] Figure 1 This invention provides an overall structural schematic diagram of a reimbursement robot device based on an AI large model for invoice discrimination. Figure 2 A left view of a reimbursement robot device based on AI large model for invoice discrimination provided by the present invention; Figure 3 The right view of a reimbursement robot device based on AI large model for invoice discrimination provided by the present invention; Figure 4 This invention provides a schematic diagram of a printing device for a reimbursement robot based on an AI large-scale model for invoice discrimination.

[0016] Figure 5 This invention provides a system multimodal recognition engine architecture diagram for a reimbursement robot device based on AI large model for invoice discrimination.

[0017] The image shows: 1. Order delivery compartment; 2. Electronic lock; 3. Electric slide rail; 4. Robotic arm gripper; 5. Horizontal drive motor; 6. Vertical drive motor; 7. Horizontal guide rail; 8. Vertical guide rail; 9. Gripper; 10. Drive motor; 11. Touch screen; 12. Temporary storage compartment; 13. Fully automatic scanning position; 14. High-definition camera; 15. Position to be scanned; 16. Scanning position; 17. Archiving compartment; 18. Printer; 19. Scanning and transporting mechanism; 20. Printing compartment; 21. Fan. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0019] To enable those skilled in the art to better understand the present invention, the following will be described in conjunction with the appendix. Figure 1-5 The technical solutions in the embodiments of the present invention will be clearly and completely described.

[0020] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the present invention can be combined with each other.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] A reimbursement robot device based on AI large model for invoice discrimination includes a printing unit, a robotic arm gripper 4-component, a printing compartment 20, and a background linkage control module. The printing unit is connected to the AI ​​processing unit and approval system of the reimbursement robot. The robotic arm gripper 4-component is adapted to the printing compartment 20, the temporary storage compartment 12, and the archiving compartment 17 to realize invoice grabbing and transfer.

[0023] The printing unit is equipped with an encrypted data receiving interface, which can receive structured JSON data output by the AI ​​processing unit and approval instructions from the approval system, and automatically generate electronic invoices and expense summary sheets.

[0024] The printing compartment 20 is equipped with an arrival detection sensor. When the printed part is output to the printing compartment 20, the sensor sends a transfer signal to the robotic arm gripper 4 assembly, triggering the document merging operation.

[0025] The robotic arm gripper 4 assembly includes a horizontal drive motor 5, a vertical drive motor 6, a horizontal guide rail 7, a vertical guide rail 8, a gripper 9, and a gripper 9 drive motor 10, which can accurately grasp and merge the original paper documents and the output of the printing unit in the temporary storage compartment 12.

[0026] The back-end linkage control module connects with the AI ​​trend prediction and analysis system, which can receive multi-dimensional analysis data and support printing structured management reports containing trend curves, pie charts, and prediction tables.

[0027] The printing unit's printing trigger logic is linked to the reimbursement approval process. The printing program is automatically started only after the AI ​​intelligent review passes and compliance verification and duplicate reimbursement detection are completed.

[0028] The robotic arm gripper 4 components are adapted to the partition grid design of the filing compartment 17, which can accurately place the merged paper documents and printouts into the corresponding filing compartments according to department / month classification rules, with a positioning error of ≤1mm.

[0029] After receiving the order delivery completion signal, the background linkage control module sends a gripping and transfer command to the robotic arm gripper 4 component. Driven by the coordinated action of the horizontal drive motor 5 and the vertical drive motor 6, the robotic arm gripper 4 component moves precisely along the horizontal guide rail 7 and the vertical guide rail 8. Simultaneously, the gripper 9 drive motor 10 controls the gripper 9 to open, using infrared positioning technology to pinpoint the exact location of the document within the order delivery compartment 1. The document is then stably gripped using a combination of vacuum adsorption and gripper 9. Subsequently, the robotic arm transfers the document to the waiting-to-scan position 15 of the fully automatic scanning position 13 according to a preset trajectory and places it smoothly. After placement, the robotic arm gripper 4 component resets to its initial standby position and simultaneously sends a "scanning ready" signal to the fully automatic scanning position 13. Upon receiving a signal, the fully automatic scanning station 13 activates the fan 21 of the scanning and transporting mechanism 19. This fan uses suction to flatten and fix the document at the scanning station 15, preventing wrinkles or shifting during scanning. Subsequently, the high-definition camera 14 activates, capturing images of the document page by page. During this process, the fan 21 continuously operates to keep the document flat. After acquisition, the scanning and transporting mechanism 19 smoothly transports the document from the scanning station 15 to the scanning station 16 for temporary storage. Simultaneously, the acquired image data is uploaded to the AI ​​processing unit via an encrypted channel. The AI ​​processing unit runs the ViT+Qwen multimodal large model. The ViT image encoder encodes the document image in blocks, extracting global visual features. Combined with the semantic understanding capabilities of the Qwen large language model, a cross-attention mechanism is used to achieve semantic alignment between the image and text. Then, position-aware encoding records the coordinates of OCR text blocks to assist spatial reasoning. Finally, the structured information of the document is extracted, generating standard-format JSON data and feeding it back to the backend system.

[0030] The applicant retrieves and associates the identified structured data of the invoices through the web or mobile app of the backend system, fills in the relevant supplementary information, and submits an electronic reimbursement application. The backend system automatically binds the electronic reimbursement application with the corresponding invoice image and structured JSON data, generates a complete reimbursement form, and pushes it to the AI ​​intelligent review module. The AI ​​intelligent review module, based on a preset strategy engine, performs compliance verification on the reimbursement form (including invoice authenticity and filling specifications), and simultaneously compares historical invoice features using a vector database to detect duplicate reimbursements (false positive rate <0.1%). During the review process, finance personnel can view the original invoice image, key field location, and parsing results through a visual interface, and manually review and intervene in abnormal documents. All review operations are logged in real time to ensure the integrity of audit traceability. After approval, the backend system sends a "Review approved, print allowed" instruction to the backend linkage control module.

[0031] After receiving the approval instruction, the back-end linkage control module immediately sends a print instruction to the printing unit, simultaneously transmitting the corresponding structured JSON data and reimbursement summary information. The printing unit receives the encrypted data via the network port, automatically parses and formats the electronic invoice content and summary sheet, and starts the printing program. Once printing is complete, the printed document slides into the printing chamber 20 via the built-in transmission mechanism. The positioning sensor in the printing chamber 20 detects the printed document in real time and immediately sends a "printing complete, ready for transport" signal to the robotic arm gripper 4 component.

[0032] After receiving the transfer signal, the robotic arm gripper 4 component restarts the horizontal drive motor 5 and vertical drive motor 6, moving along the guide rail to the temporary storage compartment 12 to precisely grasp the previously stored original paper documents. Subsequently, the robotic arm transfers to the printing compartment 20, grasps the printed electronic documents and summary sheets, and uses the gripper 9 to precisely align and merge the original paper documents with the printed documents (positioning error ≤1mm). After merging, according to the classification instructions (by department / month) issued by the backend system, the robotic arm moves to the corresponding partition grid in the archiving compartment 17, smoothly placing the merged documents into the grid to complete the archiving operation. After archiving, the robotic arm gripper 4 component returns to its initial position, awaiting the next operation instruction.

[0033] The backend AI trend prediction and analysis system collects structured historical expense reimbursement data (invoice type, amount, department, date, etc.) and process data (reimbursement status, approval time, etc.) at preset cycles (e.g., weekly / monthly). After data preprocessing (cleaning outlier data), it uses the ARIMA time series algorithm to analyze time trends, the K-means clustering algorithm to count the proportion of invoice types, and the correlation analysis of reimbursement preferences for each department to generate a structured management report containing trend curves, pie charts, and prediction tables. Managers can send report printing commands through the backend system. The backend linkage control module receives the command and transmits the report data to the printing unit. The printing unit prints the management report according to a preset format and outputs it to printing slot 20 for managers to use.

[0034] The end-to-end process of this invention is as follows: User -> Order Submission Warehouse: Tap "Submit Order" on the touchscreen. Document delivery warehouse -->> Robotic arm: Infrared positioning of document coordinates Robotic arm ->> Scanning position: Vacuum suction cup gripping and placement Scanning location -->> AI server: Upload image and trigger recognition (response < 3 seconds) AI server -->> Backend system: Returns JSON + coordinate-annotated map Robotic arm -> Temporary storage bin: Temporarily store documents by batch ID After approval -> Printer: Print electronic invoices + summary sheet Robotic arm -> Filing compartment: Merges paper documents and printouts (positioning error ≤ 1mm) In the multimodal recognition engine architecture of this invention, spatial semantic alignment embeds text coordinates (x_min, y_min, x_max, y_max) into the feature vector through position-aware encoding, enabling the model to understand spatial logic such as "the amount is in the lower right corner"; dynamic prompt fine-tuning supports dynamic adjustment of the output format by natural language commands (example command: "extract: {payee, amount, invoice date}"), without the need to retrain the model.

[0035] Intelligent review and prediction module Duplicate expense detection: Based on comparison of historical invoice features with a vector database, the false alarm rate is <0.1%; Trend Forecast Report: LSTM model analyzes cost fluctuations at the department / project level and outputs visualized forecast curves.

[0036] Based on the Qwen large model, an AI analysis model is fine-tuned to achieve trend prediction in three steps: I. Data Preprocessing: Cleaning abnormal data (such as invalid invoices and duplicate submissions), and structuring the data by "time-department-invoice type"; II. Multi-dimensional analysis: Time trend: Using the ARIMA time series algorithm, we analyze the changes in reimbursement amount over the past 6 months and predict the amount range for the next month (e.g., "The estimated reimbursement amount in November is 120,000-150,000 yuan, an increase of 8%-12% month-on-month"). Type proportion: High-frequency invoice types are statistically analyzed using the K-means clustering algorithm (e.g., "VAT invoices account for 65%, and travel expense receipts account for 20%"). Departmental characteristics: Correlation analysis of reimbursement preferences of various departments (e.g., "equipment purchase invoices account for 40% of R&D department expenses, and travel expenses account for 55% of marketing department expenses"); III. Report Generation: Automatically generates structured reports containing trend curves, pie charts, and forecast tables, supporting push to enterprise management terminals (Web / APP) or outputting paper reports through the hardware execution system's printing component.

[0037] This invention provides a full-process reimbursement robot system that integrates physical automation and multimodal AI recognition. A multi-axis robotic arm enables closed-loop circulation of documents in the document submission area, scanning area, temporary storage area, and archiving area; a ViT+Qwen multimodal large model is used to achieve end-to-end structured extraction of documents; and combined with a back-end AI review engine and multi-dimensional analysis module, the entire chain of automation is completed, from physical document storage, intelligent recognition, compliance review to predictive archiving.

[0038] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between them; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances. Obviously, the embodiments described above are only some embodiments of this invention, not all embodiments. The accompanying drawings show preferred embodiments of this invention, but do not limit the patent scope of this invention. This invention can be implemented in many different forms; on the contrary, the purpose of providing these embodiments is to make the disclosure of this invention more thorough and complete. Although the invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the patent protection scope of this invention.

Claims

1. A reimbursement robot device based on AI large-scale model for invoice discrimination, characterized in that, It includes a printing unit, a robotic arm gripper (4) assembly, a printing compartment (20), and a background linkage control module. The printing unit is connected to the AI ​​processing unit and the approval system of the reimbursement robot. The robotic arm gripper (4) assembly is adapted to the printing compartment (20), the temporary storage compartment (12), and the archiving compartment (17) to realize the grabbing and transfer of invoices.

2. The reimbursement robot device based on AI large-scale model for invoice discrimination according to claim 1, characterized in that, The printing unit is equipped with an encrypted data receiving interface, which can receive structured JSON data output by the AI ​​processing unit and approval instructions from the approval system, and automatically generate electronic invoices and reimbursement summary sheets.

3. The reimbursement robot device based on AI large-scale model for invoice discrimination according to claim 1, characterized in that, The printing compartment (20) is equipped with an arrival detection sensor. When the printed part is output to the printing compartment (20), the sensor sends a transfer signal to the robotic arm gripper (4) assembly to trigger the document merging operation.

4. The reimbursement robot device based on AI large-scale model for invoice discrimination according to claim 1, characterized in that, The robotic arm gripper (4) assembly includes a horizontal drive motor (5), a vertical drive motor (6), a horizontal guide rail (7), a vertical guide rail (8), a gripper (9), and a gripper (9) drive motor (10), which can accurately grasp and merge the original paper tickets and the output of the printing unit in the temporary storage bin (12).

5. The reimbursement robot device based on AI large-scale model for invoice discrimination according to claim 1, characterized in that, The background linkage control module is connected to the AI ​​trend prediction and analysis system, which can receive multi-dimensional analysis data and support printing structured management reports containing trend curves, pie charts, and prediction tables.

6. The reimbursement robot device based on AI large-scale model for invoice discrimination according to claim 1, characterized in that, The printing unit's printing trigger logic is linked to the reimbursement approval process. The printing program is automatically started only after the AI ​​intelligent review passes and compliance verification and duplicate reimbursement detection are completed.

7. The reimbursement robot device based on AI large-scale model for invoice discrimination according to claim 1, characterized in that, The robotic arm gripper (4) assembly is adapted to the partition grid design of the filing compartment (17), and can accurately place the merged paper documents and printouts into the corresponding filing compartments according to the department / month classification rules, with a positioning error of ≤1mm.

8. The usage process of the reimbursement robot device based on AI large model for invoice discrimination according to claim 1 is as follows: The person seeking reimbursement clicks the "Submit" button on the touchscreen (11), the electronic lock (2) of the submission compartment (1) unlocks, and the compartment door closes after the paper invoice is placed in; the robotic arm gripper (4) component grabs the invoice through infrared positioning and transfers it to the fully automatic scanning position (13) to complete image acquisition, and the image is uploaded to the AI ​​processing unit for structured extraction; the person seeking reimbursement submits an electronic reimbursement application through the back-end system, the system automatically associates the structured data to generate a reimbursement form, and after the AI ​​intelligent review is approved, the back-end linkage control module sends a printing instruction to the printing unit; the printing unit prints the electronic invoice and summary sheet, outputs it to the printing compartment (20) and triggers the sensor signal; the robotic arm gripper (4) component takes out the original paper invoice from the temporary storage compartment (12), merges it with the printed parts in the printing compartment (20), and transfers it to the corresponding grid of the archiving compartment (17) according to the department / month classification rules; the back-end AI trend prediction and analysis system can trigger the printing of management reports as needed, and the printing unit outputs paper reports.

9. The system workflow of the reimbursement robot device based on AI large model for invoice discrimination according to claim 1: After receiving the invoice completion signal, the background linkage control module sends a gripping and transfer instruction to the robotic arm gripper (4) component; the robotic arm gripper (4) component moves precisely along the horizontal guide rail (7) and the vertical guide rail (8) through the coordinated drive of the horizontal drive motor (5) and the vertical drive motor (6), while the gripper (9) drive motor (10) controls the gripper (9) to open, and locks the specific position of the invoice in the invoice delivery compartment (1) with the help of infrared positioning technology, and stably grips the invoice by combining vacuum adsorption and gripper (9) clamping; then, the robotic arm transfers the invoice to the waiting position (15) of the fully automatic scanning position (13) according to the preset trajectory, and places it stably. After placement, the robotic arm gripper (4) component resets to the initial standby position, and sends a "scanning ready" signal to the fully automatic scanning position (13); After receiving the signal, the fully automatic scanning position (13) starts the fan (21) of the scanning and transporting mechanism (19) to flatten and fix the ticket to be scanned (15) by suction force, so as to avoid wrinkles or displacement during scanning; then, the high-definition camera (14) starts to capture images of the ticket page by page. During the acquisition process, the fan (21) continues to work to keep the ticket flat. After the data collection is completed, the scanning and transporting mechanism (19) smoothly transports the invoice from the scanning position (15) to the scanning position (16) for temporary storage. At the same time, the collected image data is uploaded to the AI ​​processing unit through an encrypted channel. The AI ​​processing unit runs the ViT+Qwen multimodal big model, encodes the invoice image into blocks through the ViT image encoder, extracts global visual features, combines the semantic understanding ability of the Qwen big language model, achieves image-text semantic alignment through the cross-attention mechanism, and then records the coordinates of the OCR text blocks through position-aware encoding to assist spatial reasoning. Finally, the structured information of the invoice is extracted, and standard format JSON data is generated and fed back to the backend system. The person seeking reimbursement retrieves and associates the structured data of the invoice that has been identified through the Web or APP terminal of the backend system, fills in the relevant supplementary information for reimbursement, and submits an electronic reimbursement application. The backend system automatically binds the electronic reimbursement application with the corresponding invoice image and structured JSON data, generates a complete reimbursement form, and pushes it to the AI. The intelligent audit module, based on a preset strategy engine, performs compliance checks on expense reports and compares historical invoice features against a vector database to detect duplicate reimbursements. During the audit process, finance personnel can view the original images of invoices, locate key fields, and analyze the results through a visual interface, and manually review and intervene in abnormal documents. All audit operations are logged in real time to ensure the integrity of audit traceability. After the review is approved, the backend system sends a "Review approved, printing allowed" instruction to the backend linkage control module; After receiving the approval instruction, the back-end linkage control module immediately sends a printing instruction to the printing unit, and at the same time transmits the corresponding structured JSON data and expense summary information. After receiving encrypted data through the network port, the printing unit automatically parses and typeset the electronic invoice content and summary sheet format, and starts the printing program. After printing is completed, the printed part slides into the printing chamber (20) through the built-in transmission mechanism. After the printing chamber (20) is equipped with a positioning detection sensor, it immediately sends a "printing completed, ready for transfer" signal to the robotic arm gripper (4) component.

Images