FAQs

1. Is the system open source? Can secondary development be carried out?

• Answer: MRDVS's software package disassembly algorithm is encapsulated in the camera, and some parameters can be opened to the public. Users can adjust parameters and optimize them as needed. However, the underlying code is not open source. The system supports secondary development, provides open interfaces such as stream interfaces, and supports programming languages such as C++ and Python.

2. What types of stacking are supported by the soft package de-stacking scheme?

• Answer: The soft package disassembly solution can support various stacking types, such as five-pattern stacks, spiral stacks and three-pattern stacks, etc. Through visual algorithms, the system can flexibly identify different forms of soft packages and customize and adjust them according to customer on-site needs.

3. How does the visual algorithm work in soft package de-stacking?

• Answer: MRDVS's soft package disassembly algorithm uses 3D vision technology, combined with image recognition and coordinate conversion, to quickly locate the position and posture of the soft package and communicate directly with the robotic arm. The system has a built-in hand-eye calibration function, supports four-point calibration, and is convenient for the robotic arm to grasp and dock accurately. The algorithm has been widely used in practical projects, such as one-to-two, and one-to-three schemes.

4. Does high reflection or oil film on the surface of the soft package affect recognition?

• Answer: In certain situations, such as high reflectivity or thick oil film on the surface of the soft package, it may have some impact on the imaging effect of the camera. However, such situations are rare, and the recognition of most ordinary soft package scenes will not be affected.

5. How to integrate with the robotic arm? How to output the position signal to the robotic arm?

• Answer: The system is connected to the robotic arm through communication methods such as TCP/IP. The coordinates output by the vision system have been converted into a coordinate system that matches the robotic arm, reducing intermediate conversion steps. The system is compatible with most industrial robotic arm brands on the market, and customers only need to ensure that the robotic arm has a communication module.

• The visual system outputs the pose information of the target package, which is transmitted to the control cabinet of the robotic arm. The robotic arm captures the data based on the pose data. Different brands of robotic arms may require specific format data, and the system will configure it accordingly as needed.

6. Is Ethernet/IP protocol supported?

• Answer: Currently, the system only supports TCP/IP communication protocol and does not support Ethernet/IP protocol. If there is an Ethernet/IP requirement, additional development and adaptation work may be required.

7. How to perform hand-eye calibration?

• Answer: MRDVS provides a simplified four-point calibration method. Users can flexibly choose the calibration points on the four soft packages, and the system will automatically calculate the conversion matrix between the camera and the robotic arm. This is more convenient than the traditional nine-point or 64-point calibration.

8. How fast is the real-time processing speed of soft package unstacking?

• Answer: The system processing speed is about 2 seconds, and the specific time depends on the configuration of the camera and the industrial computer. The M4 Pro camera has a built-in computing platform with a computing power of up to 6Tops, which can independently complete the algorithm calculation of soft package disassembly, saving some of the calculation work of the industrial computer.

9. What is the connection between the M4 Pro camera and the industrial computer?

• Answer: The M4 Pro camera has a built-in computing platform that can directly output visual recognition results to the control system. The system can also choose to run algorithms on the industrial computer, depending on the customer's needs. By simplifying the configuration of the industrial computer, users can reduce hardware costs, but it is recommended to keep the industrial computer for debugging and displaying segmentation results.

10. Does the soft package disassembly plan need to be customized according to different customers?

• Answer: Yes, although the algorithm for soft package unstacking is very mature, the specific scenarios of different customers (such as stacking type, and soft package material) may require some customized adjustments. The environment of each site is different and needs to be adapted according to the specific situation.

11. Does the system support storing and reusing data from different projects?

• Answer: The system supports storing the visual data of each project on the server, and the previous algorithms and parameters can be reused in subsequent similar scenes to reduce the time of reconfiguration and training.

12. How to deal with the stack type change of different customer sites?

• Answer: If the stack type on-site changes, such as from a five-pattern stack to a spiral stack, the algorithm needs to be adjusted. This part requires algorithm adaptation but does not involve complete retraining. Existing 3D vision algorithms can handle most changes.