> For the complete documentation index, see [llms.txt](https://mrdvs.gitbook.io/mrdvs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://mrdvs.gitbook.io/mrdvs/pallet-recognition-user-guide-eagle-m-series-camera.md).
# Pallet Recognition User Guide Eagle M Series Camera
## 1. Introduction
Pallet recognition is a solution developed by MRDVS using the in-house developed Eagle-M series cameras. It's designed for forklifts to seamlessly docking with pallets. The system boasts adaptive compatibility with a range of pallet specifications, eliminating the need for specific configurations. With dynamic detection capabilities, it can identify pallets on the go, eliminating the requirement for the vehicle to come to a complete stop. The system is versatile, supporting recognition in diverse scenarios like floor-to-floor transitions, storage locations, and shelves. The system can also adaptively recognize pallets of different specifications, including wooden and plastic pallets with different shapes, as shown below.
## 2. Eagle-M camera installation
As shown in the figure below, install the Eagle-M camera at a distance of 35cm to 45cm above the base panel of the fork arm, ensuring it is close to the center of the two fork tines (with a left-right deviation of no more than 5cm).
Figure 1: Installation example
## 3. Interfaces
### 3.1 Main Interface
After launching the software, the main interface shown below will appear. You can perform operations such as camera connection, camera calibration, and pallet recognition on this screen.
Figure 2: Main Interface
The interface includes the following functions:
| Function Module | Specific Functions/Description |
| --------------- | ------------------------------------------------------------------------------------------------------------------------------------ |
| Menu Bar | Includes basic operations, advanced applications, external calibration, camera operations, advanced parameters, and save parameters |
| Camera Settings | Includes scanning, embedded mode, camera on/off, real-time display, detection, projection, and camera calibration parameter settings |
| File Import | Supports importing calibration parameter files, recognition images, and folders |
| Log Printing | Prints detection results, camera firmware version, algorithm version, etc. |
| Image Display | Displays image data |
### 3.2 Menu bar
The menu bar includes five options: External Calibration, Camera Operations, Advanced Parameters, Save Parameters, and Languages. These options are accessible under Basic Operations and Advanced Applications
Figure 3: Menu bar
### 3.3 Basic Operations
| Menu | Option | Description |
| ---------------- | -------------------- | ---------------------------------------------------------------------------------------------------- |
| Basic Operations | External Calibration | Camera calibration, including camera height, lateral offset, pitch angle, yaw angle, and roll angle. |
| Basic Operations | Save Parameters | Save the configured parameters and send them to the camera. |
| Basic Operations | Camera Operations | Retrieve the camera IP, camera ID, and software logs. |
#### 3.3.1 External Calibration Module
Click External Calibration under Basic Operations. The external calibration interface will be displayed on the right side of the software window.
Figure 4: External calibration module
Click 'External Calibration' and then select 'Calibration Notes' to display the precautions for external calibration. Please read them carefully to avoid any issues during the process.
Figure 5: External calibration notes
Information verification at the Bottom of \[Calibration Notes]: Customer Name: It is recommended to use the format "Project Name + Forklift Number" for easier troubleshooting as shown in Figure 6. Minimum Recognition Distance (mm): Keep the default value as shown in Figure 6. (If the calibration distance is relatively short, adjust it to 1.2m.)
Figure 6 : Information verification
Rotation Center to Optical Center Distance: When set to 0, it assumes the optical center overlaps with the rotation center. If they are not aligned, enter the actual distance. You must enter the actual distance from the camera's optical center to the forklift's rotation center.
Fork Arm Position (Depth to Tray Edge): This parameter defines the fork insertion depth, which means the distance forklift arms need to travel after reaching the front edge of the pallet. It acts as a safety margin to ensure the forks are fully and securely inserted under the pallet.
Figure 7: Parameters
Pallet Teaching (It is recommended to perform two calibrations): \[Pallet Teaching (Near)]: Position the pallet 1300mm away from the camera to perform near-end calibration. This step determines the pallet's height, lateral offset, heading angle, and roll angle. \[Pallet Teaching (Far)]: After near-end calibration, proceed with far-end calibration (secondary teaching). For example, if the camera recognizes the pallet distance as 1300mm during near-end calibration, drive the forklift straight backward by approximately 700mm–1100mm without any angular or lateral movement. Perform far-end calibration when the distance to the pallet reaches 2.1–2.4m. This step determines the camera's pitch angle. Note: \[Pallet Teaching (Far)] tab is only available after \[Pallet Teaching (Near)] is successful.
Figure 8: Pallet Teaching
Clicking 'Reset Calibration' will restore all external calibration parameters to zero.
Figure 9: Reset Calibration
Lateral deviation means the lateral deviation between the camera's optical center and forklift's coordinate center, and it is automatically determined through the calibration process.
#### 3.3.2 Camera Operations
The \[Camera Operations] section displays the camera's internal parameters.
Figure 10: Camera Operations
\[Camera IP]: Displays the IP address of the connected camera.
\[Camera ID]: Displays the ID of the connected camera.
\[Get lanxin vision log]: Click this button to save the camera detection log to the platform folder as file lanxin\_vision\_log.
\[Camera docking system Configuration]: This section is for administrators only to configure the docking system. Do not modify these settings.
Figure 11: Camera docking system Configuration
\[Advanced Parameters Configuration]: By clicking \[Pull], the current advanced parameters will be loaded into the platform folder as /docking\_config.json. Next, click \[Configure the algorithm parameters] to show and adjust the parameters within the platform, and finally click \[Upload].
Figure 12: Advanced Parameters Configuration
### 3.4 Advanced applications Interface
#### 3.4.1 Advanced parameters
Camera configuration file: The camera configuration file "pallet\_settings.json", which contains the camera algorithm parameters, will be downloaded and loaded into PalletPro.
Camera IP/ID: Displays the IP address and ID of the connected camera.
Leg width: The width of the legs on the side facing the camera.
Pallet width: The width of the pallet on the side facing the camera.
Crossbar width: The width of the pallet crossbar on the side facing the camera.
| Setting | Mode Option | Parameter / Description |
| ---------------------------- | ------------------ | --------------------------------- |
| **Leg Width Selection** | Standard Mode | 10-15cm |
| | Thin Fork Mode | 5cm |
| | Ultra-thin Pallet | 3-4cm |
| | Wider | 15cm |
| | Extra Wide | 20cm+ |
| | 8 - 32 cm | 8-32cm |
| | Custom number | Custom number |
| **Pallet Width Selection** | Standard Mode | 0.7-1.3m |
| | Wide Pallet | 1.1-1.6m |
| | Custom number | Custom number |
| **Crossbar Width Selection** | Standard (Default) | No changes needed for general use |
| | Thinner | 80% |
| | Very Thin | 60% |
| | Sparse | 10% |
Figure 13: Advanced parameters 1
**Camera algorithm parameters which can be modified in advanced settings:**
| Parameter | Value/Status | Description |
| ----------- | ---------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| R | \[0.0, 0.0, 0.0] | Angle parameters; the first group is 0.0 pitch angle, the second group is 0.0 heading angle, and the third group is 0.0 roll angle. |
| t | \[0.0, 0.0, 0.0] | Offset parameters; the first group is 0.0 lateral offset parameter, the second group is 0.0 (not effective), and the third group is 0.0 depth offset (forklift front and rear direction). |
| two\_leg | 0 | Whether to enable two legs; if there is no two-legged pallet in the application, it is not recommended to enable it. 0 is closed, 1 is open. |
| ground\_y | 331 | Installation height in mm. |
| min\_leg | 30 | Minimum leg width supported. |
| max\_leg | 170 | Maximum leg width supported. |
| thickness | 15 | Algorithm extraction thickness. |
| x-client | "nonetest" | Customer name (and forklift serial number) for future maintenance. |
| orientation | 0 | Installation method; 0 is upright, 2 is inverted; 1, 3 are side-mounted. |
| min\_z | 1000 | Minimum range for recognition; it does not recognize within 1000. |
| max\_z | 3200 | Maximum range for recognition; it does not recognize beyond 3200. |
| min\_x | -980 | Horizontal direction filtering; not recognized beyond 980mm to the left of the center. |
| max\_x | 980 | Horizontal direction filtering; not recognized beyond 980mm to the right of the center. |
Expanded Parameters: Expansion of existing parameters, but modification is not recommended.
Figure 14: Advanced parameters 2
#### 3.4.2 Display 3D
Select 'Advanced Application' and click the \[Display 3D] tab to display the current 3D point cloud. You can adjust the page dimensions by dragging its edges, hold down the left mouse button and use the scroll wheel to zoom in or out，or hold down the left mouse button and drag to rotate or tumble the image. This feature can also be used during offline testing to check the completeness of the pallet point cloud.
Figure 15: Display 3D image
#### 3.4.3 Offline Testing Module
Select \[Advanced Application] and click the \[Offline Testing] tab. To recognize offline images, import a JSON calibration parameter file, a single image or an image folder into PalletPro, choose \[File] or \[Folder] option, and then click the \[Start Detection] button.
Figure 16: Offline testing module
Normally, a JSON parameter file is automatically generated into the PalletPro application folder upon connecting to the camera. However, if you wish to test the PalletPro algorithm offline, you can create your own JSON parameter file following the format as shown below. Place this file in the root directory of the application, and specify its path in the \[Parameter file] field as shown in Figure 16.
```json
{
"R": {
"key": "R",
"value": [0.0, 0.0, 0.0],
"level": 1,
"desc": "euler angles"
},
"t": {
"key": "t",
"value": [0.0, 0.0, 0.0],
"level": 1,
"desc": "translation vector"
},
"ground_y": {
"key": "ground_y",
"value": -100,
"level": 1,
"desc": "地面高度，单位毫米"
},
"fork_space": {
"key": "fork_space",
"value": 20,
"level": 1,
"desc": "相机与叉臂间隙"
},
"thickness": {
"key": "thickness",
"value": 15,
"level": 1,
"desc": "算法提取厚度"
},
"multi_level": {
"key": "multi_level",
"value": 0,
"level": 1,
"desc": "是否支持多托盘检测"
},
"x-client": {
"key": "x-client",
"value": "none",
"level": 1,
"desc": "客户名称"
},
"fork_angle": {
"key": "fork_angle",
"value": 0,
"level": 1,
"desc": "叉臂与叉车直线运动方向的夹角，一般为0"
},
"two_leg": {
"key": "two_leg",
"value": 0,
"level": 1,
"desc": "是否支持双腿"
},
"fork_insert": {
"key": "fork_insert",
"value": 0,
"level": 1,
"desc": "叉车旋转中心到托盘前沿后，需要继续插入的距离"
},
"orientation": {
"key": "orientation",
"value": 0,
"level": 1,
"desc": "安装方式；0为正装，2为倒装；1、3侧装"
},
"min_z": {
"key": "min_z",
"value": 800,
"level": 1,
"desc": "z方向最小值"
},
"max_z": {
"key": "max_z",
"value": 3200,
"level": 1,
"desc": "z方向最大值"
},
"min_x": {
"key": "min_x",
"value": -980,
"level": 1,
"desc": "水平方向最小值"
},
"max_x": {
"key": "max_x",
"value": 980,
"level": 1,
"desc": "水平方向最大值"
},
"quad": {
"key": "quad",
"value": 0,
"level": 2,
"desc": "是否支持四腿托盘"
},
"scan_step": {
"key": "scan_step",
"value": 1,
"level": 2,
"desc": ""
},
"min_line_count": {
"key": "min_line_count",
"value": 60,
"level": 2,
"desc": ""
},
"log_level": {
"key": "log_level",
"value": 0,
"level": 2,
"desc": ""
},
"cutting_height": {
"key": "cutting_height",
"value": 50,
"level": 2,
"desc": ""
},
"space_dist": {
"key": "space_dist",
"value": 25,
"level": 2,
"desc": ""
},
"max_legwidth_diff": {
"key": "max_legwidth_diff",
"value": 50,
"level": 2,
"desc": "左右腿最大差异"
},
"min_hole": {
"key": "min_hole",
"value": 280,
"level": 2,
"desc": ""
},
"max_hole": {
"key": "max_hole",
"value": 700,
"level": 2,
"desc": ""
},
"leg_dist_min": {
"key": "leg_dist_min",
"value": 500,
"level": 2,
"desc": "<双腿检测>左右腿最小距离"
},
"leg_dist_max": {
"key": "leg_dist_max",
"value": 1600,
"level": 2,
"desc": "<双腿检测>左右腿最大距离"
},
"denoise_search_radius": {
"key": "denoise_search_radius",
"value": 25,
"level": 2,
"desc": ""
},
"denoise_min_pts": {
"key": "denoise_min_pts",
"value": 20,
"level": 2,
"desc": ""
},
"bar_check_dist": {
"key": "bar_check_dist",
"value": 100,
"level": 2,
"desc": ""
},
"bar_search_y": {
"key": "bar_search_y",
"value": 12,
"level": 2,
"desc": ""
},
"max_hole_diff": {
"key": "max_hole_diff",
"value": 91,
"level": 2,
"desc": "最大托盘孔距差异"
}
}
```
Figure 17: Offline testing module
Then add the target pointcloud image to the PalletPro folder and specify its path in the \[Input file] section. Then, click \[Start Detection].
Figure 18: Add offline pointcloud image
Or you can also specify the path of the folder containing the images in the \[Folder] section, enter the starting image number you want to test from, and then, click \[Start Detection], and the Algorithm will automatically find the picture inside the folder.
Figure 19: Add offline pointcloud image folder
**Example:**
Figure 20: Example format
Note: The offline testing module will automatically perform the detection, so you don't need to do \[Pallet Teaching (far)] and \[Pallet Teaching (near)] separately if the forklift's moving range is within 2000 mm of the pallet.
## 4. Initial setup and Calibration going through
Install LxCameraViewer and PalletPro software on a Windows operating system, set up and prepare the software before testing.
### 4.1 Camera installation
As shown in the figure below, it is advisable to mount the camera at a height of around 400mm. Position the pallet directly in front of the camera at a distance of 1300mm. Note: Ensure that the front surface of the camera is positioned close to the edge of the object supporting the camera to prevent any blockage of emitted or reflected infrared light.
Figure 21: Camera installation example
### 4.2 Camera configuration (MRDVS)
1. Before connecting the camera, please complete the following setup:
* The factory default IP address of the camera is 192.168.100.82. Ensure that your local network port (or Ethernet adapter) is configured to be in the same subnet before establishing a connection.
* Disable the firewall.
* For more details about IP and Firewall configuration, please see the LxCameraViewer use manual at
Figure 22: IP configuration
2. Open the LxCameraViewer to test the camera view. Click the \[3D settings] tab. The default camera high exposure value is 650, and the low exposure value is 200. Adjust the camera exposure value based on different application scenarios.
Figure 23: Camera configuration
4. Select the pallet positioning algorithm and set the working mode to \[Always-On]. Note: LxCameraViewer is only used for configuring camera internal parameters, while PalletPro handles all other algorithm parameters. For better recognition performance, we recommend upgrading to the latest version of PalletPro via a software update.
Figure 24: Camera configuration
### 4.3 Pallet algorithm configuration (PalletPro)
1. Before opening PalletPro, exit LxCameraViewer first. Then open the Pallet Recognition software PalletPro.
2. After opening PalletPro, it will automatically scan the camera. Click on the \[start camera] tab to open and obtain the camera data stream.
Figure 25: Start Camera
3. Click on \[parameter settings] tab, which shows the current position of the camera. If the parameters are all default 0 while connecting the PalletPro to the camera first time, click the \[auto-adjusting high] button to get the approximate installation height of the current camera. Then fine-tune it, as shown in the diagram, adjusting the auxiliary line value to the intersection position of the pallet and the ground. Click save parameters, check \[real-time display], and \[real-time detection] to achieve pallet recognition display functionality.
Figure 26: Parameter Settings
### 4.4 Pallet calibration preparation
1. Preparation:
* A standard European blue plastic pallet, 1200mm×1200mm×150mm, as shown in the diagram.
* A tape measure.
* A forklift that can move normally
Note: The ground should be relatively flat during calibration. Before calibration, refer to the simple test method mentioned in 3.1 to set the approximate installation height of the camera. Then, when the forklift moves back and forth, both the far-end calibration and the near-end calibration can recognize the pallet stably.
Figure 27: Pallet
### 4.5 Near-end calibration
The near-end calibration scheme uses a so-called teaching calibration method. The forklift needs to manually control the docking of the pallet once. First, move the forklift under the pallet, and pick up the pallet once, which makes sure that the forklift is aligned with the pallet center. Then, lower the lift arm, and in a straight line, drive the forklift (the pallet is on the ground without lifting) to make the camera about 1300mm - 1500mm away from the front edge of the pallet (as shown in the diagram). At this time, click on the \[pallet teaching (near)] to calibrate the installation height, lateral offset, heading angle, and roll angle. After calibration, click \[Real-time display] and \[Detection] to continue detection.
Figure 28: Pickup first
Figure 29: Lift down and move backward
Figure 30: Lift down and move backward
### 4.6 Far-end calibration
The far-end calibration function is to calibrate the pitch angle of the camera installation. After completing the near-end calibration, ensure that the pallet position remains unchanged, control the forklift to further move away from the pallet, about 500mm, and click on the \[pallet teaching (far)] when the distance displayed on PalletPro is in the range of 1800mm-2000mm. This will calibrate the pitch angle of the camera installation. Click save parameters to complete the calibration. As shown in the diagram.
Note: The \[Pallet Teaching (far)] is not mandatory, if the forklift's moving range is within 2000 mm of the pallet.
Figure 31: Distance 2000mm
Figure 32: Far-end Calibration Distance
### 4.7 Calibration verification
Once the near-end and far-end calibrations are complete, you can begin verifying the calibration accuracy. Move the forklift back and forth between the near-end and far-end calibration points while monitoring the R output for distance (X value), lateral deviation (Y value), and angle. Calibration is considered successful when the deviation between the detected position and the actual position is within 10mm and the recognition is stable with no missed detections. The forklift can then be put into normal operation.
Example: x, y, theta = (-1987, 81, -2.1) This signifies that the pallet is 1987 millimeters away from the camera's optical center, shifted right by 81 millimeters, with an angular deviation of -2.1° (top view, counterclockwise direction). The data result is X: -1987; Y: 81; Theta: -2100.
Figure 33: Verification
## 5. Data communication methods
Here, we used NetAssist as the communication configuration tool for TCP and UDP. You can download it from , as shown in Figure 33.
You can also use NetAssistant from , which includes English documentation, but it has not been tested with this instruction.
### 5.1 API Invocation
The API invocation method supports C++, C#, Java, ROS1, ROS2, and other environments on Windows, Linux, ARM, etc. For the Windows environment, after installing the LxCameraViewer software, SDK files, and sample program codes are available in the installation path (e.g., D:\Program Files\Lanxin-MRDVS).
* Document Folder: Contains SDK and upper computer usage documentation.
* Firmware Folder: Stores camera firmware packages.
* Sample Folder: Holds sample code source files for various development environments.
* SDK Folder: Contains SDK library files, and environment variables can be configured according to the development environment.
### 5.2 UDP Communication
**Port Configuration:** Camera Host Port Number: 8000
**Sending Content:**
1. 0x60 0x04 0x00 0x00 means Start Recognition Task
2. 0x60 0x04 0x00 0x01 means End Recognition Task
Note: When using UDP communication, you must set the algorithm working mode to \[Always-On] Mode in the NetAssist.
### Response Data Structure
| Byte Position | Content |
| ------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| 1-2 | Code, return value, 16-bit unsigned integer (unsigned int). 0 indicates successful recognition; other values indicate recognition errors: 1: Camera open error 2: Tray positioning failed 3: Camera internal error |
| 3-4 | Total byte length (number of bytes), 16-bit unsigned integer (unsigned int). Currently 32 |
| 5-8 | Signs for tray recognition X, Y, theta, Z. 0x00 is positive, 0x01 is negative. |
| 9-12 | Tray front center point position X, 4-byte unsigned integer, unit: mm\*1000 |
| 13-16 | Tray front center point position Y, 4-byte unsigned integer, unit: mm\*1000 |
| 17-20 | Tray front center point angle theta (unit: degrees), 4-byte unsigned integer, unit: degrees\*1000 |
| 21-24 | Tray height (distance from camera optical center as origin to tray crossbeam, downwards is positive), 4-byte unsigned integer, unit: mm\*1000 |
| 25-32 | 8 bytes of extension, set to 0 |
### 5.3 TCP Communication
**Port Configuration:** As shown in the figure, the camera acts as a TCP Server on port 5501.
**Commands:** Send 0x60 0x04 0x00 0x00 to start the recognition task. Send 0x60 0x04 0x00 0x01 to stop the recognition task.
Note: When using TCP communication, you must set the algorithm working mode to \[Always-On] Mode in the NetAssist.
### Response Data Structure
| Byte Position | Content |
| ------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| 1-2 | Code, return value, 16-bit unsigned integer (unsigned int). 0 indicates successful recognition; other values indicate recognition errors: 1: Camera open error 2: Tray positioning failed 3: Camera internal error |
| 3-4 | Total byte length (number of bytes), 16-bit unsigned integer (unsigned int). Currently 32 |
| 5-8 | Signs for tray recognition X, Y, theta, Z. 0x00 is positive, 0x01 is negative. |
| 9-12 | Tray front center point position X, 4-byte unsigned integer, unit: mm\*1000 |
| 13-16 | Tray front center point position Y, 4-byte unsigned integer, unit: mm\*1000 |
| 17-20 | Tray front center point angle theta (unit: degrees), 4-byte unsigned integer, unit: degrees\*1000 |
| 21-24 | Tray height (distance from camera optical center as origin to tray crossbeam, downwards is positive), 4-byte unsigned integer, unit: mm\*1000 |
| 25-32 | 8 bytes of extension, set to 0 |
### 5.4 CAN Communication
**Note:** The current CAN protocol only supports the **M4/M4mega** series cameras.
The communication protocol uses **CANopen**, which is an application layer protocol running on top of the standard CAN bus. Its communication mode follows the "Master-Slave" architecture commonly used in industrial communication protocols. In this network, there is one Master station and multiple Slave stations. Slave stations do not communicate directly with each other; all communication occurs between the Master and the Slaves. The Master station is also referred to as the "Client," and the Slave station is referred to as the "Server."
The underlying communication uses the **CAN Standard Frame format**, meaning the CAN ID is 11 bits (0x000 \~ 0x7FF) and the data payload is 8 bytes.
#### CAN Communication protocol:
**1. CAN ID Usage Convention**
The 11-bit ID is divided into a **4-bit Function Code** and a **7-bit Node ID**. The CAN ID is also referred to as the **COB-ID**:
| Bit Position | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | ... (down to 0) |
| :------------: | :---------------: | :-: | :-: | :-: | :---------: | :-: | :-: | :-: | :-------------: |
| **Field Name** | **Function Code** | | | | **Node ID** | | | | |
**Note:** The CAN baud rate is defined as **250K**.
The **SDO (Service Data Object)** communication message is a basic protocol format in CANopen. The 8-byte data payload in the CAN message is defined by the communication protocol as shown below:
| Byte Index | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| ----------- | -------------------------- | -------------------- | --------------------- | ------------- | -------- | - | - | - |
| **Content** | **Command Specifier (CS)** | **Index (Low Byte)** | **Index (High Byte)** | **Sub-index** | **Data** | | | |
***
**2. Host to Device (Control Command)**
The main host sends the control command to the device via **COB ID: `0x212`**.
| Byte Index | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| ---------- | ------ | ------ | ------ | ------ | --------- | ------ | ------ | ------ |
| **Data** | `0x40` | `0x00` | `0x01` | `0x00` | MODE CTRL | `0x00` | `0x00` | `0x00` |
**Parameter Definition**
* **MODE CTRL**: Controls the switching state.
* `0`: OFF
* `1`: ON
**Reference Example** To enable the mode (send `1`), the hex string is: `40 00 01 00 01 00 00 00`
***
**3. Device to Host (Feedback Data)**
The device reports status and positioning data back to the main body using two different COB IDs. The variables `x`, `y`, and `yaw` represent actual physical values multiplied by 1000 (Integer type).
**A. Status Feedback** **COB ID: `0x192`** *Note: The `count` value increments by 1 for each query/response cycle.*
| Byte Index | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| ---------- | ----- | - | ------ | - | --- | --- | - | - |
| **Data** | count | | result | | yaw | ... | | |
> **Note:** Based on standard integer mapping, Bytes 0-1 likely represent `count`, Bytes 2-3 represent `result`, and Bytes 4-7 represent `yaw`.
**B. Position Feedback** **COB ID: `0x292`**
| Byte Index | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| ---------- | - | --- | - | - | --- | - | - | - |
| **Data** | x | ... | | y | ... | | | |
> **Note:** `x` and `y` coordinates are transmitted as integers (Actual Value × 1000).
## 6. Special Instructions
### 6.1 Black Pallet Recognition
Due to the TOF principle, M-series cameras are affected by the reflectivity of black pallets. Imaging effects for black pallets with 5% reflectivity can be effectively imaged within a range of 2m. During debugging, optimize the imaging effect of black pallets by adjusting the high integration time and low signal threshold through the upper computer software. The maximum supported lateral offset for black pallet recognition is ±400mm (distance from the camera center to the pallet center), and the maximum supported pallet rotation angle is ±10°.
### 6.2 Deployment of High-Level Pickup and Reflective Columns
When forklifts pick up and deposit goods in stereoscopic warehouse locations, navigation is achieved using reflective columns for positioning. However, the camera can be easily affected by reflective columns, causing recognition issues. During deployment, it is recommended to avoid the impact of reflective columns on pallet recognition. Deploy reflective columns in a way that they are staggered relative to pallet legs, as shown in the diagram.
Figure 34: Reflective Columns
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://mrdvs.gitbook.io/mrdvs/pallet-recognition-user-guide-eagle-m-series-camera.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.