🛠V1 Pro Visual Mapping Deployment Instructions

Overview

Key Features

• Overhead view positioning, repeatable positioning accuracy to the point ±1cm

• For different ceiling heights, there are 6-meter and 12-meter versions according to the focal length.

• Built-in arithmetic, no need for an external industrial computer

• Positioning mode inputs robot odometer data and directly outputs robot’s pose.

• Default with supporting LiDAR to assist in map building, and can also import the laser data through the SDK.

Coordinate system of the camera

Method of use

• Windows host computer. Visualize and operate the camera building and positioning processes through the camera host software.

• SDK. When practically deployed, it can be developed by calling the camera SDK to realize various functions.

Application scenario

• Overhead view navigation, horizontal mounting required

• Use either the 6-meter or 12-meter version of the camera depending on the actual ceiling height to achieve the best positioning results.

• Suitable for overhead view scenes with rich visual features, large areas with fewer visual features or solid colors can not be used and can be improved by applying reflective stickers.

• The camera is mounted reliably, with no obstruction of vision, and operates in a relatively smooth environment, free of visible dust.

• Low-speed and smooth operation when building the map, the speed does not exceed 0.2m/s; the speed does not exceed 1m/s when positioning, and the frame rate of the returned positioning results is about 10Hz.

Mapping process

1. The odometer is necessary for map building and positioning, and odometer data are absolute coordinates;

2. Single line laser data is needed to assist positioning when building maps. Some better scenarios do not rely on laser data to assist building maps. At this time, virtual laser data can be input. Please contact the relevant technical support staff;

3. If you need to unify the coordinate system of the laser map and the visual map (the positioning results of the visual map and the laser map can not be completely overlapped, and there is usually a deviation of a few centimeters), you need to pass in the laser pose of the robot when building the map, and then you can input the virtual laser data;

4. You can use the map building tool to build maps offline. Some scenarios can not be realized to build maps or have problems. Please contact the relevant technical support staff;

5. Virtual data indicates that there is real-time data coming in, but there can be no requirement for the data to be real data from the actual scene, For example, virtual laser data where all range values are set to zero；

6. If no actual laser is actually input, no laser raster map is provided and no laser raster map is displayed in the host computer display interface;

7. After enabling mapping, data recording will only start when the robot is in motion; no data will be recorded while it remains stationary;

8. When exporting data, please turn off the map building enable in advance and wait patiently for the data to be exported;

9. Plan the approximate recording path in advance, keeping the robot moving at a low and steady speed (<0.2 m/s), with straight-line motion and right-angle turns. Avoid overlapping paths or random shaking. Ensure that paths in the same direction maintain about one-third overlap in the field of view. In general, keep a parallel distance of around 2 meters, so as to maximize coverage of the mapping area or the robot’s travel path. The ideal situation is illustrated in the figure below:

Localization process

1. Please import the built visual map before positioning, and switch to the corresponding map when positioning;

2. The initial positioning needs to send a repositioning request, and the correct positioning result can only be received after successful repositioning;

3. Enabling mapping is required when sending a relocalization request.

4. The positioning module records the pose in the normal positioning state, and if the camera is turned off and then restarted, it automatically repositions itself according to the last recorded pose;

5. Knowing the approximate coordinates of the robot in the vision map is required for retargeting.(x, y, yaw), starting point for recording data on a build map(0, 0, 0)perform repositioning. Generally，x、ydeviation in less than 30cm，yawangular deviation less than 20°；

6. In the normal positioning state, the robot coordinates at the fixed station are recorded, and repositioning can be performed at that fixed station using the recorded robot coordinates;

7. If repositioning fails, please move the robot's actual position or fine-tune the repositioning coordinates, or move it to the area with rich visual features and reposition it again; if repositioning still fails, please check whether the SDK program, data communication, and camera output are working properly;

8. During normal positioning, it is necessary to run within the range of the camera's built-in map. Running in an area where no map has been built will output incorrect position information;

9. Factors such as unsatisfactory visual features in some areas, low match between overhead view height and camera focal length, etc. can affect the localization accuracy

Preparation work

• Check that the camera is properly connected and that the output and communication are normal

• If you are using a LIDAR kit, please connect it to the corresponding network port.

• The camera is mounted horizontally and the external reference relative to the center of the robot is determined according to the mounting position and orientation.[x, y, z, yaw, pitch, roll]

• The single-line lidar is mounted horizontally and provides an external reference of the lidar with respect to the center of the robot.[x, y, yaw]

• If you use both the Windows-side host computer and the SDK for the industrial control unit, you can connect the camera, the industrial control unit and the Windows computer via a switch for communication.

Note: The camera supports up to two accesses at the same time, e.g., it can be accessed through a single uplink and a single SDK at the same time, or through two uplinks at the same time

Example of Use

Build a map

Preparation work

1. Open the host computer program

2. Connecting the camera

3. Select the Application Algorithm

4. Selection of visual localization in application algorithms

5. At this time, you can check whether the data status of each sensor is normal through the visual localization view, if there is any abnormality, please check the communication configuration.

6. This is the state returned by the algorithm. When in the "Normal Positioning" state, the robot pose is returned and can be used for localization.

User input data

• Robot odometer data in absolute coordinate system, >30Hz

• LiDAR data (virtual data if LiDAR-equipped is used), >= 10 Hz

• Pose of the robot under the laser map (may not be entered if the coordinate system is not required to be harmonized with the laser map), >=10Hz

Recording data

• Configuration parameters
Map Name.(If not, use the default "example_map1".)
Camera relative to the robot's external reference.In meters and degrees
Lidar versus robotic external reference.In meters and degrees

• Recorded map path planning

• Activation chart

• Export data

Offline building maps

Importing visual maps

Localization

Preparation work

• Confirm that the camera and the communication is working.

• Map of your area has been imported into the camera

User input data

• After the successful completion of the map recording, map building, and map importing operations, the real sensor data can be transferred into the camera for positioning purposes according to the SDK sample program, and the data includes:
• Robot odometer data in absolute coordinate system, >30Hz

Enabling positioning

• Configuration parameters
• Map Name.(Use of already constructed maps for the corresponding area)
• Camera relative to the robot's external reference.In meters and degrees
• Lidar versus robotic external reference.In meters and degrees

• Enabling positioning

• Setting the positioning enable, i.e. enabling the positioning mode

Repositioning request

Normal positioning (math.)

UDP communication

Vision Module->Robot

• Localization value

• Repositioning results

Robot->Vision Module

• Speedometer (of a vehicle)

• Repositioning request

• Robot pose (to be passed in if the vision map and laser map coordinate systems need to be harmonized, otherwise it can be left out)

Parameter Configuration