JSON Settings

The JSON settings are the interface for all parameters. It contains information about cameras, robots, networks, objects, and pipelines (Fig. 2).
Exactly which parameters are needed depends on the application.

A flowchart showing settings options linked to various backends including camera, network, robot, objects, experience, geometry, graphics, and motion, which then connect to outcomes, UI, detection 2D, and detection 3D, with outcomes further branching into calibration and geometry

A project folder is created for each application. All necessary folders and files are stored in it.
The meaning of the standard elements of the JSON settings is explained in
Table 1: LumiScan VGR: Default elements of the JSON settings.

Table 1. Table 1: LumiScan VGR: Default elements of the JSON settings
Parameter	Description
cameras	All (hardware) cameras are listed under `cameras`. Usually only one camera is needed, but any number can be created. There are only two entries here: `id` (the serial number of the LumiScan camera) and `calibrationFile`.
recordings	Under `recordings`, you define how a LumiScan camera should capture new data. This includes settings such as filters, depth of field range and exposure settings. Each entry corresponds to one physical LumiScan camera. Any number of settings can be made for precise coverage of different scenes and objects.
networks	This lists all the neural networks for the application and defines how a single neural network is imported into LumiScan VGR. Multiple objects may require multiple networks, which in turn may require their own architectures.
robots	Contains information about the geometry of the robot and its spatial position relative to the cameras. A physical robot can be described by several robot objects. For example, several robot objects can be created to represent different grippers. Each entry must have a successful hand-eye calibration transformation and refer to a file that describes the robot. This file has the format `.urdf`. Additional 3D data is required for collision avoidance. Robots and grippers can also be visualized if CAD models are available in `.stl` format. If a collision with the box is to be calculated, it is also listed as a robot.
objects	Corresponds to an object to be detected. All objects to be picked up are listed individually here. The system works simultaneously with 2D and 3D data to recognize and localize objects in space. A distinction is made here between 3D object recognition and 2D recognition using neural networks. Which method is used depends on the application. They are not incompatible with each other, but can even be combined for more accurate detection. Note that we can also define abstract objects, such as composite part characteristics or classification results (e.g. `emptyContainer`).
pipelines	A pipeline defines how to process a point cloud to obtain object orientations (in the camera image). It contains steps for post-processing the point cloud, a definition of the ROI, and a list of objects to be calculated. Any number of pipelines can be created. They are selected by the PLC. Together with the robots and cameras, they are required to perform a gripping operation.
user	(optional) To prevent admin rights from being available to all users, they can be password protected (`adminPassword, manualModePassword`).
system	(optional) The system settings are used for recording and debugging. The data to be recorded by the system is defined here.

Detailed information about each parameter of the JSON settings is available on the Settings page.

JSON Settings: New Meta Format

The meta-format of the JSON parameters has changed since the last LumiScan VGR version 1.7.0.0.
See below for a comparison of the old and new format.

Old Format

New Format

{
     "inputFileType": "LumiScanVGRSettings",
     "inputFileVersion": "1.2.0",
     "VGRVersion": "1.7.0.4",
     ...
}

  {
        "meta": {
            "inputFileType": "LumiScanVGRSettings",
            "inputFileVersion": "1.2.0",
            "targetMachine": "Inspection1",
            "programName": "LumiScanVGR",
            "programVersion": "1.7.0.5"
        },
        ....
    }

Before installing the LumiScan VGR software, please read the information about network settings, Windows Firewall and virus protection in the
LumiScanX camera manual.

In order to install the LumiScan VGR software, you must first install CUDA 11.8.0 and cuDNN 8.8.1. Both installers are provided by HD Vision Systems.