Project is the core working unit of PickWiz. It includes Robot configuration and communication, eye-hand calibration, Vision Parameter settings, and multiple other feature and task configurations, providing an efficient and flexible solution for the Robot to complete picking tasks and meeting the needs of various industrial scenarios such as sack and carton depalletizing, random picking, ordered loading and unloading, and positioning assembly.

The workflow for building a Project is as follows:

1. Create a New Project

When PickWiz is opened for the first time, you will enter the welcome page. If you need to create a new empty Project, refer to 1.1 Create a New Empty Project;

For common typical application scenarios, you can load the Project for the required Scene from the Case Library. It contains the basic configuration data of the Scene, allowing faster commissioning and configuration. For details, refer to 1.2 Create a New Project from the Case Library.

1.1 Create a New Empty Project

Select a task Scene and create a new empty Project for that Scene.

① Select a task Scene

Click a Scene image to display the introduction to that task Scene

If the materials are sacks and cartons, refer to Depalletizing Scene

If the materials are stacked Target Objects arranged randomly and irregularly, refer to Random Picking Scene

If the materials are stacked Target Objects arranged in an ordered and regular manner, refer to Ordered Loading and Unloading Scene

If there is only one Target Object within the Camera field of view that needs positioning, refer to Positioning Assembly

Scene 1: Depalletizing Scene

Depalletizing refers to the process of separating stacked materials one by one and moving them to a specific position.

• If the material is sacks, select Single Sack Depalletizing

• If the material is cartons, select Single Carton Depalletizing

Scene 2: Random Picking Scene

A random picking Scene means that the Robot recognizes and picks randomly and irregularly arranged Target Objects to complete loading and unloading tasks.

• If the Target Object is circular-surface type and arranged randomly and irregularly, select Random Picking Based on Circular Surfaces

• If the Target Object is a cylinder and arranged randomly and irregularly, select Random Picking Based on Cylinders

• If the Target Object is neither a circular surface nor a cylinder, has a large difference between the front and back sides, and is arranged randomly and irregularly, select General Target Object Random Picking. This Scene is suitable for most random picking scenarios, but it is not suitable for Target Objects with only a small difference between the front and back sides.

• If the Target Object is neither a circular surface nor a cylinder, has a small difference between the front and back sides, and is arranged randomly and irregularly, select Random Picking of Surface-Type Target Objects

• If the Target Object is neither a circular surface nor a cylinder, has a small difference between the front and back sides, is isolated from other Target Objects, and is arranged randomly and irregularly, select Loading and Unloading of Surface-Type Target Objects (Materials Isolated from Each Other)

Scene 3: Ordered Loading and Unloading Scene

An ordered loading and unloading Scene means that the Robot recognizes and picks regularly and orderly arranged Target Objects to complete material loading and unloading tasks.

• If the Target Object is circular-surface type and arranged regularly and orderly, select Ordered Loading and Unloading Based on Circular Surfaces

• If the Target Object is a cylinder and arranged regularly and orderly, select Ordered Loading and Unloading Based on Cylinders

• If the Target Object is a quadrilateral and arranged regularly and orderly, select Ordered Loading and Unloading Based on Quadrilaterals

• If the Target Object is not a circular surface, cylinder, or quadrilateral, has a large difference between the front and back sides, and is arranged regularly and orderly, select General Target Object Ordered Loading and Unloading. This Scene is suitable for most ordered loading and unloading scenarios, but it is not suitable for Target Objects with only a small difference between the front and back sides.

• If the Target Object is not a circular surface, cylinder, or quadrilateral, has a small difference between the front and back sides, and is arranged regularly and orderly, select Ordered Loading and Unloading of Surface-Type Target Objects

• If the Target Object is not a circular surface, cylinder, or quadrilateral, has a small difference between the front and back sides, is arranged regularly and orderly, and you also need to significantly reduce the time consumed by vision calculation, select Ordered Loading and Unloading of Surface-Type Target Objects (Parallelized)

• If the Target Object is not a circular surface, cylinder, or quadrilateral, has a small difference between the front and back sides, is arranged regularly and orderly, and rapid deployment is also required, select Ordered Loading and Unloading of Surface-Type Target Objects (Image Matching)

Scene 4: Positioning Assembly

Positioning assembly means that there is only one Target Object within the Camera field of view that needs positioning. For example, a charging head is precisely positioned as a single target so that the Robot can accurately plug and unplug the charging head.

• If there is only one Target Object within the Camera field of view that needs positioning, the difference between the front and back sides is small, and the Target Object position changes only slightly (ROI 3D can completely enclose the Target Object. For ROI 3D, refer to ROI Operation Guide), select Positioning Assembly of Surface-Type Target Objects (Point Cloud Matching Only)

• If there is only one Target Object within the Camera field of view that needs positioning, the difference between the front and back sides is small, and the Target Object position does not change much (ROI 3D can completely enclose the Target Object. For ROI 3D, refer to ROI Operation Guide), select Positioning Assembly of Surface-Type Target Objects (Image Matching Only)

• If there is only one Target Object within the Camera field of view that needs positioning, the difference between the front and back sides is small, and the Target Object position changes greatly (ROI 3D cannot completely enclose the Target Object. For ROI 3D, refer to ROI Operation Guide), select Positioning Assembly of Surface-Type Target Objects

• If there is only one Target Object within the Camera field of view that needs positioning and the difference between the front and back sides is large, select General Target Object Positioning Assembly. It is not suitable for Target Objects with only a small difference between the front and back sides.

② After selecting a task Scene, fill in the Project Name. The default is "new_project" + the Project sequence number, and it can be modified.

The Project name cannot contain Chinese characters or spaces

③ Select the Project Path. The default Project storage path is C:\Users\dex\Documents\PickWiz, and it can be modified.

④ Select feature options

Each Scene has multiple task types, and each task type corresponds to multiple feature options, which can be selected according to the actual working conditions.

Feature Option	Description
Instance Optimization	Optimizes the instances generated by the model and processes the instance Mask.
Pallet Detection	Used to detect pallets. After it is selected, pallet detection-related Parameters are added to the Vision Parameter page. For the detailed process, refer to [Pallet Detection Guide](../功能选项/底托检测使用指南.md)
Separator and Pallet Detection	Applicable to scenarios where there are separators between each layer of material and pallets are present. Through Vision Classification, three data types can be entered separately: separators, pallets, and Target Objects. For the detailed process, refer to [Separator and Pallet Detection Guide](../功能选项/隔板和底托检测使用指南.md)
Collision Detection	Used to detect collisions between the Tool and the container and filter Picking Poses that may collide. After it is selected, container-related Parameters must be configured in "Scene Object". At the same time, collision detection-related Parameters are added to the Vision Parameter page. For the detailed process, refer to [Collision Detection Guide](../功能选项/碰撞检测使用指南.md).
Recognition Type	Recognizes multiple types of Target Objects through one Deep Learning model. After it is selected, multiple types of Target Objects need to be configured on the Target Object page according to requirements, and then the built-in one-click end-to-end training model is used. For the detailed process, refer to [Recognition Type Guide](../功能选项/识别类型使用指南.md)
Local Feature Recognition	Positions the Target Object through local features of the Target Object. After it is selected, the overall information and local information of the Target Object need to be configured on the Target Object page according to requirements, and then the built-in one-click end-to-end training model is used. For the detailed process, refer to [Local Feature Recognition Guide](../功能选项/识别局部特征使用指南.md)
Front/Back Recognition (via Point Cloud Template)	You can import Point Cloud Templates for the front and back sides of the Target Object to match the front or back side of the picked Target Object. Pick Point settings can also be configured separately for the front and back sides of the Target Object. For the detailed process, refer to [Front/Back Recognition (via Point Cloud Template) Guide](../功能选项/识别正反（通过点云模版）使用指南.md)
Vision Classification	Used to recognize different textures, orientations, and other features of the same Target Object. After it is enabled, feature types of different categories of the same object can be entered on the Target Object page, and related Parameters can be set on the Vision Parameter page. For the detailed process, refer to [Vision Classification Guide](../功能选项/视觉分类使用指南.md)

1.2 Create a New Project from the Case Library

• On the welcome page, switch to Create from Case Library and select the Project for the required Scene

• If this is the first time you are using the case, click the Download button to download the Project configuration

• After downloading the Project configuration of the case, fill in the Project Name and Project Path.

2. Load a Project

If you need to open an existing Project on the disk, click Load Project on the welcome page

Open the corresponding Project folder, select the Project, and click Select Folder. The default Project storage path is C:\Users\dex\Documents\PickWiz.

If the following incompatibility issue appears when opening the Project, use a lower version of PickWiz to open the Project. If necessary, contact Dexforce technical support personnel.

---

3. Project Configuration

Open the Project configuration folder config to view the Project configuration data, as shown below.

Note

It is not recommended to modify the Project configuration folder. Otherwise, PickWiz may report errors and fail to run properly.

No.	Folder Name	Configuration Name	Description
1	calibration	eye-hand calibration (new version)	Stores the eye-hand calibration (new version) configuration used when the current task is running
2	camera	Camera	Stores the Camera and Camera Parameters used when the current task is running
3	end_effector	Tool	Stores the Tool configuration used when the current task is running
5	PickLightMidData	Visualization configuration	Stores the configuration corresponding to the visualization window
6	product	Target Object	Stores the Target Object configuration of the current task
7	robot	Robot	Stores the Robot used when the current task is running
8	roi	ROI	Stores the ROI configuration used when the current task is running
9	roi2d	ROI 2D	Stores the ROI 2D configuration used when the current task is running
10	roi3d	ROI 3D	Stores the ROI 3D configuration used when the current task is running
11	scene_object	Scene Object	Stores various objects in the actual Scene of the current task, including material bins, pallets, and more
12	strategy	Vision Parameter	Stores the Vision Parameter configuration used when the current task is running
13	dexforce_proj.json	Root configuration file	The root configuration file of the Project. All configuration items corresponding to all tasks in the Project are specified in this file

After a new Project is created, Pickwiz creates a default task. The Project configuration data can be cross-checked against the task information

Label	Description
Task Name	The name of the current task and can be modified
Task Type	The type of the current task, that is, the task Scene. It corresponds to the task Scene selected when creating a new task and cannot be modified
Task ID	Identifies the current task. The Robot can switch tasks through this ID
Camera	The Camera and Camera Parameters used when the current task is running. For connecting the Camera and adjusting camera imaging quality, refer to Camera Connection and Parameter Adjustment Guide
Robot	The Robot used when the current task is running. For establishing communication with the Robot, refer to Robot Configuration and Communication Operation Guide
Target Object	The Target Object configuration of the current task. After adding it on the Target Object page, configure it into the task information. For adding a Target Object, refer to Target Object Operation Guide
Tool	The Tool configuration used when the current task is running. After adding it on the Tool page, configure it into the task information. For adding a Tool, refer to Tool User Guide
Scene Object	Various objects in the current task Scene, including material bins, pallets, and more. The Scene Object page appears only after Collision Detection, Pallet Detection, and Separator and Pallet Detection are selected in Feature Options. After a Scene Object is added, it needs to be configured into the task information. For the detailed process, refer to Pallet Detection Guide, Collision Detection Guide, and Separator and Pallet Detection Guide
eye-hand calibration	The eye-hand calibration configuration used when the current task is running. After adding it on the eye-hand calibration page, configure it into the task information. For the detailed process, refer to Eye-Hand Calibration Operation Guide
ROI	The ROI configuration used when the current task is running. After adding it on the ROI page, configure it into the task information. For the detailed process, refer to ROI Operation Guide
Vision Parameter	The Vision Parameter configuration used when the current task is running. Configure it on the Vision Parameter page. For the detailed process, refer to Vision Parameter Adjustment Guide
Feature Options	Click "Edit" to select working-condition attributes such as "Collision Detection", "Recognition Type", "Local Feature Recognition", and "Pallet Detection". After selecting working-condition attributes related to the Project, PickWiz matches feature configurations suitable for those working conditions. That is, it adds configurations related to the selected working conditions into related configurations such as Target Objects and Vision Parameters. For related features, refer to the Feature Options User Guide

Previous page3D Vision Guidance Kit Hardware Setup

Next pageTask Operation Guide