How to set up a Pickit 3D robot vision application in minutes

Configuring a 3D vision system with Pickit is faster and easier than you might expect. The platform is designed to eliminate coding and programming, allowing you to set up a bin picking application in just a few minutes using only your mouse.

This quick guide walks through the entire process — from defining the region of interest to enabling your robot to execute reliable picks.

Understanding the Pickit interface

The Pickit interface is designed for clarity and simplicity. On the top left, you have the viewer, showing both a 2D image and the 3D point cloud captured by the camera. The lower left displays detection results, while the right panel provides configuration panel.

Configuration is organized into three tabs:
- Setup – where to look
- Detection – what to look for
- Picking – how to pick the parts

This structure guides you step by step from environment definition to picking execution.

Defining the region of interest

Setup begins with defining a region of interest (ROI) - the area where Pickit will perform detection. A blue box represents this ROI. Align it with your bin to ensure the system focuses only where needed.

To position the box accurately, QR markers can be placed on the bin’s corners. Pickit automatically detects these markers and aligns the box accordingly. You can then fine-tune its position by adjusting edges to cut the corners.

Additionally, Pickit allows you to model the collision environment. By generating collision bins around the ROI, you ensure that the robot avoids unintended contact during picking.

Setting up object detection

Once the ROI is defined, the next step is telling Pickit what objects to detect. This is done by either uploading a CAD model of the object or showing it to the camera. Pickit supports most 3D file formats, making it easy to import your parts.

For example, if you’re detecting a shaft, upload its CAD file. The system then overlays this model onto the 3D point cloud and automatically identifies matching objects inside the bin.

At this stage, the objects are detected but not yet pickable - they will appear highlighted (in orange) but labeled as unpickable until picking is configured.

Configuring Picking

With detection in place, the next step is to define how the robot should pick the parts. This involves:

Tool model: Define the geometry of your picking tool. For example, a simple cylinder can represent a suction gripper. You can also upload a CAD model of your tool for more precise collision checking.

Flexibility: Specify the angles your tool can tolerate while still successfully gripping parts. This is crucial for maximizing picking opportunities while minimizing unnecessary robot motion. Green and red indicators in the interface visualize these tilt allowances.

By allowing flexibility, you increase the number of valid pick poses the system can generate, making your application more robust.

Pick points: Tell Pickit where exactly to grip the object. To add one, select a location on the object model and align the tool approach correctly. You can adjust the orientation so the tool aligns with the part surface.

You can also define a symmetry axis, giving the system freedom to pick from any angle around that axis. More flexibility means better optimization and higher success rates.

Validating Picks

Once pick points are defined, Pickit reevaluates the detected objects. Those previously marked as “unpickable” now become ready to go to the robot, turning from orange to color-coded pickable objects.

If a part is blocked or would cause a collision, it remains unpickable until the obstructing part is removed. This layered approach ensures reliable and safe picking even in cluttered bins.

Ready. Set. Go!

With setup, detection, and picking complete, your system is ready to operate. Within minutes, you have a fully functional 3D vision-guided bin picking application.

In summary; setting up a Pickit application follows a clear and easy three-step flow:

1. Setup – Define the region of interest and collision environment.
2. Detection – Upload a CAD model or teach it to the camera and detect your objects.
3. Picking – Configure your tool, define pick points, and validate pickable objects.

Pickit’s intuitive interface ensures that anyone - from automation engineers to operators - can configure robust 3D vision applications quickly and confidently.

Want us to guide you through the process with a real demo?… Then watch this short video👇