The Cambrian vision system has been developed for cost-effective, fast and robust bin picking, pick-and-place, machine feeding, feature recognition, assembly, supermarket and warehouse logistics, as well as other tasks in the fields of automation and manufacturing.

The system is based on artificial intelligence. To make precise predictions, the AI model uses only the image data from the stereo cameras, which can be mounted on the robot or above the workstation. The image processing system requires no external lighting such as structured light or infrared light. The system is highly robust against changes in external lighting conditions, even when it becomes very bright or very dark. The AI model calculates the exact positions of the parts and suitable gripping points within 0.2 seconds. The system can handle a wide variety of different parts, including typical plastic and metal parts; it also works even if these are transparent, thin or highly reflective. The standard camera module can handle parts ranging from the sub-millimetre range up to those that fit within a 20-cm cube. For larger parts, we offer a slightly larger camera module with a different camera lens. Set-up is quick and easy. The user simply needs to define the gripping points, and the AI trains itself without any further user input. Mounting plates for most grippers are available and included in the scope of delivery.

The Cambrian system utilises the power of state-of-the-art neural networks to recognise a wide variety of different parts, including shiny, reflective, transparent and black parts that are difficult or impossible for conventional vision systems to detect.

The AI model responsible for part recognition is trained using simulated data and exposed to a wide range of conditions, enabling it to learn to recognise even the most challenging parts.

Brief description of function:

• Load the CAD model you want to manipulate into the Cambrian software.
• Define your ideal grip points on the CAD model (approx. 10 minutes or less) -> send them to Cambrian.
• Get an AI model on your local system that can reliably and accurately locate the part.
• Insert the Cambrian nodes for localisation and approximation in the UR programme and start the extraction.
• The system does not need to be trained using the actual robot or the actual part.

Features/Capabilities:

• Reliable bin picking of a wide variety of parts
• Precise detection and  gripping of transparent, black or shiny parts
• Cable management and wiring harnesses
• State-of-the-art AI recognition
• Fast prediction, less than 0.17 seconds
• Accuracy in the sub-millimetre range
• Quick setup
• Handles collisions with the pit
• Does not require external lighting
• Reliable in a variety of lighting conditions
• Versions available for mounting on a robot arm or on the top

Licence: One-off purchase

Requirements: A robot and gripper hardware are required for order picking. The Cambrian Vision System is compatible with all major robot brands (e.g. ABB, KUKA, FANUC, DENSO, UR, DOOSAN, Stäubli, Dobot, Hyundai, Yaskawa)

Certifications & Standards:

Cameras and computing units are CE-certified.

Contents: 

• IPC with Cambrian software
• Camera module
• Network switch
• Cables (network & power)
• Calibration board
• Test parts
• Mounting plate (on request)
• Fixing screws
  
  

Dimensions:

Dimensions: LxWxH 500x400x400 mm

Camera module: LxWxH 81x100x32 mm

Weight of the camera module: 220 g

Dimensions of the computing module: LxWxH 283x268x139 mm

Weight of the computing module: 4.2 kg