Case Studies
Robotic picking of croissants
POSTED 03/18/2022
Partner:
IMR – Irish Manufacturing Research
Problem:
Bin picking of randomly oriented parts is of great interest to manufacturers as automation of these processes can save a lot of costs and space. The market offers a large number of bin picking solutions but many of them struggle with recognizing and identifying transparent, deformable, variable, and reflective objects that may even be piled, entangled, and jumbled up.
The Irish Manufacturing Research built a bin picking cell for the picking of 3 different types of randomly oriented bread products.
This project presented several technical challenges:
- Identification and picking of objects of non-symmetrical and complex shapes
- Random orientation of the objects
- Picking of 3 different types of objects
- Achieving optimal process speed
Solution:
Bin Picking Studio+PhoXi 3D Scanner
The solution comprises a bin picking cell with Photoneo robotic intelligence software for bin picking – the Bin Picking Studio, a PhoXi 3D Scanner, and a UR 5 robot. The supplier of Photoneo technology in the UK is Multipix Imaging.
Before implementation, the following setup and configuration requirements needed to be met:
Pre-planning:
- Defining a suitable end-of-arm tooling design
- Mounting an EOAT gripper and testing its operation on the robot
- Ensuring correct safety requirements (Collaborative mode or Industrial mode)
- Having CAD models (in the STL file format) of:
- The product to be picked
- The EOAT gripper
- The cell layout
- Having the required software files of the UR5 robot for integration with the Bin Picking Studio
- Having the required network cables for a physical connection between Bin Picking Studio, UR5, and PhoXi 3D Scanner
- Having an HDMI cable for connection between Bin Picking Studio hardware and the monitor screen
Setup:
After all the above requirements had been met, the next implementation steps were:
- Building the cell and integrating hardware
- Integrating Bin Picking studio hardware and software
- Planning gripper grasping, including all potential grasping orientations and picking points
- Calibration of robot vision and the end-of-arm tooling
- Setting parameters
- Optimization of layout, path planning, and robot parameters
- Repeatability testing
The system enables repeatable picking and emptying of random bins filled with 3 different bread products of complex shapes at a picking rate of 6ppm. If the products had a symmetrical shape, the processing speed would be even faster.
The system enables an easy changeover between different products.
Discussion:
This project provides IMR partners and customers insights into the requirements and capabilities of a typical bin picking solution. IMR has invested in a commercial bin picking solution that Irish manufacturers and IMR partners can avail of.
The system can be further optimized by:
- Setting the robot speed to maximum
- Restricting the robot’s joint motions for shorter trajectories
- Using smart memory and fast scanning option
- Optimizing localization parameters available in the system
- Using grasp invariance feature that allows multiple possible grasp points on the product