Collaborative R&D project · COMPETE 2030
Feeding difficult parts, without reconfiguring the machine.
VI2F is developing an automatic feeder that combines two 2D machine vision stations, artificial intelligence and two SCARA robots to separate, inspect and deliver bulk parts, including flexible, fragile or complex-shaped parts.
Reference in production: CX-07Project status
Month 5 of 36
Current phase: Preliminary studies: requirements and technical specifications
- 36months of research and development
- 2co-promoting entities
- 4industrial validation partner companies
- TRL 7technology readiness to reach by the end
The problem
Every new reference means stopping and reconfiguring.
Conventional feeders are designed for a narrow family of parts. Changing reference means swapping tools, adjusting guides and reprogramming paths, with line stoppages and specialised labour. With flexible, fragile or irregular parts, such as springs, O-rings, hooks or silicone elements, feeding is still often done by hand.
The answer
The machine adapts to the part.
Instead of mechanically adapting the equipment to each part, VI2F recognises it. Computer vision and artificial intelligence identify, inspect and locate every part; adaptive robotic handling separates the good ones, removes the rest and returns to the loop those not yet picked.
How it works
Seven steps, one reference at a time.
From bulk to the right part in the right place. The images illustrate the concept, not the final equipment.
01Bulk feeding
Parts enter in bulk and are spread on the belt in a controlled, spaced way to reduce overlapping. The feeder works with one reference at a time.
02First inspection: 2D vision with AI
Checks in real time the conformity of every part and its separation on the belt. Parts of another reference that appear on the belt are always treated as rejects.
03Rejection and redistribution
A SCARA robot removes non-conforming parts and redistributes valid parts that overlap or are badly placed, leaving the belt ready for picking.
04Second inspection: position for picking
A second 2D camera, further along, confirms each eligible part and determines its position and orientation, which it sends to the picking robot.
05Picking on the moving belt
The picking robot collects the conforming parts while the belt keeps moving, with vision and robot synchronised, and hands them to the next process.
06Automatic tool change
Each robot has a gripper store next to it, including soft and flexible grippers, to change tool without operator intervention.
07Recirculation
Valid parts that were not picked fall onto a return conveyor and an elevator takes them back to the feeder, with no waste.
Objectives
Four research fronts.
- 01
Flexible multi-product feeder
Handle rigid, semi-rigid, hollow and irregular parts, with no physical reconfiguration between references.
- 02
Robotic perception with 2D vision and AI
Real-time quality control and part location, with models trained per product family rather than per reference.
- 03
Active redistribution and rejection
SCARA robots with soft, flexible grippers to separate overlapping parts and segregate non-conforming ones.
- 04
Modular interface for Industry 4.0
Standard industrial protocols (OPC UA, Profinet, EtherCAT), secure remote access, monitoring and integration with MES and ERP systems.
Consortium
Industry and science, side by side.
Lead partner
Neosepec
Industrial SME based in São João da Madeira, specialised in special machines, production lines and mechatronic solutions. In VI2F, it designs and builds the equipment and leads industrial integration and real-world validation.
Co-promoter
INESC TEC
Institution of the Portuguese science and technology system, based in Porto, with recognised expertise in robotics, machine vision and computational intelligence. In VI2F, it develops the artificial intelligence algorithms, the computer vision models and the digital control architecture.
Industrial validation partners
- Sporting armsPartner company
- High-precision plastic injectionPartner company
- Automotive componentsPartner company
- Die casting and machiningPartner company
Funding
Co-funded by the European Union.
Innovation and Digital Transition Programme · COMPETE 2030 · ERDF
- 1 650 853,02 €total eligible cost
- 1 283 972,30 €European Union support
- 78 %co-financing rate