Improving robot vision with AI
‘We knew that Flanders Make has strong expertise in AI and computer vision: exactly what we needed. With the available data from customers already using our systems, we put the Flanders Make team to work. We quickly saw promising results.’
Challenge
Making robots see and understand better
Industrial robots are ideal for repetitive tasks such as bin picking, but they struggle in unpredictable environments. This can lead to errors, quality issues and wasted time. Pickit wanted robots to not only “see” with cameras, but also “understand” what they see.
Technologies
Zero-shot AI segmentation
Together with Flanders Make, Pickit developed an AI solution that does not require specific training for each production environment. The software can draw contours around relevant objects, even in unfamiliar situations, making robots more flexible for use in a wide range of sectors.
Situational and self-awareness technologies
To identify, interrogate, and evaluate both environmental conditions and system states.
Result
From proof-of-concept to pilots
Promising results were achieved with customer data and practical experience. The solution has now been validated by several customers, and pilot projects with companies in the mechanical engineering and automotive sectors are planned.
