The project IntVIS4Insp intends to design, develop and demonstrate more flexible and adaptable concepts in the field of automatic inspection using complex object manipulation through computer vision technology and artificial intelligence.
The automatic inspection systems based on computer vision are aimed at assessing the raw material/ intermediary products/ and final products quality. The generalized application of this kind of technology, as well as the growing demand in terms of quality control, entails the inspection of increasingly complex materials/objects, which in some cases implies object manipulation. Applications intended for object manipulation often use robotic systems that are led by computer vision, making the previously defined operations more flexible.
These systems lead to significant efficiency gains when compared with other conventional mechanisms. They are increasingly being applied to analyse complex objects (with specular and textured characteristics; complex tridimensional geometry and features only visible through manipulation) with tight thresholds in terms of inspection, quality and detection requirements.
Factors such as the product to be inspected, its surrounding environment, its distance to the inspection system, the lighting, the analysis perspective, among others, reflect each system’s specific methodological and technical requirements, which entail applying several image acquisition and processing paradigms.
In this regard, this project has the following main technological goals:
● Design an image acquisition platform that is more flexible and adaptable to inspection process requirements, regarding the image acquisition rate, as well as the various hardware and acquisition protocol alternatives;
● Develop a base image (2D and 3D) processing and analysis and artificial intelligence generic tools platform that can be available for automatic inspection of distinct objects;
● Develop computer vision tools dedicated to extracting tridimensional information from objects, in order to increase the robotic systems’ flexibility;
● Develop a two-hand manipulation robotic system to support complex object inspections, together with the vision system, taking an active vision approach;
● Develop an interoperability architecture, which should ensure the connection between the inspection system and the production line, and must be able to conform to the increased flexibility and adaptability of vision computer and artificial intelligence’s algorithms.