ViTroVo: in vitro assembly search for in vivo adaptive operator guidance

Abstract

Product customisation is a topic of growing interest in Smart Manufacturing. Allowing customers to design intended products brings additional challenges to the manufacturing task, such as the increase in flexibility of the assembly theatre, the compilation of assembly instructions for possibly unique products, and stress-related risks for human operators. This work introduces ViTroVo, an artificial intelligence framework capable of (1) autonomously building a graph of assembly steps via trial-and-error (in vitro Assembly Search) and (2) presenting relevant instructions to a human operator and, by autonomously detecting her progress and affective state, adapting accordingly (in vivo Adaptive Operator Guidance). The power of ViTroVo resides in its versatile way to manipulate a given product’s component Augmented Computer Aided Design (CAD+) models throughout the whole assembly task. We conducted an empirical evaluation involving participants instructed to assemble a previously unseen product. The encouraging results make us believe ViTroVo’s architecture could become the foundations of highly customised flexible manufacturing.

Enhancing manufacturing operations with synthetic data: a systematic framework for data generation, accuracy, and utility

“…Synthetic data has found diverse applications such as multispectral data classification in plastic bottle sorting (Maliks and Kadikis, 2021), and autonomous navigation in unstructured industrial environments using Unreal Engine 4 (Outón et al, 2021). Virtual environments like ViTroVo have been employed for in vitro assembly search (Grappiolo et al, 2021) and virtual prototyping has been utilized for detecting modules in modular integrated construction (Zheng et al, 2020).…”