The Synergistic Role of Additive Manufacturing and Artificial Intelligence for the Design of New Advanced Intelligent Systems

Abstract

Intelligent systems are, by definition, setups that are able to adapt their features and functions for a specific set of applications. Systems intended to be "intelligent," or "smart," usually have a specific arrangement of active and passive structural components. The first class is the key of a setup and includes all those parts of a device made of materials that are activated upon a triggering stimulus (e.g., pH, heat) against nonreactive components. Smart systems, in a broader sense, also include those made of materials able to perform living-like functions, such as healing, sensing, and actuating. [1] Designing intelligent systems has pushed scientists to develop new strategies in terms of behavior enhancement and property predictability in relation to each specific application. For instance, any device intended for the biomedical sector has to comply with the challenging needs of dimensions and biocompatibility to prevent any mechanical and biological issue to living tissues. In contrast, applications for industrial robotics require high adaptability and reliability for manipulating objects and, in case of collaborative robotic operations, to ensure operator safety. While previous design approaches have included mathematical techniques like linear elastic topology optimization or level-set topology, [2] multiphysics finite-element models (FEMs) have widened the design space, giving scientists new tools to design, simulate, and predict the behavioral capabilities of complex structures, also by introducing material and geometric nonlinearities and solving multiple physics problems in parallel. Remarkable examples are the optimization of structures included in geometrical and material nonlinear environments, [3] studies on failure mechanisms difficult to observe experimentally, [4] or the development of compliant mechanisms for mechatronic and medical applications that exploit nonlinear deformations. [5] However, the traditional path of a device from the in silico design to manufacturing has two main limitations. From the design standpoint, FEMs represent a powerful tool that can handle parametric models by sweeping parameters across specific ranges of values. Yet, they cannot smartly cover the large number of design parameter combinations without an expensive computational cost. [6] This scenario is also complemented by the traditional fabrication technologies that narrow the topologies

Artificial Intelligence-Empowered 3D and 4D Printing Technologies toward Smarter Biomedical Materials and Approaches

“…Furthermore, as recently reported by Milazzo and Libonati [ 58 ], we give an outlook on the “5D printing” technique, in which AI and ML will assume the role of the fifth dimension to empower the effectiveness of AM in biomedical approaches in real-time. Lastly, we will briefly discuss the regulatory standpoint for managing AI technologies.…”

“…With the evolution of this technology, the concept of 4D-printing has been expanded by incorporating the product design into a flexible and intelligent material based on 3D-printing [58]. Therefore, the structures can deform, swell, self-assemble, or self-repair according to a pre-designed path under specific conditions of time and upon exposure to external stimuli.…”

Bibliometric Analysis of Fourth Industrial Revolution Applied to Material Sciences Based on Web of Science and Scopus Databases from 2017 to 2021

“…Additive manufacturing, also known as 3D printing, is used to determine fabrication parameters, quality of the workpieces, and processing time. Furthermore, it is expected that additive manufacturing will achieve 5d-printing, through the implementation of time and artificial intelligence tools as the fourth and fifth dimensions, respectively [50]. In the case of molecular dynamics, artificial intelligence is employed to contribute to understanding materials' properties by simulating the interaction of atoms and molecules.…”

Machine-Guided Discovery of Acrylate Photopolymer Compositions

“…A solution to this problem that has made major inroads in recent years is active learning (AL), which is a process to build an ML model (and dataset) iteratively from a small starting set of training points using the principles of Bayesian optimization (BO). − This creates a path to obtain informed recommendations for successive experiments, minimizing the number of experiments required (to meet a target material or performance metric) and thus expediting the development of new materials. AL has been used to accelerate material design problems in polymers, alloys, and ceramics. − Within the AM space, ML has been used to enhance in situ process monitoring to improve print quality through optimizing printing parameters, resin composition, and component attributes. ,− AL has been used to accomplish such tasks with automated decision making. ,, A small subset of recent literature deals with the modeling of acrylates for several properties, albeit in a limited capacity. Notable contributions include the modeling of glass transition temperatures for copolymers, tested on an expansive acrylate dataset, and the application of physics-constrained BO to enhance tensile strength and toughness of thermoplastic polymers. , The expanding literature highlights the potential for further investigation into optimizing multiple acrylate properties in data-scarce situations using novel data recommendation approaches and ML algorithms.…”