Abstarct Engineering design problems typically require optimizing a quality measure by finding the right combination of controllable input parameters. In Additive Manufacturing (AM), the output characteristics of the process can often be non-stationary functions of the process parameters. Bayesian Optimization (BO) is a methodology to optimize such “black-box” functions, i.e., the input–output relationship is unknown and expensive to compute. Optimization tasks involving “black-box” functions widely use BO with Gaussian Process (GP) regression surrogate model. Using GPs with standard kernels is insufficient for modeling non-stationary functions, while GPs with non-stationary kernels are typically over-parameterized. On the other hand, a Deep Gaussian Process (DGP) can overcome GPs’ shortcomings by considering a composition of multiple GPs. Inference in a DGP is challenging due to its structure resulting in a non-Gaussian posterior, and using DGP as a surrogate model for BO is not straightforward. Stochastic Imputation (SI)-based inference is promising in speed and accuracy for BO. This work proposes a bootstrap aggregation-based procedure to effectively utilize the SI-based inference for BO with a DGP surrogate model. The proposed BO algorithm DGP-SI-BO is faster and empirically better than the state-of-the-art BO method in optimizing non-stationary functions. Several analytical test functions and a case study in metal AM simulation demonstrate the applicability of the proposed method.

Human-Robot Collaboration (HRC) is key to achieving the flexible automation required by the mass personalization trend, especially towards human-centric intelligent manufacturing. Nevertheless, existing HRC systems suffer from poor task understanding and poor ergonomic satisfaction, which impede empathetic teamwork skills in task execution. To overcome the bottleneck, a Mixed Reality (MR) and visual reasoning-based method is proposed in this research, providing mutual-cognitive task assignment for human and robotic agents’ operations. Firstly, an MR-enabled mutual-cognitive HRC architecture is proposed, with the characteristic of monitoring Digital Twins states, reasoning co-working strategies, and providing cognitive services. Secondly, a visual reasoning approach is introduced, which learns scene interpretation from the visual perception of each agent’s actions and environmental changes to make task planning strategies satisfying human–robot operation needs. Lastly, a safe, ergonomic, and proactive robot motion planning algorithm is proposed to let a robot execute generated co-working strategies, while a human operator is supported with intuitive task operation guidance in the MR environment, achieving empathetic collaboration. Through a demonstration of a disassembly task of aging Electric Vehicle Batteries, the experimental result facilitates cognitive intelligence in Proactive HRC for flexible automation.

This article examines a real-time parking reservation service that dynamically allocates a limited number of parking slots (which can be reused) to serve parking requests in a crowded area. The study takes into account two stochastic factors: (i) random arrivals of parking requests and (ii) driver parking unpunctuality, which are seldom simultaneously considered in the existing literature. This study formulates the dynamic parking resource allocation problem as a joint chance-constrained model to maximize the expected total revenue over a finite horizon. Due to the complex nature of the model considering the two stochastic factors, the study proposes a data-driven approach using joint chance constraint decomposition and sample average approximation to approximate the model to a deterministic mixed-integer programming model. Furthermore, a stratified ranked set sampling method is introduced to construct a high-quality sample set as the input to the deterministic model, and valid inequalities are designed to accelerate the solution process. Numerical experiments based on real-world data are conducted to validate the performance of the proposed approach under multiple parking scenarios and provide findings.

Landmarks are essential in non-rigid shape registration for identifying the correspondence between designs and actual products. In 3D printing, manual selection of landmarks becomes labor-intensive, due to complex product geometries and their non-uniform shape deviations. Automatic selection, however, has to pinpoint landmarks indicative of geometric regions prone to deviations for accuracy qualification. Existing automatic landmarking methods often generate clustered and redundant landmarks for prominent features with high curvatures, compromising the balance between global and local registration errors. To address these issues, we propose an automatic landmark selection method through deviation-aware segmentation and landmarking. As opposed to segmentation for semantic feature identification, deviation-aware segmentation partitions a freeform product for high-curvature region identification. Prone to deviation, these regions are generated through curvature-sensitive remeshing to extract vertices of high curvature and automatic clustering of vertices based on vertex density. Within each segment or high-curvature region, a curvature-weighted function is tailored for the Gaussian process landmarking to sequentially select landmarks with the highest local curvatures. Furthermore, we propose a new evaluation criterion to assess the effectiveness of selected landmarks through registration. The proposed approach is tested through automatic landmarking of printed dental models.

Bayesian Optimization (BO) links Gaussian Process (GP) surrogates with sequential design toward optimizing expensive-to-evaluate black-box functions. Example design heuristics, or so-called acquisition functions, like expected improvement, balance exploration and exploitation to furnish global solutions under stringent evaluation budgets. However, they fall short when solving for robust optima, meaning a preference for solutions in a wider domain of attraction. Robust solutions are useful when inputs are imprecisely specified, or where a series of solutions is desired. A common mathematical programming technique in such settings involves an adversarial objective, biasing a local solver away from “sharp” troughs. Here we propose a surrogate modeling and active learning technique called robust expected improvement that ports adversarial methodology into the BO/GP framework. After describing the methods, we illustrate and draw comparisons to several competitors on benchmark synthetic exercises and real problems of varying complexity.

Recent years have witnessed the emergence of many new mobile apps and user-centered systems that interact with users by offering choices with rewards. These applications have been promising to address challenging societal problems such as congestion in transportation and behavior changes for healthier lifestyles. Considerable research efforts have been devoted to model the user behaviors in these new applications. However, as real-world user data is often prone to data corruptions, the success of these models hinges on a robust learning method. Building on the recently proposed Latent Decision Threshold model, this article shows that, among the existing robust learning frameworks, the L 0-norm-based framework can outperform other state-of-the-art methods in terms of prediction accuracy and model estimation. And based on the L 0-norm framework, we further develop a user screening algorithm to identify potential bad actors.

Motivated by make-to-order applications with committed delivery dates in a variety of industries, we investigate the integrated multi-plant collaborative production, inventory, and hub–spoke delivery problem in a complex production–distribution network. This network includes multi-location heterogeneous plants, distribution centers, and customers, for producing customized and splittable orders with one or more general-size multi-type jobs. Completed jobs are transported from plants to distribution centers, and then the orders whose all constituent jobs have arrived are delivered from distribution centers to customer sites. The objective is to make integrated scheduling decisions for production, inventory, and delivery, for minimizing total cost composed of production, transportation, tardiness, and inventory. We first formulate this problem as a mixed-integer programming model, and analyze its intractability by proving that the problem is NP-hard and no approximation algorithms exist with a constant worst-case ratio. We then reformulate this problem as a binary integer linear programming model to select a feasible schedule for each job, and propose a combined column generation and two-layer column enumeration algorithm to solve it. Through extensive numerical experiments, we demonstrate that our proposed algorithm is capable of generating optimal or near-optimal solutions expeditiously and outperforms four benchmark approaches, and gain valuable managerial insights for practitioners.

We introduce the Influence Coverage Optimization Problem (ICOP), which is an influence maximization problem where the activation of nodes also depends on their location on the plane. Specifically, the ICOP assumes that there is a network where nodes become active (i.e., influenced) either by the influence they receive from interactions with active in-neighbors or by entering the coverage area of a physical ad or a Geo-fence. The objective is to locate a fixed number of ads or Geo-fences and modify the network influence rates to minimize the network activation time. Assuming a Markovian influence model, we prove that the ICOP is NP -hard, and then we present mixed-integer programming formulations for three different types of coverage modes. A reformulation of the non-linear “big-M” constraints, two types of valid cuts, and a fast heuristic based on the k-means algorithm are used as enhancements that facilitate solving the ICOP via an Iterative Decomposition Branch-and-Cut (IDBC) algorithm. In addition, we present an alternative discrete formulation of the ICOP using critical intersection points. Several experiments under various parameter configurations across instances with more than a hundred nodes and thousand arcs are conducted, showing the IDBC’s capability to provide optimal solutions within seconds or minutes for most instances. Moreover, the experiments reveal that the ICOP can significantly outperform a Geo-fence coverage model that does not consider network interactions to make location decisions.

We consider a continuous-review inventory control problem for a hybrid manufacturing and remanufacturing system with product recovery and setup costs. A pull policy has been proposed in the literature, however, finding its optimal parameters requires an exhaustive search and a single cost function evaluation is itself complex. We propose a tractable approach to finding these parameters using an interim policy called double (r, Q) with parameters ( r m , Q m , r r , Q r ) . When the inventory position of the serviceable item reaches rr , a remanufacturing lot size Qr is setup if recoverable inventory is sufficient. Otherwise, we allow the inventory position to decrease further. As it drops to rm , a manufacturing lot size Qm is placed. Unlike the pull policy, this interim policy suspends the remanufacturing option when the inventory position is less than rr . This facilitates an efficient approximation of the recoverable inventory which decouples the double (r, Q) problem into two standard (r, Q) problems. It can then be efficiently solved using a modification of existing (r, Q) algorithms for two instances. Numerical studies show that our approach performs well relative to the optimal pull policy with parameters estimated from an extensive Simulation-Optimization method and other heuristics found in the literature. The approach is also extended to correlated demand and return arrivals.