Domain-specific Design Research Articles

Mixed Integer Programming based Placement Refinement by RSMT Model with Movable Pins

Placement is a critical step in the physical design for digital application specific integrated circuits (ASICs), as it can directly affect the design qualities such as wirelength and timing. For many domain specific designs, the demands for high performance parallel computing result in repetitive hardware instances, such as the processing elements in the neural network accelerators. As these instances can dominate the area of the designs, the runtime of the complete design’s placement can be traded for optimizing and reusing one instance’s placement to achieve higher quality. Therefore, this work proposes a mixed integer programming (MIP)-based placement refinement algorithm for the repetitive instances. By efficiently modeling the rectilinear steiner tree wirelength, the placement can be precisely refined for better quality. Besides, the MIP formulations for timing-driven placement are proposed. A theoretical proof is then provided to show the correctness of the proposed wirelength model. For the instances in various popular fields, the experiments show that given the placement from the commercial placers, the proposed algorithm can perform further placement refinement to reduce 3.76%/3.64% detailed routing wirelength and 1.68%/2.42% critical path delay under wirelength/timing-driven mode, respectively, and also outperforms the state-of-the-art previous work.

Intelligent floor plan design of modular high-rise residential building based on graph-constrained generative adversarial networks

In the context of Modular High-Rise Residential Buildings (MHRBs), designing floor plans involves intricate complexities due to the need for adherence to numerous domain-specific design rules. To address this issue, our research introduces a novel framework based on a Graph-Constrained Generative Adversarial Network (GC-GAN) specialized for generating MHRB floor plans. This enhanced GC-GAN incorporates knowledge graphs that encapsulate domain-specific constraints and guidelines, thereby generating floor plans that exhibit realism, diversity, and conformity to established design principles. Additionally, the framework integrates a sophisticated image-to-vector conversion algorithm that enables seamless alignment with a predefined flat-design standardization library. A salient feature of this framework is the automated generation of Building Information Modeling (BIM) models, which rigorously conform to the modularity specifications essential for efficient modular construction. The efficacy and practical applicability of our approach have been validated through an exhaustive analysis covering fifteen cases across five diverse scenarios.

Systematic synthesis of design prompts for large language models in conceptual design

Recent advancements in large language models (LLMs) demonstrate great potential in supporting engineering design, especially conceptual design. Prompt engineering plays an important role in facilitating designer-LLM collaboration in conceptual design. This paper proposes a new classification scheme that categorizes design-specific prompts into multiple classes. It also introduces different patterns for synthesizing design prompts, being grounded in the theoretical foundations of prompt engineering and domain-specific design methodology. A design experiment, utilizing ChatGPT, was conducted to investigate the impacts of different syntheses of design prompts on the effectiveness of LLM in concept generation, as measured by the metrics of novelty and diversity.

PartLabeling: A Label Management Framework in 3D Space

In this work, we focus on the label layout problem: specifying the positions of overlaid virtual annotations in Virtual/Augmented Reality scenarios. Designing a layout of labels that does not violate domain-specific design requirements, while at the same time satisfying aesthetic and functional principles of good design, can be a daunting task even for skilled visual designers. Presenting the annotations in 3D object space instead of projection space, allows for the preservation of spatial and depth cues. This results in stable layouts in dynamic environments, since the annotations are anchored in 3D space. In this paper we make two major contributions. First, we propose a technique for managing the layout and rendering of annotations in Virtual/Augmented Reality scenarios by manipulating the annotations directly in 3D space. For this, we make use of Artificial Potential Fields and use 3D geometric constraints to adapt them in 3D space. Second, we introduce PartLabeling: an open source platform in the form of a web application that acts as a much-needed generic framework allowing to easily add labeling algorithms and 3D models. This serves as a catalyst for researchers in this field to make their algorithms and implementations publicly available, as well as ensure research reproducibility. The PartLabeling framework relies on a dataset that we generate as a subset of the original PartNet dataset [17] consisting of models suitable for the label management task. The dataset consists of 1,000 3D models with part annotations.

Systematic review of the performance evaluation of clinicians with or without the aid of machine learning clinical decision support system.

For the adoption of machine learning clinical decision support systems (ML-CDSS) it is critical to understand the performance aid of the ML-CDSS. However, it is not trivial, how the performance aid should be evaluated. To design reliable performance evaluation study, both the knowledge from the practical framework of experimental study design and the understanding of domain specific design factors are required. The aim of this review study was to form a practical framework and identify key design factors for experimental design in evaluating the performance of clinicians with or without the aid of ML-CDSS. The study was based on published ML-CDSS performance evaluation studies. We systematically searched articles published between January 2016 and December 2022. From the articles we collected a set of design factors. Only the articles comparing the performance of clinicians with or without the aid of ML-CDSS using experimental study methods were considered. The identified key design factors for the practical framework of ML-CDSS experimental study design were performance measures, user interface, ground truth data and the selection of samples and participants. In addition, we identified the importance of randomization, crossover design and training and practice rounds. Previous studies had shortcomings in the rationale and documentation of choices regarding the number of participants and the duration of the experiment. The design factors of ML-CDSS experimental study are interdependent and all factors must be considered in individual choices. The online version contains supplementary material available at 10.1007/s12553-023-00763-1.

Compiler Technologies in Deep Learning Co-Design: A Survey

With the rapid development of deep learning applications, general-purpose processors no longer suffice for deep learning workloads because of the dying of Moore’s Law. Thus, computer architecture innovation has entered a golden age for domain-specific design, which has led to a demand for new compilation technologies to facilitate cross-layer optimization. Historically, hardware and software have been collaboratively designed. Today, these co-design ideas still benefit the deep learning field in both academia and industry, encompassing additional aspects and layers. In this study, we elaborate on past and recent works on deep learning compilers and co-design while focusing on the combination of these two technologies, which we believe is the trend in the new deep learning era. After summarizing the existing compilation technologies and co-design approaches, we propose a domain-specific compilation framework, the Buddy Compiler, for a typical deep learning co-design system.

Customized FPGA Design and Analysis of Soft-Core Processor for DNN

Transistor scaling has continued to yield significant performance gains. The general-purpose architecture will not suffice in this dark silicon era with the big data boom and growing emphasis on performance. Artificial neural networks and deep learning are sophisticated computing activities that cannot be optimally executed (accuracy, performance, and energy) on today's general-purpose computer architecture. The demand for domain-specific computer architectures is increasing. This paper suggests and evaluates a highly reconfigurable domain-specific design for deep learning and artificial neural networks to overcome these problems. An FPGA-based soft-core processor for deep neural network (DNN) acceleration is suggested in this paper. Using a combination of unicast and multicast for collective operations, the proposed FPGA accelerates the performance of the deep learning network. The suggested design outperforms the current general-purpose processor by 6.6 times, according to a thorough examination and comparison. The existing architecture's level of accuracy is retained.

Decomposing style, content, and motion for videos

In this paper, we present the first video decomposition framework, named SyCoMo, that factorizes a video into style, content, and motion. Such a fine-grained decomposition enables flexible video editing, and for the first time allows for tripartite video synthesis. SyCoMo is a unified and domain-agnostic learning framework which can process videos of various object categories without domain-specific design or supervision. Different from other motion decomposition work, SyCoMo derives motion from style-free content by isolating style from content in the first place. Content is organized into subchannels, each of which corresponds to an atomic motion. This design naturally forms an information bottleneck which facilitates a clean decomposition. Experiments show that SyCoMo decomposes videos of various categories into interpretable content subchannels and meaningful motion patterns. Ablation studies also show that deriving motion from style-free content makes the keypoints or landmarks of the object more accurate. We demonstrate the photorealistic quality of the novel tripartite video synthesis in addition to three bipartite synthesis tasks named as style, content, and motion transfer.

Tailoring the V-Model for Optics: A Methodology for Optomechatronic Systems

The integration of optical technologies into once purely mechatronic systems enables innovative functions, but simultaneously increases the complexity of previous mechatronic system development. Therefore, a process has been elaborated to develop these so-called optomechatronic systems by Knöchelmann at the Institute of Product Development at Leibniz University Hanover, which is based on the V-Model of VDI 2206 and can be applied to various fields of application. For a target-oriented development in a specific product context and for systems with competing main requirements, detailing and adapting the process is recommended. High-resolution lighting systems are one of them, where requirements for high optical efficiency and image quality lead to a conflict of objectives. Focusing on the optics domain, Ley elaborated methods for the preliminary and detailed design of high-resolution lighting systems to address the aforementioned conflict of objectives. This contribution focuses on the integration of Ley’s design methods into Knöchelmann’s process model within the phases of system design and domain-specific design, allowing us to analyze the impact of the system design on the fulfillment of main requirements to achieve an optimal solution of the conflict of objectives. To illustrate this, the integrated process model is described using an example from automotive lighting technology.

Integrating Model-Based Design of Mechatronic Systems with Domain-Specific Design Approaches

AbstractIn addition to the known approaches for product development new or supplementary approaches have emerged. An important approach in this field is Systems Engineering (SE) and Model-Based Systems Engineering. Through these approaches, new procedures, level-focused description concepts and terms come into product development. However there are still some uncertainties as to how the known approaches of product development can be combined with the SE approaches. This paper aims to show how the known development approaches can be extended by and integrated with SE approaches.

Automation of the distributed applications development based on the description of the subject area

An approach to automation of the distributed information systems development is proposed, which makes it possible to radically reduce the role of traditional development, up to the complete transfer this function to specialists in the subject area of an automated enterprise. It is proposed to achieve this through the development of a specialized multi-level language for describing subject areas, built on the ideas basis of the domain driven design, event storming and other means for describing business processes. A certain phasing in the implementation of this approach is proposed. Examples of implementation of some components are given. The necessity of the intermediate representation standardization for the subject area description is shown. Keywords distributed information system, development automation, domain-specific design, domain description language

Unsupervised symbol emergence for supervised autonomy using multi-modal latent Dirichlet allocations

In future Mars exploration scenarios, astronauts orbiting the planet will control robots on the surface with supervised autonomy to construct infrastructure necessary for human habitation. Symbol-based planning enables intuitive supervised teleoperation by presenting relevant action possibilities to the astronaut. While our initial analog experiments aboard the International Space Station (ISS) proved this scenario to be very effective, the complexity of the problem puts high demands on domain models. However, the symbols used in symbolic planning are error-prone as they are often hand-crafted and lack a mapping to actual sensor information. While this may lead to biased action definitions, the lack of feedback is even more critical. To overcome these issues, this paper explores the possibility of learning the mapping between multi-modal sensor information and high-level preconditions and effects of robot actions. To achieve this, we propose to utilize a Multi-modal Latent Dirichlet Allocation (MLDA) for unsupervised symbol emergence. The learned representation is used to identify domain-specific design flaws and assist in supervised autonomy robot operation by predicting action feasibility and assessing the execution outcome. The approach is evaluated in a realistic telerobotics experiment conducted with the humanoid robot Rollin' Justin.

Mez: An Adaptive Messaging System for Latency-Sensitive Multi-Camera Machine Vision at the IoT Edge

Mez is a novel publish-subscribe messaging system for latency sensitive multi-camera machine vision applications at the IoT Edge. The unlicensed wireless communication in IoT Edge systems are characterized by large latency variations due to intermittent channel interference. To achieve user specified latency in the presence of wireless channel interference, Mez takes advantage of the ability of machine vision applications to temporarily tolerate lower quality video frames if overall application accuracy is not too adversely affected. Control knobs that involve lossy image transformation techniques that modify the frame size, and thereby the video frame transfer latency, are identified. Mez implements a network latency feedback controller that adapts to channel conditions by dynamically adjusting the video frame quality using the image transformation control knobs, so as to simultaneously satisfy latency and application accuracy requirements. Additionally, Mez uses an application domain specific design of the storage layer to provide low latency operations. Experimental evaluation on an IoT Edge testbed with a pedestrian detection machine vision application indicates that Mez is able to tolerate latency variations of up to 10x with a worst-case reduction of 4.2% of the application accuracy F1 score metric. The performance of Mez is also experimentally evaluated against state-of-the-art low latency NATS messaging system.

A virtual engineering based approach to verify structural complexity of component-based automation systems in early design phase

Highly diverse factors including technological advancements, uncertain global market and mass personalisation are believed to be main causes of ever-growing complexity of manufacturing systems. Although complex systems may be needed to achieve global manufacturing requirements, complexity affects on various factors, such as: system development effort and cost, ease of re-configuration, level of skill required across the system life-cycle (e.g. design, operate and maintain). This article aims to develop a scientifically valid and industrially applicable complexity assessment approach to support early life-cycle phases of component-based automation systems against unwanted implications of structural system complexity. The presented approach defines component-based automation system as a constellation of basic components which can be represented in various design domains, such as: mechanical, electrical, pneumatic, control, etc. Accordingly, structural complexity is expressed as the combination of both inherent complexity of system entities and topological complexity resulting from the integration of elements of such constellations in a multi-layered network. The proposed approach is used to specify and implement a complexity assessment module which can be integrated into a series of virtual system design software solutions, in order to add complexity assessment as part of the design support and validation tools used by manufacturing engineers. Consequently, the proposed approach is integrated into the vueOne virtual engineering tool, wherein virtual automation system design data can be streamlined and used as input to the theoretical complexity model. In the developed tool, only mechanical and logical design domains are considered due to the limited data availability in early design phase. Inherent complexity of both mechanical and logical system entities and their interactions are expressed as a function of domain-specific design elements, and topological complexity is defined as the graph energy of the corresponding design connectivity matrix. Furthermore, the values of mathematical model parameters are determined based on an optimisation study, where subjective opinions of system/control engineers regarding the effort/difficulty associated with the development of thirty different component-based automation system designs are correlated with the corresponding complexity model outputs to minimise the prediction errors. The proposed approach is also demonstrated on a modular production system consisting of four sub-modules. The study shows that the approach can help designers/managers to better identify root causes of structural system complexity, and provides a systemic approach to compare alternate system designs during early system planning phase.

Understanding the motivations, challenges and needs of Blockchain software developers: a survey

The blockchain technology has potential applications in various areas such as smart-contracts, Internet of Things (IoT), land registry, supply chain management, storing medical data, and identity management. Although GitHub currently hosts more than six thousand active Blockchain software (BCS) projects, few software engineering researchers have investigated these projects and their contributors. Although the number of BCS projects is growing rapidly, the motivations, challenges, and needs of BCS developers remain a puzzle. Therefore, the primary objective of this study is to understand the motivations, challenges, and needs of BCS developers and analyze the differences between BCS and non-BCS development. On this goal, we sent an online survey to 1,604 active BCS developers identified by mining the GitHub repositories of 145 popular BCS projects. The survey received 156 responses that met our criteria for analysis. The results suggest that the majority of the BCS developers are experienced in non-BCS development and are primarily motivated by the ideology of creating a decentralized financial system. Although most of the BCS projects are Open Source Software (OSS) projects by nature, more than 93% of our respondents found BCS development somewhat different from a non-BCS development as BCS projects have higher emphasis on security and reliability than most of the non-BCS projects. Other differences include: higher costs of defects, decentralized and hostile environment, technological complexity, and difficulty in upgrading the software after release. These differences were also the primary sources of challenges to them. Software development tools that are tuned for non-BCS development are inadequate for BCS and the ecosystem needs an array of new or improved tools, such as: customized IDE for BCS development tasks, debuggers for smart-contracts, testing support, easily deployable simulators, and BCS domain specific design notations.

On Semantic Interoperability of Model-based Definition of Product Design

Fundamentally, the engineering designs are collections of information, which contain the structural, qualitative and the behavioral (kinematic and static) specifications of some physical products that satisfy some customer requirements. Vast suits of information exchange standards, most notably ISO10303 (STEP), are currently being used for capturing product design in manufacturing industries. In recent years, a modern practice of capturing the product design information using model-based definition (MBD) (STEP AP242, ASME Y14.5) is becoming increasingly popular due to its ability to capture geometrical dimension and tolerances (GD&T) against managed models of 3D design features instead of traditional 2D drafting, facilitating communication of the design intents, and product model information (PMI). However, MBD suffers from difficulties in standardization of managed models due to diversity of views, stemming from cultural and traditional differences among the product designers and organizations. In the wake of semantic web technologies, researchers proposed number of semantic models for capturing product design concepts based on ontological taxonomy and rules, aiming to integrate heterogeneous sources of data based on contextual meaning of the information. In this article, we present an upper level ontology for capturing MBD based product designs, including axiomatic definitions of concepts related to product design (e.g. part, topology, feature, dimension, and tolerance). The primary goal of such semantic model is to increase the interoperability of MBD based product design and integrate domain-specific design features from manufacturability perspective (CNC, sheet metal, molding, 3D printing). In order to achieve such cross-domain interoperability, we based our design on Basic Formal Ontology (BFO). The applicability of the proposed ontology is elucidated by presenting usage pattern based on example design.

Integrating environmental sustainability in undergraduate mechanical engineering courses using guided discovery instruction

In this paper we discuss a guided discovery instruction approach for integrating environmental sustainability in undergraduate mechanical engineering courses. To validate the proposed approach, we conducted two studies with students in a computer-aided design and prototyping course. The first study verified the feasibility of incorporating guided discovery instruction for teaching environmental sustainability using a structural shape synthesis design task. The second study compared the influences of the guided discovery instruction approach and traditional lecture-based instruction on students' understanding of environmental sustainability concepts. Results show the guided discovery instruction approach facilitated a better understanding of interactions among design parameters and the resulting environmental impact. We also found that students in the guided discovery instruction group gave more prominence to modifying design parameters specific to mechanical engineering concepts taught in the course. These findings suggest that using guided discovery instruction to teach environmental sustainability in undergraduate mechanical engineering courses is beneficial for promoting students' understanding of complex relationships between domain-specific design parameters and environmental sustainability.

Model and Tool Integration Platforms for Cyber–Physical System Design

Design methods and tools evolved to support the principle of separation of concerns in order to manage engineering complexity. Accordingly, most engineering tool suites are vertically integrated but have limited support for integration across disciplinary boundaries. Cyber–physical systems (CPSs) challenge these established boundaries between disciplines, and thus, the status quo on the tools market. The question is how to create the foundations and technologies for semantically precise model and tool integration that enable reuse of existing commercial and open source tools in domain-specific design flows. In this paper, we describe the lessons learned in the design and implementation of an experimental design automation tool suite, OpenMETA, for complex CPS in the vehicle domain. The conceptual foundation for the integration approach is platform-based design: OpenMETA is architected by introducing two key platforms: the model integration platform and the tool integration platform. The model integration platform includes methods and tools for the precise representation of semantic interfaces among modeling domains. The key new components of the model integration platform are model integration languages and the mathematical framework and tool for the compositional specification of their semantics. The tool integration platform is designed for executing highly automated design-space exploration. Key components of the platform are tools for constructing design spaces and model composers for analytics workflows. The paper concludes with describing experience and lessons learned by using OpenMETA in drivetrain design and by adapting OpenMETA to substantially different CPS application domains.

Quantifying the Impact of Domain Knowledge and Problem Framing on Sequential Decisions in Engineering Design

Many decisions within engineering systems design are typically made by humans. These decisions significantly affect the design outcomes and the resources used within design processes. While decision theory is increasingly being used from a normative standpoint to develop computational methods for engineering design, there is still a significant gap in our understanding of how humans make decisions within the design process. Particularly, there is lack of knowledge about how an individual's domain knowledge and framing of the design problem affect information acquisition decisions. To address this gap, the objective of this paper is to quantify the impact of a designer's domain knowledge and problem framing on their information acquisition decisions and the corresponding design outcomes. The objective is achieved by (i) developing a descriptive model of information acquisition decisions, based on an optimal one-step look ahead sequential strategy, utilizing expected improvement maximization, and (ii) using the model in conjunction with a controlled behavioral experiment. The domain knowledge of an individual is measured in the experiment using a concept inventory, whereas the problem framing is controlled as a treatment variable in the experiment. A design optimization problem is framed in two different ways: a domain-specific track design problem and a domain-independent function optimization problem (FOP). The results indicate that when the problem is framed as a domain-specific design task, the design solutions are better and individuals have a better state of knowledge about the problem, as compared to the domain-independent task. The design solutions are found to be better when individuals have a higher knowledge of the domain and they follow the modeled strategy closely.

Domain-Specific Aspects of Technological Pedagogical Content Knowledge: Music Education and the Importance of Affect

The present study addresses the lack of a theoretical framework for the integration of technology in music teaching and learning, and explores, within the framework of Technological Pedagogical Content Knowledge (TPACK), the importance of affect in the instructional design of music lessons. The purpose of this study is twofold: (a) to extend the theoretical framework of TPACK to a design framework by enriching it with domain-specific design guidelines for the teaching of music, and, (b) to identify and propose design guidelines, which can facilitate associations of affect with cognition in learning music with technology. The research has practical significance for teachers and practitioners as it provides explicit instructional design guidance of how music lessons can be designed based on TPACK principles that address both the cognitive and the affective domain.