Modular construction aims at overcoming challenges faced by the traditional construction process such as the shortage of skilled workers, fast-track project requirements, and cost associated with on-site productivity losses and recurrent rework. Since manufacturing is done off-site in controlled factory settings, modular construction is associated with increased productivity and better quality control. However, because every construction project is unique and results in distinct work pieces and building elements to be assembled, modular construction factories necessitate better mechanisms to assist workers during the assembly process in order to minimize errors in selecting the pieces to be assembled and idle times while figuring out the next step in an assembly sequence. Machine intelligence provides opportunities for such assistance; however, a challenge is to rapidly generate large datasets with rich contextual data to train such intelligent agents. This work overviews a mechanism to generate such datasets in virtual environments and evaluates the performance of AI models trained using data generated in virtual environments in recognizing the next installation step in modular assembly sequences. Performance of the trained MV-CNN models (with accuracy of 0.97) shows that virtual environments can potentially be used to generate the required datasets for AI without the costly, time-consuming, and labor-intensive investments needed upfront for capturing real-world data.
more »
« less
Towards AI-Assisted Smart Training Platform for Future Manufacturing Workforce
With the fast development of Industry 4.0, the ways in which manufacturing workers handle machines, materials, and products also change drastically. Such changes post several demanding challenges to the training of future workforce. First, personalized manufacturing will lead to small batch and fast
changing tasks. The training procedure must demonstrate agility. Second, new interfaces to interact with human or robots will change the training procedure. Last but not least, in addition to handling the physical objects, a worker also needs to be trained to digest and respond to rich data generated at the manufacturing site. To respond to these challenges, in this paper we describe the design of an AI-assisted training platform for manufacturing
workforce. The platform will collect rich data from both machines and workers. It will capture and analyze both macro and micro movement of trainees with the help of AI algorithms. At the same time, training for interaction with robot/cobot will also be covered. Mixed reality will be used to create in-situ experiences for the trainee.
more »
« less
- NSF-PAR ID:
- 10184361
- Date Published:
- Journal Name:
- AAAI 2020 Spring Symposium, AI in Manufacturing
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
In September 2019, the fourth and final workshop on the Future of Mechatronics and Robotics Education (FoMRE) was held at a Lawrence Technological University in Southfield, MI. This workshop was organized by faculty at several universities with financial support from industry partners and the National Science Foundation. The purpose of the workshops was to create a cohesive effort among mechatronics and robotics courses, minors and degree programs. Mechatronics and Robotics Engineering (MRE) is an integration of mechanics, controls, electronics, and software, which provides a unique opportunity for engineering students to function on multidisciplinary teams. Due to its multidisciplinary nature, it attracts diverse and innovative students, and graduates better-prepared professional engineers. In this fast growing field, there is a great need to standardize educational material and make MRE education more widely available and easier to adopt. This can only be accomplished if the community comes together to speak with one clear voice about not only the benefits, but also the best ways to teach it. These efforts would also aid in establishing more of these degree programs and integrating minors or majors into existing computer science, mechanical engineering, or electrical engineering departments. The final workshop was attended by approximately 50 practitioners from industry and academia. Participants identified many practical skills required for students to succeed in an MRE curriculum and as practicing engineers after graduation. These skills were then organized into the following categories: professional, independent learning, controller design, numerical simulation and analysis, electronics, software development, and system design. For example, professional skills include technical reports, presentations, and documentation. Independent learning includes reading data sheets, performing internet searches, doing a literature review, and having a maker mindset. Numerical simulation skills include understanding data, presenting data graphically, solving and simulating in software such as MATLAB, Simulink and Excel. Controller design involves selecting a controller, tuning a controller, designing to meet specifications, and understanding when the results are good enough. Electronics skills include selecting sensors, interfacing sensors, interfacing actuators, creating printed circuit boards, wiring on a breadboard, soldering, installing drivers, using integrated circuits, and using microcontrollers. Software development of embedded systems includes agile program design, state machines, analyzing and evaluating code results, commenting code, troubleshooting, debugging, AI and machine learning. Finally, system design includes prototyping, creating CAD models, design for manufacturing, breaking a system down into subsystems, integrating and interfacing subcomponents, having a multidisciplinary perspective, robustness, evaluating tradeoffs, testing, validation, and verification, failure, effect, and mode analysis. A survey was prepared and sent out to the participants from all four workshops as well as other robotics faculty, researchers and industry personnel in order to elicit a broader community response. Because one of the biggest challenges in mechatronics and robotics education is the absence of standardized curricula, textbooks, platforms, syllabi, assignments, and learning outcomes, this was a vital part of the process to achieve some level of consensus. This paper presents an introduction to MRE education, related work on existing programs, methods, results of the practical skills survey, and then draws conclusions based upon these results. It aims to create the foundation for standardizing the development of student skills in mechatronics and robotics curricula across institutions, disciplines, majors and minors. The survey was completed by 94 participants and it was clear that there is a consensus that the primary skills students should have upon completion of MRE courses or a program is a broader multidisciplinary systems-level perspective, an ability to problem solve, and an ability to design a system to meet specifications.more » « less
-
The necessity for educational programs in advanced manufacturing became prominent during the economic crisis in 2007 when the demand of industrial plants was for already trained highly-skilled laborers. To respond to this demand, many advanced manufacturing educational pro-grams, such as mechatronics, were developed in community and technical colleges. Since it was officially defined in the United States Congress in 2015, Smart Manufacturing (SM) has increasingly been under the spotlight. However, current efforts in deploying SM technologies in the US do not provide a workforce trained to utilize and perform SM technologies and techniques. Graduates of mechatronics and other advanced manufacturing programs remain mostly unaware of the technologies of Smart Manufacturing, such as Internet of Things (IoT) and Cyber Physical Systems (CPS), Industry 4.0 standards, and the capacity and range of applications of additive manufacturing and high-precision subtractive manufacturing technologies from tooling to end-user products. The programs currently available do not provide workforce training on SM technologies that target community and technical colleges, which supply a significant percentage of the industrial workforce. In the project Smart Manufacturing for America’s Revolutionizing Technological Transformation (SMART2), this gap in workforce training is met by providing the needed training to career technical education (CTE) and STEM educators in mechatronics and engineering technology. This project is a collaborative effort among three institutions and provides professional training for faculty of advanced manufacturing education programs and an online knowledge-base platform for educators and manufacturers, as well as on-ground training work-shops and educational modules.more » « less
-
null (Ed.)Advancements in Artificial Intelligence (AI), Information Technology, Augmented Reality (AR) and Virtual Reality (VR), and Robotic Automation is transforming jobs in the Architecture, Engineering and Construction (AEC) industries. However, it is also expected that these technologies will lead to job displacement, alter skill profiles for existing jobs, and change how people work. Therefore, preparing the workforce for an economy defined by these technologies is imperative. This ongoing research focuses on developing an immersive learning training curriculum to prepare the future workforce of the building industry. In this paper we are demonstrating a prototype of a mobile AR application to deliver lessons for training in robotic automation for construction industry workers. The application allows a user to interact with a virtual robot manipulator to learn its basic operations. The goal is to evaluate the effectiveness of the AR application by gauging participants' performance using pre and post surveys.more » « less
-
International collaboration between collections, aggregators, and researchers within the biodiversity community and beyond is becoming increasingly important in our efforts to support biodiversity, conservation and the life of the planet. The social, technical, logistical and financial aspects of an equitable biodiversity data landscape – from workforce training and mobilization of linked specimen data, to data integration, use and publication – must be considered globally and within the context of a growing biodiversity crisis. In recent years, several initiatives have outlined paths forward that describe how digital versions of natural history specimens can be extended and linked with associated data. In the United States, Webster (2017) presented the “extended specimen”, which was expanded upon by Lendemer et al. (2019) through the work of the Biodiversity Collections Network (BCoN). At the same time, a “digital specimen” concept was developed by DiSSCo in Europe (Hardisty 2020). Both the extended and digital specimen concepts depict a digital proxy of an analog natural history specimen, whose digital nature provides greater capabilities such as being machine-processable, linkages with associated data, globally accessible information-rich biodiversity data, improved tracking, attribution and annotation, additional opportunities for data use and cross-disciplinary collaborations forming the basis for FAIR (Findable, Accessible, Interoperable, Reproducible) and equitable sharing of benefits worldwide, and innumerable other advantages, with slight variation in how an extended or digital specimen model would be executed. Recognizing the need to align the two closely-related concepts, and to provide a place for open discussion around various topics of the Digital Extended Specimen (DES; the current working name for the joined concepts), we initiated a virtual consultation on the discourse platform hosted by the Alliance for Biodiversity Knowledge through GBIF. This platform provided a forum for threaded discussions around topics related and relevant to the DES. The goals of the consultation align with the goals of the Alliance for Biodiversity Knowledge: expand participation in the process, build support for further collaboration, identify use cases, identify significant challenges and obstacles, and develop a comprehensive roadmap towards achieving the vision for a global specification for data integration. In early 2021, Phase 1 launched with five topics: Making FAIR data for specimens accessible; Extending, enriching and integrating data; Annotating specimens and other data; Data attribution; and Analyzing/mining specimen data for novel applications. This round of full discussion was productive and engaged dozens of contributors, with hundreds of posts and thousands of views. During Phase 1, several deeper, more technical, or additional topics of relevance were identified and formed the foundation for Phase 2 which began in May 2021 with the following topics: Robust access points and data infrastructure alignment; Persistent identifier (PID) scheme(s); Meeting legal/regulatory, ethical and sensitive data obligations; Workforce capacity development and inclusivity; Transactional mechanisms and provenance; and Partnerships to collaborate more effectively. In Phase 2 fruitful progress was made towards solutions to some of these complex functional and technical long-term goals. Simultaneously, our commitment to open participation was reinforced, through increased efforts to involve new voices from allied and complementary fields. Among a wealth of ideas expressed, the community highlighted the need for unambiguous persistent identifiers and a dedicated agent to assign them, support for a fully linked system that includes robust publishing mechanisms, strong support for social structures that build trustworthiness of the system, appropriate attribution of legacy and new work, a system that is inclusive, removed from colonial practices, and supportive of creative use of biodiversity data, building a truly global data infrastructure, balancing open access with legal obligations and ethical responsibilities, and the partnerships necessary for success. These two consultation periods, and the myriad activities surrounding the online discussion, produced a wide variety of perspectives, strategies, and approaches to converging the digital and extended specimen concepts, and progressing plans for the DES -- steps necessary to improve access to research-ready data to advance our understanding of the diversity and distribution of life. Discussions continue and we hope to include your contributions to the DES in future implementation plans.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
