The goal of the research profile Virtual Factory with Knowledge-Driven Optimization (VF-KDO) is to deliver radical innovations in manufacturing research essential to design and operate next-generation manufacturing systems.
It will focus on KDO, interactive analytics, as well as multi-level and multi-disciplinary modeling and simulation, including the modeling of humans. A unique concept proposed in VF-KDO is: knowledge extracted for decision support is achieved through systematically exploring, e.g., using advanced data mining and other interactive visual analytics techniques on optimal solutions generated via many-objective optimizations on virtual factory models. In this regard, we will give some updates about a toolset that we have been developing for knowledge extraction in this newsletter.
University of Skövde
FPM data mining and graph visualization
We start by presenting the FPM data mining and graph visualization tool prototype developed through the synergy between OPT-KNOW and INTERACT for extracting knowledge from the industrial case studies. In particular, the snapshot below shows how the robotic simulation-based optimization results from a model developed together with ABB and ElektroAutomatik (a former partner in Synergy-KDDS project) can be analyzed to generate the rules that capture the attributes of the optimal trade-off (cycle time vs. energy consumption) solutions. In collaborations with the FLOW team who is helping Arla and IKEA Industry in building Discrete-Event Simulation (DES) models for their production lines (Figure 2 and Figure 3, respectively), we anticipate more FPM results will be generated for line-level in 2020Q3-4.
Figure 1. Extracting rules from a robotic cell optimization using the FPM data mining and graph visualization prototype.
Figure 2: The digital twin concept implementation for Arla’s Bregott production line.
Figure 3: The DES model developed for IKEA’s Hultsfred factory.
OP30 Tool Indexing Optimization
OP30 at VCC represents so far the most successful, multi-level company showcase that we started in the Synergy-KDDS project but being carried over to continue in VF-KDO led by the PROCESS team. Some important tool indexing optimization results have been obtained in which the tool indexing time can be reduced by 60%.
The significance of this potential improvement is still being validated by VCC’s engineers, but the current results are already reported in Kaveh Amouzgar’s publications. You can also read his whitepaper on the site. If this significant improvement can be validated by VCC, it will represent a very significant impact on the manufacturing process of the company because the same type of tool indexing is used in multiple machining centers.
HUMAN & ROBOT in VGTO and Scania
In collaboration with Scania and FCC, HUMAN researchers have developed a digital twin solution wherein one can record real-time motion data which can then be transmitted to a DHM software. This enables visualization and ergonomic analysis in the virtual world - of work done in the real world. They also developed a flexible platform that can receive movement data from a variety of sources, such as DHM simulations, motion capture, smart textiles or data-driven motion prediction methods.
Programmed into platform
Based on incoming data, a variety of posture-based ergonomics assessments may be made when programmed into the platform. The evaluation can be done in real-time or as an aggregated load dose. The development is carried out in collaboration with the mosim.eu project and involves VGTO and Scania.
Digital platform for simulation-based multi-objective optimization is developed
Additionally, in collaboration with OPT-KNOW and LINK, a digital platform for simulation-based multi-objective optimization is developed for the simultaneous analysis of ergonomics and production criteria. The platform is general-purpose and can communicate with most production flow and DHM tools as long as a specific input/output module per software is created as data from/to different software differs. The technology is developed to be used at both the workstation level and production line level, as shown in the figure below.
Figure 4: The optimization platform connecting to a DHM software, being applied to the Scania go-cart assembly stations. Read PhD student Aitor Iriondo Pascual's whitepaper, Optimizing Ergonomics and Productivity by Connecting Digital Human Modelling and Production Flow Simulation Software.
Application study about collaborative robot in the final assembly
Particularly at VGTO, the ROBOT team has mainly focused on the application study about introducing a collaborative robot in the final assembly in Skövde at station S2700. The case study uses a virtual engineering framework jointly developed with Volvo staff, in which various solutions can be effectively evaluated and tested virtually.
In this study, we also investigated a potential quality increase of the station through vision systems that are integrated into the collaboration robot. The results of this study were also published and presented at the International Conference on Manufacturing Research, click to check out Niklas Land’s whitepaper.
Figure 5: Virtually testing and evaluating the HRC solutions for VGTO’s S2700 engine assembly station.
New competencies developed in this case
In terms of new competencies, VF-KDO has partially taken over the good research results from the EU-ManuWork project, particularly both the load balancing optimization methods as well as the visualization of load balancing results using Augmented Reality (Microsoft HoloLens). These research topics are now explicitly included in VF-KDO and immediately find applications, like load-balancing optimization (LBO) for the Scania go-cart assembly, as seen in Figure 4 as well as the VDM Room (see below).
Actively Planning Joint Company Studies
Within VF-KDO, joint company studies are highly encouraged right from the beginning. Already at the end of November 2019, FlexLink and Arla had started the discussions on a joint business case in conveyor cell re-design. Together with FlexLink, the OPT-KNOW researchers and some Master student project have extended their FlexLink Design Tool (FLDT) to connect to our optimization platform, which can be tested on Flexlink’s existing models as well as for the joint business case mentioned.
Recently, on May 28 we had the first IKEA-ABB meeting about some collaboration or business case within VF-KDO within a wide range of topics which are related to AI and Robotics, including vision-guided robots, adaptive gripping and autonomous picking and human-robot collaborations (HRC) in production and warehouses. Apart from the joint business case, co-production at the profile level is promoted by identifying the common scientific questions and/or industrial problems that spread across different companies and subject areas. A good example is that both VGTO and Scania have defined two different two-level assembly line design and balancing cases that involve HRC. Therefore, there have been some joint activities between OPT-KNOW, ROBOT and HUMAN to look into some common solutions to the two cases in the two truck manufacturing companies.
Inspired by the “Content Marketing” principles, the virtualfactories.se website and its contents will be the main focus that we will work on to “reach” the target groups; through social media like Twitter @Virtualdriven and LinkedIn #vfkdo, conferences like Katrineholm Konferens, Swedish Production Symposium, etc., together with Search Engine Optimization (SEO) techniques; we would like to point all interesting parties in our target groups to locate and access the web site. In our communication plan, there are clear descriptions of our target groups:
Employees of the partner companies who are not directly working with the co-production but have heard about the project; the web’s rich content is intended to ease our contact persons at the companies to introduce the project to their colleagues.
Swedish industrial companies (especially SME) which are not our partners in VF-KDO (or any projects) but want to know the university’s key research.
Other national and international researchers who are conducting research related to Digital Twins, Industry 4.0, Cyper-physical Production Systems and Artificial Intelligence applications, etc., can easily find the web and its publication list when searching these keywords.
We have a new document server
It is worthy of mentioning the web allows users (researchers and partners) to directly login to the new documents server, i.e., NextCloud, as well as all the news and events about the profile, the subject areas, or related company visits/activities, are listed. With all these design features, we aim at having the web as a common platform for information dissemination.
In addition, with the establishment of virtualfactories.se, the complete list of publications affiliated to VF-KDO can be found under virtualfactories.se/publications. Publications by a specific subject area can be found under the subject area description on the site.
ExpertKompetens Education Programme
VF-KDO has some strong connections with the ExpertKompetens programme, Virtual Factory, to develop industry courses as one form of knowledge transfer of the research outcomes. The idea is that there will eventually be a course for each subject area within the VF-KDO, where researchers from each sub-area hold the different courses. In this way, we promote a knowledge transfer between research and industry, and we enable companies to absorb the results of VF-KDO in a wide spectrum.
During 2020Q1, there have been five pilot courses ran within ExpertKompetens, Stage 1, in which four of them are responsible by researchers/staffs affiliated with VF-KDO using some of the research outcomes as planned:
Opportunities and benefits of simulation as decision support
Integrated virtual commissioning
Optimization methods for industry
Industrial human-robot collaboration through virtual tools
An application to Stage 2, called WISER, has been submitted, we are looking forward to developing the complete programme when the project is granted.
The VDM Room
On the 16th December 2019 ASSAR research demonstration day, we presented our vision about building a Virtual Decision Making (VDM) room at ASSAR. It is identical to the “war room” concept in business or industry: equipped with big screens surrounding inside the room and VR/AR devices for the participants to have group meetings that involve strategic, tactical and decision-making activities. Nonetheless, the VDM room will have a stark contrast with an ordinary war room: it will be used, not only to demonstrate our unique simulation, optimization and group decision-making support methodology introduced in VF-KDO, but allow industrial partners and companies to borrow/hire the room for their decision-making activities with the support of technical staff from ASSAR and the The University of Skövde.
that an analyst has run a many-objective optimization on a 3D robotic cell design model and generated a large set of optimized trade-off solutions. A group of manufacturing managers/executives, sitting in the room, everyone is wearing a HoloLens. One of them points to a solution that s/he prefers to check the value of the decision variables representing the configuration of the robot cell design. By clicking the screen on the wall, the 3D model of the solution will be visualized on the HoloLens so that everyone, in the room (and remotely connecting to the room through the Internet) can view the design in her/his own perspective. By supporting this way of ‘transparent’ group decision-making process with the support of state-of-the-art VR/AR and visualization technologies, combined with our unique many-objective optimization and knowledge extraction methods, the industry will experience and has the possibility to utilize a novel way of designing/improving their manufacturing systems, which is the goal of VF-KDO.
The construction of the VDM room will be completed in early July 2020 with the inspection date booked on July 3. Have it up and running for demonstration purposes is anticipated by the end of the year 2020.