< Project Overview >
This project proposes to contribute a methodology to identify, classify and select low cost technologies compliant with the Digital Manufacturing on a Shoestring approach. In order to validate our approach, we also propose the development of demonstrators derived from the activities of the Digital Manufacturing on a Shoestring project. As a result, we will identify and capture key enablers that would underpin the industrial adoption and deployment of these demonstrators in manufacturing small and medium enterprises (SMEs).
In order to deliver this project, the implementation of digital solutions for manufacturing (demonstrators) were proposed to address the following barriers: a) lack of knowledge on how to deploy low cost IoT systems and c) lack of understanding in how IoT will/can generate value in a given application domain.
What was done?
The main activities this project focused on were 1) contributing a systematic approach for recognising, classifying and selecting off-the-shelf technologies to enable low cost industrial IoT deployments and 2) the development of a legacy control panel’s information capture demonstrator validated through interviews with SME partners of the Digital Manufacturing on a Shoestring project.
This project has successfully delivered a systematic approach for the classification of low cost digital technologies.
Specifically, this approach focused on establishing the requirements for low cost technologies to be part of low cost industrial digital solutions, proposing an approach to classify low cost technologies to support such type of solutions, and demonstrating how this approach provides a clear pathway to provide appropriate technologies. In addition, one demonstrator within the manufacturing domain was also delivered. Specifically, this demonstrator focused on legacy control panels data extraction using cameras, edge computing systems and real-time object detection to recognise and capture legacy panel status.
Deliverables and other tangible outputs
1. Low Cost Digital Technologies Classification
One of the challenges associated with the development of low cost industrial digital solutions is collecting and selecting the right low cost technologies to realise an effective system implementation. Thus, the purpose of this approach is to offer an initial three-fold response that addresses this challenge by establishing the requirements for low cost technologies to be part of low cost industrial digital solutions, proposing an approach to classify low cost technologies to support such types of solutions, and demonstrating how this approach provides a clear pathway to provide appropriate technologies. In particular, the applied approach enlarges on a methodology that has been successfully employed to create classification systems on a myriad of domains such as mobile solutions, software failures and blockchain applications to name a few. Hence, we demonstrated its potential to create classification systems for five specific technology groups: data collection, data storage, visualisation, analytics and computing hardware. Figure 1 shows the relationship between technologies capture processes realised either manually or semi-automatically, technologies classification system and their storage.
The resulting classification systems are realised and made available throughout a digital platform from where a solution design tool retrieves the information needed to help end-users make an informed implementation without falling into low level technical specifications and, as a consequence, eliminating the need for specifically skilled assistance (see Figure 2).
2. Legacy Control Panel Data Extraction
Vision systems can be applied to various tasks in manufacturing since they offer many advantages such as enabling non-invasive and non-touch monitoring as well as information capture by using imaging software. However, these type of systems could be cost-prohibitive for a small to medium size business, not just because of the perceived monetary value of the hardware and software, but also because of the perceived complexity of integrating such solution with production processes and its associated maintenance. The vision system demonstrator reported here uses a low cost camera connected to a small single-board computer that runs a software module implemented with OpenCV (see Figure 3 left). This is a non-invasive solution device designed to stand at a specific distance from the control panel of interest, i.e. no tampering or physical connections to the control panel required (see Figure 3 right). In particular, the software module is an image processing unit trained to recognise when individual panel components such as knobs, dials, lights or displays change their associated state, e.g. a light goes from on to off or a dial moves sideways. All this information change is captured in real time, stored and linked to underlying manufacturing processes.
The impact this project had centres on general public engagement activities and sector stakeholder engagement activities. The general public engagement activities involved:
- Showcasing and installing the Legacy Panel Data Recognition demonstrator in Photofabrication Ltd, a local SME dedicated to chemical etching based manufacturing.
- A two-day virtual hackathon event organised for nearly 150 students of the University of Cambridge, at the Institute for Manufacturing, on 24-25 October 2020.
The sector stakeholder engagement activities involved:
- Demonstrators were shown virtually in online Digital Manufacturing workshops and webinars, organised by the Institute for Manufacturing, University of Cambridge.
As a result of the project outcomes, further foreseen activities will centre on:
- Providing both research and technical support for the EPSRC project Digital Manufacturing on a Shoestring. In particular, these activities will centre on facilitating a digital platform to integrate within the digital Solution Configurator tool.
- Providing both research and technical support for the Pitch-In project IoT-based Devices for Non-critical Support in Hospitals. In particular, these activities will centre on how to adapt and translate existing IoT demonstrators developed for manufacturing into non life-critical medical devices.
This project has demonstrated 1) how to systematically collect and select the right low cost technologies to realise an effective system implementation and 2) how manufacturing SMEs could gain access to digital vision-based manufacturing capabilities by building one digital low-cost IoT solution while showcasing its outcomes and transferring knowledge through public engagement activities. As a result, this University has benefited from: 1) exposing students and academics to a novel methodology for prototyping digital manufacturing capabilities; and 2) broadcasting our approach for building low cost digital manufacturing capabilities through different sectors of the manufacturing community.
The value of this work lies in the practicality of applying a classification method to already defined groups of openly available, off-the-shelf and low cost technologies. Since the resulting classification systems provide a non-hierarchical structure built in terms of application-related features, it also offers a wide range of end-users from manufacturing SMEs the ability to effortlessly access and interpret the achieved results. Therefore, having had an opportunity to validate our approach with a group of SMEs, it would have been beneficial to measure its practical impact as well as to assess several technical aspects.
One of the key aspects of this research project was its underpinning interdisciplinary approach. Therefore, having the ability to collaborate with related research programmes, e.g. those exploiting low-cost, off-the-shelf digital IoT technologies in different domains or those looking at digitally enhancing legacy manufacturing equipment with vision systems would have been useful to both validate our methodology and knowledge exchange.
What has Pitch-In done for you?
The results of this Pitch-In programme has been crucial to help quick start the EPSRC Digital Manufacturing on a Shoestring project since it has enabled an initial exploration of IoT technologies to deliver proof-of-concept demonstrators for low-cost digital manufacturing capabilities.
Professor Duncan McFarlane, The University of Cambridge