Demonstrating feasibility of autonomous supply chains with IoT

Posted on

< Project Overview >

Introduction

This project aimed to demonstrate the feasibility of autonomous supply chains by coupling software agents with Internet of Things. The project team (Dr Alexandra Brintrup and Dr Liming Xu) in the Manufacturing Analytics Group at IfM collaborated with a local Cambridge tech company – Fetch.ai (https://fetch.ai/) to develop an autonomous supply chain management (SCM) demonstrator, in which fetch.ai offered their new decentralized autonomous agent framework and corresponding technical support for implementing such a prototype using their agent development technology.

Project aims

Several researchers and practitioners have in the past proposed to automate data sharing and low level operational decision making in supply chains. Doing so, would enable more seamless and coordinate operations, removing wastage points from the supply chain. However a lack of technical infrastructure makes investment into these systems difficult. Although multiple innovative and successful approaches for autonomous SCM have been developed in academic realm for many years, they have not been implemented in industrial settings.

Accordingly, the goal of this project is to address the challenges that impede the realization of autonomous SCM in industry by integrating research outputs, bringing them to higher TRL; so that transition to industrial settings is possible and the corresponding companies which are keen to promote supply chain automation would find thoughts and guidelines from this research.

What was done?

The main activities this project focused on were as follows:

• Literature review on the agent-based supply chain management;
• Screening relevant technologies for realising autonomous SCM;
• Demonstrator development that showcased how sensor data could be shared between organisational boundaries on a food logistics chain, so that product quality could be guaranteed and traced;
• Dissemination activities: including exhibiting supply chain automation demonstrator (v1) in Advanced Engineering, NEC Birmingham, Nov 2019; webinar (autonomous supply chains with agent-based system) for industrial audience that interested in fetch.ai agent development framework; presenting the work in Pitch-in manufacturing day, Dec 2019, Nov 2020; Exhibition (demonstrator v2) in Digital Manufacturing Week (Nov 2020);
• Submission of a Journal paper that reviewed learnings from the project.

Results

  • This project has successfully delivered two integrative demonstrators for autonomous supply chains. Specifically, the first is demonstrating the automatic selection of suppliers, the use of IoT in monitoring the ambient conditions of the transport vehicles, rerouting when emergent events (e.g., traffic jam) occur, and prevenance and analytic summary of the product. The second demonstrates an integrated agent-based autonomous SCM platform, including procurement, transport monitoring, negotiation, inventory update, and product quality summary.
  • Additionally, a manuscript titled “Will bots take over the supply chain? Revisiting agent-based supply chain automation” was submitted for a Journal publication and another one is in preparation.
  • Importantly, a continuous collaboration with Fetch.ai was established, who, after our project together, who have continued to develop Autonomous Supply Chains as a product. They have developed collaborations with company Festo to implement a similar concept, and our research group is actively involved in the project providing research directions.
  • A successful grant application was made to the Connected Everything Network Plus (EPSRC), to further develop the Autonomous Supply Chains concept for facilitating collaborative logistics.

Deliverables and other tangible outputs

1. Software demonstrator: IoT based Smart Supply Chain Automation Demonstrator (v1)

Figure 1. Screenshots (gif) of the 1st version demonstrator

This demonstrator is an example to showcase how legacy systems could be integrated with Agent Based Systems (ABS) to build an IoT-based supply chain automation platform. Key functionalities of this demonstrator are automatic procurement and supplier selection, monitoring the transport, product quality tracking and analysis, dynamic pricing. This demonstrator screens and explores the technologies that can support the implementation of such systems: web interfaces (GIS, visualisation) together with multi-agent backend were eventually adopted, in which messaging broker was employed to connect the web systems with the multi-agent system. Figure 1 shows the screenshots (in GIF) of the demonstrator, which was exhibited in Advanced Engineering, NEC Birmingham, Nov 2019. The demonstrator attracted interests from the exhibition audience and lead to the follow-up Connected Everything (https://connectedeverything.ac.uk/) grant partnered with a London-based supply chain analytic company. This demonstrator (1st version) shed light on the initial design of an integrated end-to-end supply chain automation system.

2. Software demonstrator: IoT based Smart Supply Chain Automation Demonstrator (v2)

Figure 2. Screenshots (gif) of the 2nd Version demonstrator

The lessons learned from the development of this system contribute to the design of the 2nd version autonomous supply chain demonstrator. We almost redesigned the platform: refactored the 1st version demonstrator and streamlined the supply chain automation pipeline. Through the partnership with Fetch.ai, the 2nd demonstrator used its new developed autonomous economic agent framework to create a multi-agent backend that support negotiation using FIPA contract-net protocol. This demonstrator mainly includes four parts: ordering, transport monitoring and service analysis, agent chatting box, and IoT reading visualisation. While only two types of agents were designed in the 1st version demonstrator, this demonstrator implemented five types of agents that represent primary stakeholders in supply chains: retailer, wholesaler, supplier, logistics agent, and third-party logistics agent. This demonstrator, as shown in Figure 2, was presented to industrial audience through a Fetch.ai organised webinar. This demonstrator attracted many industrial attentions and has been covered by multiple news outlets. A video presentation of this demonstrator can be found in Fetch.ai official Youtube channel (https://www.youtube.com/watch?v=WFcLAif-cHo) or vimeo (https://vimeo.com/440352160).

3. Publications
A paper, entitled “Will Bots Take Over the Supply Chain: Revisting Agent-based Supply Chain Automation” was submitted for publication. This paper systematically reviewed the work in the domain of agent-based systems (ABS) in supply chain management, discussed the identified barriers that impede that adoption of agent technology in industrial sector and presented the future research directions in the intersection of ABS and SCM.

We are preparing other papers that describe the technical aspects of the two demonstrators and are expecting to submit for publication in next coming months.

Impact

The impact of this project had centres on general public engagement activities and sector stakeholder engagement activities. The engagement activities include:

  • The demonstrator (v1) was exhibited in Advanced Engineering, NEC Birmingham, 30-32 Oct 2019
  • The two demonstrators were presented in the Pitch-in Manufacturing Day on 08 Dec 2019 and on 09-08 Nov 2020, respectively
  • We report the work on autonomous supply chain to industrial audience (around 70 attendees) in Fetch.ai organised webinar on 22 Oct 2020
  • We exhibited the demonstrator (v2) in the Digital Manufacturing Week (online) in 09-13 Nov
    2020

Social media and online outlets have also help to disseminate this research, such as:

An impact for our partner Fetch.ai was its investment into developing Autonomous Supply Chains as a product. They have developed a follow on collaboration with a machine manufacturer, Festo, and a number of others, to implement a similar concept, and our research group is actively involved in the project providing research directions.

An impact for our research group was a successful grant application which was made to the Connected Everything Network Plus (EPSRC), to further develop the Autonomous Supply Chains concept for facilitating collaborative logistics. That project will use the same agent based platform, and is partnered with Value Chain Lab (VCL) to implement the concept.

Next steps

As a result of the project outcomes, further foreseen activities will centre on:

  • Establishing our group’s technical expertise on Autonomous Supply Chains through a number of case studies including the follow-up Connected Everything project on collaborative logistics.
  • Contributing to the Fetch.ai / Festo project on agent based supply chains
  • Producing a follow on Journal paper on the technical results.
  • Actively seeking how ASC can help in emerging concepts such as Supply Chain Digital Twins. We have a workshop planned in mid-April 2021 with more than 10 industrialists to discuss how similar concepts can be applied.
  • Providing technical foundation to apply grants from such as EPSRC Digital Manufacturing theme to conduct research on building a practical, scalable autonomous supply chain platform and the underlying technologies.

Lessons learned

This project has demonstrated how IoT, agent technologies, and web technology can be combined together to achieve an integrated IoT-based smart autonomous supply chain platform by developing two relevant demonstrators. As a result, this University has benefited from the technological recognition of industrial stakeholders in the supply chain and manufacturing sector. We as the researchers undertake this project have benefited from :1) collaborating with industrial partners; 2) working in an intra-university research team; 3) industrially grounded our expertise in using ABS, decentralized technology, interface development, visualisation of IoT data, for supply chain operations.

With the accumulated knowledge in the project, it would have been better to systematically review the agent-based supply chain systems at the very beginning of this project rather than in the middle of this project. Due to the pressure in demonstrator deliverables we have dived into multiple agent development frameworks.

As this project aimed to promote the TRL and adoption of autonomous supply chain, regularly talking to various industrial stakeholders helped in shedding light on the design of platforms or systems that are desirable for relevant companies. Well-written technical documentation would greatly flatten the learning curve of a specific technology, e.g., the documentation of the agent development framework.

While collaborating with industrial partners, it is better to have a clear research plan that is not vulnerable to be affected by the industrial partner’s changing priorities. For example, mid-way through the project our partner’s software framework has changed, which meant that we had to start from scratch to move onto the next version of the platform – although this was not particularly beneficial from an academic perspective as the same functionality existed in an earlier version, doing so aided the partner in testing how their new version works and helped them identify software bugs.

Access to deliverables, resources and media content

The shareable resources of this project can be found at:

• Project website: https://www.ifm.eng.cam.ac.uk/research/manufacturing-analytics/current-projects/autonomous-supply-chains-with-iot/

• Information about IoT-based supply chain automation demonstrator (v1): https://www.ifm.eng.cam.ac.uk/insights/digital-manufacturing/iot-smart-supply-chain-automation-system/

• Video presentation of IoT-based supply chain automation demonstrator (v2): https://vimeo.com/438586015 and https://www.youtube.com/watch?v=WFcLAif-cHo

What has Pitch-In done for you?

The results of this Pitch-In programme have been crucial to help the successful bid of the Connected Everything grant, facilitated collaboration with industrial partners (Fetch.ai, VCL, Festo), that would not have been possible otherwise.

We submitted a journal paper with one other under preparation establishing our academic credentials in this topic area.

Project lead:

Dr Alexandra Brintrup, Lecturer in Digital Manufacturing at the University of Cambridge.

Partners:

< Related Content >