Multi-agent AI for coordinating a hardware gig economy

Gig economies have turned many traditional business models on their heads, and are poised to do the same yet again, but for one of the most traditional economies: undergraduate education. In particular, universities are including ever more activity in their courses so as to enhance the attributes of their graduates. Yet many of the most traditional campuses face strict estates constraints that are opening the way for online, remote laboratory alternatives to flourish. Using a browser to access camera feeds, live data, and control remote hardware that could be across town, or half a world away, is already a proven concept in the distance learning sector, where the lead supervisor’s work at the Open University attracted awards from Times Higher Education, The Guardian, The Global Online Labs Consortium and National Instruments. Educational literatures on non-traditional laboratories (online real, virtual, and simulated labs) are unanimous in pointing out that a mix of traditional and non-traditional laboratories offers better overall education than traditional laboratories alone. Now at the University of Edinburgh, the challenge is to build organically-growable remote laboratory installations that can occupy small portions of otherwise unusable estate, and federate together across multiple campuses in the University of Edinburgh. Subsequently, the federation will be extended to similar forthcoming installations by collaborators at the Universities of Manchester and Cardiff. Unlike human-driven gig economies such as providing transport or services, the remote laboratory experiments are not attended to by humans. This is because many experiments in engineering, physics, maths, and computer science do not require it, although it is also possible to put humans in the loop for remote experiments in areas like chemistry, medicine and veterinary medicine where materials handling requirements can make it cost-effective. Hence, there is a need for the experiments to be self-aware, self-sufficient, and able to earn their keep by servicing students from both their local university, as well as other universities. This requires a complex multi-agent system with accounting for usage (micro-payments), trust (reliability, data provenance), and efficacy (evaluation). The lead supervisor’s work on a self-sufficient network of agents in a centralised facility has shown the utility of adopting the multi-agent approach in this context. For this next step in the challenge, an entirely new decentralized and larger scale architecture is needed. The research work will encompass contributing to an open-source multi-agent infrastructure code-base and exploring multi-agent interactions in both simulation and in real networks of remote experiments, generating research outputs in theoretical and applied aspects ranging from reliability in distributed systems to dynamic pricing in time-sensitive markets, with potential applications in the our forthcoming networks of remote laboratories as well as more diverse areas such as networks of renewable energy generators, and autonomous transport systems amongst others. In this way the work will inspire new approaches to multi-agent AI systems as well as provide a test-bed for exploring them.

Further Information: 

Presentations on remote lab infrastructure pooling from an academic user point of view are available at An earlier centralised agent network was coordinated using AMQP protocols, and led to several awards for the teaching offered by the remote laboratory. For scaling to multi-institution usage, scalability protocols offer advantages. A demonstration implementation of a small-scale multi-agent communication network in python 3 using scalability protocols is available here Ideally a more performant language with good concurrency support will be used in the research work, for example golang, or possibly rust.


This post will remain open until filled.

Closing Date: 

Wednesday, March 25, 2020

Principal Supervisor: 

Prof Timothy Drysdale


Minimum entry qualification - an Honours degree at 2:1 or above (or International equivalent) in a relevant science or engineering discipline, possibly supported by an MSc Degree.

The project is suitable for a student with a top MSc or first-class bachelor's degree in computer science, electronics, mathematics, physics, or a related discipline.

Previous coursework or experience in concurrent programming languages, distributed systems, artificial intelligence and economics is desirable, although we do not expect students to have all four of these. Because of the scale of the distributed agent networks involved, a strong programming background will be essential for this project.


Further information on English language requirements for EU/Overseas applicants.


Tuition fees + stipend are available for Home/EU students (International students can apply, but the funding only covers the Home/EU fee rate).

Further information and other funding options.

Informal Enquiries: