We engineer swarm systems across scales, from trillions of nanoparticles, or millions of cells, to thousands of robots.

The challenge is to discover which individual actions give rise to desired swarm behaviours.

We take inspiration from nature, use machine learning, or the power of the crowd to engineer swarms.


Robot Swarms

Swarm Logistics

Many robots (10-1000+) working together to facilitate logistics in industrial settings promise to improve productivity, and enable specialised pipelines for a variety of items, including bespoke, complex form, high-value products. In swarm robotics, many simple robots, following simple rules based on local interactions with other robots or their environment, give rise to seemingly complex behaviours. By taking a swarm engineering approach, we will create systems that are adaptable to changing environments and conditions, scalable to chosen numbers of robots, robust to individual robot failures, and usable out of the box without dedicated infrastructure.

Main Researchers: Emma Milner, Simon Jones
Supervisors: Manuel Giuliani, Mahesh Sooriyabandara, Sabine Hauert
Funding: EPSRC FARSCOPE CDT, EPSRC IAA Commercialisation Award, EPSRC ICASE Award
More information: Frontiers in Robotics and AI paper, Advanced Intelligent Systems paper, swarm video.

Bio-inspired robot fish

Certain fish are better at shoaling than others. We aim to learn from fish morphology and behaviour studies to design a new type of underwater robot swarm.

Main Researcher: Elliott Scott
Supervisors: Christos Ioannou, Martin Genner, Sabine Hauert
Funding: EPSRC DTP

Expressive swarms

For swarms to be useful in real-world applications, users will need to easily interact with and understand them. By making swarms expressive, we are designing systems that are more intuitive and visual, facilitating human-swarm interactions and new artistic displays.

Main Researcher: Merihan Alhafnawi
Supervisors: Paul O’Dowd, Sabine Hauert
Funding: UAE Scholarship
More information: ALIFE paper and video.

Automatic design of supervisory control of robot swarms

Supervisory control of swarms is essential to their deployment in real world scenarios to both monitor their operation and provide guidance. We explore mechanisms by which humans can provide supervisory control to swarms to improve their performance. Rather than have humans guess the correct form of supervisory control, we use artificial evolution to learn effective human-readable strategies.

Main Researcher: Elliott Hogg
Supervisors: Sabine Hauert, Arthur Richards
Funding: EPSRC FARSCOPE CDT, EPSRC Prosperity Partnership, Thales
More information: Swarm paper.

Stochastic swarms for environmental monitoring

Single-use jumping robots that are mass-produceable and biodegradable could be quickly released for environmental sensing applications. Such robots would be preloaded to perform a set number of jumps, at random directions and distances, removing the need for onboard energy and careful control. Stochastic swarms however require new algorithms that build on randomness and large-scale deployments to perform useful work.

Main Researcher: Julian Hird
Supervisors: Andrew Conn, Sabine Hauert

Aerial swarms for forest fire monitoring and mitigation

Swarms of large unmanned aerial vehicles have the potential to monitor high-risk landscapes for fires, and rapidly extinguish them by delivering an extinguishing payload. To be efficient and safe, and robust, we are designing new swarm algorithms tailored to the next generation of aerial platforms.

Main Researcher: Georgios Tzoumas
Supervisors: Tom Richardson, Sabine Hauert
Funding: Windracers LTD

Risk-sensitive robot swarms for effective environmental monitoring

A robot swarm could be used to deploy a dynamic network of sensors in a hazardous environment with a subset of robots that move closer to detected anomalies to inspect them at a higher resolution. We are using insight from behavioural biology and financial risk management to develop risk-sensitive swarm control algorithms.

Main Researcher: Edmund Hunt
Supervisor: Sabine Hauert
Funding: Royal Academy of Engineering UK Intelligence Community Postdoctoral Research Fellowship
More information: Nature Machine Intelligence commentary, and ALIFE paper.

Micro and Nano Swarms

Dynamic Optical Control of Micro-environments (DOME)

Control of microscopic systems such as engineered bacterial cells, mammalian cells and functionalised microparticles is challenging due to the difficulty in designing manipulation tools with sufficiently high spatiotemporal resolution to impact agents on a local level. Widely employed methods include magnetic field control and chemical or thermal induction, all of which have limited capability to provide a truly localised environment. By instead using light as a control mechanism, it is possible to achieve very high-resolution and rapid response. Owing to the prevalence of light responsive behaviour at small scales, an effective method for localised light control has potential in an array of fields including synthetic biology, microrobotics, optogenetics and micro-scale assembly.

Main Researchers: Ana Rubio Denniss
Supervisors: Thomas Gorochowski, Sabine Hauert
Funding: EPSRC DTP, EPSRC Institutional Support, EPSRC IAA Commercialisation Award
More information: MARSS paper, “Game of light” video.

Automatic design of swarming nanomedicine

Understanding how trillions of nanoparticles move and interact in tumour tissue could prove instrumental to improve tissue penetration and cellular update. We’re designing a computational framework to automatically design nanoparticles for optimal treatment and tools to validate results using a tumour-on-a-chip microfluidic device.

Main Researchers: Namid Stillman, Scott McCormick, Matthew Hockley
Supervisors: Adam Perriman, Sabine Hauert
Funding: EPSRC DTP, EBI Blackwell Institute, FET-OPEN H2020 EVO-NANO
More information: EVO-NANO website, npj Computational Materials review paper, Trends in Biotechnology review paper.

Automatic extraction and translation of swarm behaviours

Automatic understanding of natural swarms could help unlock mysteries of self-organisation in biology and serve as an inspiration in the design of swarm robot controllers. We are building a framework to automatically analyse swarm behaviours, extract their intention and the individual rules of agents, and translate them to new artificial swarms.

Main Researcher: Khulud Alharthi
Supervisors: Sabine Hauert
Funding: Kingdom of Saudi Arabia Scholarship

Growing de novo microbots

Complex functional structures emerge from the self-organisation of millions of cells continuously reconfiguring as they grow. We are exploring new building blocks and methodologies that would allow us to grow tiny controllable microrobots, and then combine these microrobots in swarms.

Main Researcher: Matthew Uppington
Supervisors: Sabine Hauert and Helmut Hauser

Social Impact

Leaving No One Behind: Educating Those Most Impacted by Artificial Intelligence

Robotics and AI are becoming more common in the workplace and everyday life. As has happened with previous technological advances, there is a risk of certain groups of the population being left behind as robotics and AI grows. Education is one method of minimising the number of people being left behind. Often those at risk are the same people who could not or would not avail of other formal education offerings, and whose employers would be unlikely to provide such education. Thus, any education needs to be specifically designed with these people and their needs in mind. We are conducting research to understand how to design such an educational framework, and co-creating new educational resources to empower those who may be left behind.

Main Researcher: Laura Gemmell
Supervisors: Lucy Wenham, Sabine Hauert


Morphogenesis in robot swarms

Morphogenesis allows millions of cells to self-organize into functional shapes during embryogenesis. This process emerges from local interactions of cells under the control of gene circuits that are identical in every cell, robust to intrinsic noise, and adaptable to changing environments. Constructing human technology with these properties presents an important opportunity in swarm robotic applications ranging from construction to exploration.

Main Researcher: Daniel Carrillo Zapata
Supervisors: James Sharpe, Luca Giuggioli, Alan Winfield, Sabine Hauert
Funding: EPSRC Farscope CDT
More information: Science Robotics paper, IROS/RA-L paper, video, and blog, Frontiers in Robotics and AI paper.

Swarm engineering with embodied reality modelling

Designing controllers for swarm robots to produce a desired collective behaviour is hard, and current off-line methods are brittle in the face of changing environments. By giving each swarm agent sufficient processing power to conduct on-board reality simulations, we hope to be able to continually evolve new controllers in an adaptive and distributed way. To this end, we are building a swarm of Xpucks (eXtended e-pucks) with a collective processing power of 2 Teraflops.

Main Researcher: Simon Jones
Supervisors: Sabine Hauert, Matthew Studley, Alan Winfield
Funding: EPSRC Farscope CDT
More information: DARS paper, Frontiers in robotics and AI paper, Advanced Intelligent Systems paper and video.