PhD Studentships: Continual Learning in Meta-heuristic Optimisation and Evolutionary Optimisation and Robotics

Edinburgh Napier University is ranked the top modern University in Scotland in the 2022 Times World University Rankings. The School of Computing, Engineering and the Built Environment is highly regarded and has invested recently heavily in research in terms of both staff and facilities to conduct world class research in a wide range of disciplines.  

In the 2021 Research Excellence Framework (REF), our research was ranked top modern university in Scotland in terms of research power. 

As part of our recent significant investments in research, we have recently recruited additional academics with outstanding research capabilities.  

The investment in research is continuing with a large number fully funded 3-year PhD studentships being made available of which this is one. The studentship will cover full UK or international tuition fees and will include a standard living allowance at the RCUK rate (Currently £17,668 pa). 

At this stage, we are recruiting students for following projects:

• Generating diverse and functional robots by jointly optimising their body-plan and controllers
• Adaptive Robot Behaviours in dynamic and outdoor settings
• Continual Learning for Combinatorial Optimisation

A brief description of the projects is shown below. More information on requirements and how to apply is available following the provided links.

The studentship is expected to start in October 2023.  All applications must be received by 14^th April. Those who have not been contacted by 28^th April 2023 should assume that they have been unsuccessful. 

More information about PhD degrees at Napier can be found at https://www.napier.ac.uk/research-and-innovation/research-degrees. 

PROJECT DESCRIPTIONS

Continual Learning in Meta-Heuristic Optimisation

Director of Studies: Professor Emma Hart (e.hart@napier.ac.uk)

Optimisation problems are ubiquitous across many sectors. In a typical scenario, instances arrive in a continual stream and a solution needs to be quickly produced.  Meta-heuristic search techniques have proved useful in providing high-quality solutions, but rather than simply being designed and deployed in one-off process, they should  (a) be capable of continually adapting to changing instances to ensure they deliver the best quality solutions and (2) improve over time, by learning from experience gleaned from solving previous solutions. This project will focus on one or more aspects of creating a continual learning system, for instance developing novel algorithm-selection methods that are capable of selecting the most appropriate method; using  algorithm-generation methods (e.g. genetic programming) to generate or tune algorithms to work well on instances that occur in novel regions of the instance space; developing methods that are capable of learning from experience, for example  using transfer learning or warm-starting methods using knowledge  learned from solving past instances. The project is likely to mix techniques from meta-heuristic optimisation and machine-learning, particularly borrowing ideas from the transfer learning or continual learning literature.

Generating diverse and functional robots by jointly optimising their body-plan and controllers

Director(s) of Studies:  Dr Léni Le Goff (L.LeGoff2@napier.ac.uk)  and Professor Emma Hart (e.hart@napier.ac.uk)

The PhD project will explore optimisation, learning and/or adaptation in the context of evolutionary robotics. Areas of interest include the co-evolution of morphology and control of robots the interaction of evolution and learning mechanisms to produce bodies and behaviours that are specialised to specific environments and tasks. Alternative projects might focus on adaptation of behaviour only, using learning methods (e.g. evolution, reinforcement learning) to adapt controllers in real time to adapt to new environments, or learning repertoires of behaviours to enable robust performance.  Another promising area is in the use of state-of-the art methods from the quality-diversity literature to fully explore rich search spaces of both morphologies and controller.  Projects can be conducted in simulation only but there is also the possibility to utilise our robotics laboratory to conduct experiments on physical robots.  A software and hardware framework to jointly optimise the body and controllers of real robots developed by the ARE project will also be availablefor candidates to expand.

Adaptive Robot Behaviours in dynamic and outdoor settings

Director of Studies: Dr Leni Le Goff (L.LeGoff2@napier.ac.uk)

In controlled settings such as factories, robots are able to achieve many tasks efficiently and accurately. However, it is still a challenge to enable robots to operate in unstructured, dynamic and outdoor environments. In such settings, changes can occur that can render the skills and knowledge of the robot ineffective. Robots must therefore be able to adapt previously learned behaviours to new tasks and settings. The approach proposed to be investigated is in two steps. First, existing resource intensive algorithms will be applied to learn robust behaviours and perceptual representation for the robot to tackle complex tasks and environments. In this first step the environments will be static. Then, light weight algorithms, i.e. with fast convergence, will be explored to adapt quickly these learned behaviours and representations to face dynamic environments. The ultimate goal of this project is to enable robots to achieve complex task in outdoor settings where the conditions can change suddenly or progressively. Using mobile legged robot such as dog or hexapod robot, the Ph.D. work will focus first on testing the viability of the methods in simulation before eventually testing them on a real robotic platform.

For further details and how to apply please visit

https://www.napier.ac.uk/research-and-innovation/research-degrees/how-to-apply

Advert information