Goal and Plan Recognition for in Complex Settings

About the Project

These projects are open to students worldwide, but have no funding attached. Therefore, the successful applicant will be expected to fund tuition fees at the relevant level (home or international) and any applicable additional research costs. Please consider this before applying.

Goal recognition is the task of identifying the intended state of affairs of an agent, given evidence about its currently executing plans. While recent approaches have improved runtime performance for classical planning settings, the expressivity of the domain formalism has remained largely the same, providing ample opportunities to define more realistic and challenging goal recognition settings. Specifically, domains where the recognizer needs to learn target agent behaviour, or needs to recognise the goal of multiple target agents, remain poorly explored in goal recognition research. This project aims to give the first steps towards a number of novel goal recognition techniques in more complex domains, beyond those based on classical planning formalisms. The resulting techniques should prove invaluable for goal recognition in the real world, and where one needs to identify the behaviour of teams of agents.

One potential avenue for goal recognition with multiple agents relies on Linear Programming and other Model Checking approaches, possibly taking into consideration information on landmarks.

Most GR approaches rely on an arduous process to inform the recognizer about the feasibility and likelihood of the different actions that the actor can execute. This process might include manually crafting elaborate domain theories, running a planning algorithm multiple times, performing intricate domain optimizations, or any combination of these tasks. This way of designing and deploying goal recognition approaches is clearly infeasible for real-world applications, especially when the observations made by the recognizer rely primarily on image-based sensing.

To address limitations of current Goal Recognition you will investigate novel goal recognition techniques capable of recognizing goals in real-world scenarios. We aim to replace manually crafted representations and online execution with model-free Reinforcement Learning (RL) techniques that allow most of the computation to be done a-priori. Thus, we must revisit the GR problem definition to accommodate RL-based domains and develop a new framework for GR that relies on policies or utility functions derived from any model-free RL technique.

Candidates can potentially work remotely on the PhD.

Informal enquiries can be made by contacting Prof Meneguzzi (felipe.meneguzzi@abdn.ac.uk)

Decisions will be based on academic merit. The successful applicant should have, or expect to obtain, a UK Honours Degree at 2.1 (or equivalent) in Computing Science.

We expect the candidate to have experience with at least one of the following topics:

• Automated Planning algorithms (e.g., heuristic search) and formalisms (e.g., PDDL)
• Multiagent Systems
• Multiagent Planning
• Goal and Plan Recognition
• Reinforcement Learning

We encourage applications from all backgrounds and communities, and are committed to having a diverse, inclusive team.

Application Procedure:

Formal applications can be completed online: https://www.abdn.ac.uk/pgap/login.php.

You should apply for Degree of Doctor of Philosophy in Computing Science to ensure your application is passed to the correct team for processing.

Please clearly note the name of the lead supervisor and project titleon the application form. If you do not include these details, it may not be considered for the project.

Your application must include: A personal statement, an up-to-date copy of your academic CV, and clear copies of your educational certificates and transcripts.

Please note: you do not need to provide a research proposal with this application.

If you require any additional assistance in submitting your application or have any queries about the application process, please don't hesitate to contact us at researchadmissions@abdn.ac.uk

Funding Notes

This is a self-funding project open to students worldwide. Our typical start dates for this programme are February or October.

Fees for this programme can be found here Finance and Funding | Study Here | The University of Aberdeen