PhD position on epistemic planning in Copenhagen, Denmark (DTU), deadline 1 February 2025

[Thomas Bolander]

=== PhD scholarship in Logic and AI - DTU Compute, Copenhagen, Denmark ===

DTU (Technical University of Denmark) is offering a PhD scholarship in logic and AI within the research project Attention in Epistemic Planning. The project members are Gaia Belardinelli (Stanford University), Thomas Bolander (PI, DTU), Hans van Ditmarsch (IRIT Toulouse), Jasmin Grosinger (Örebro University), Jens Ulrik Hansen (Roskilde University), Nir Lipovetzky (University of Melbourne), Rineke Verbrugge (University of Groningen) and Sebastian Watzl (University of Oslo). Your principal supervisor will by Thomas Bolander.  

Short project description: The project takes departure in epistemic planning based on dynamic epistemic logic (DEL), a line of research within symbolic AI initiated by the PI in 2011. It allows agents, e.g. robots, to reason about other agents, e.g. humans, as part of their planning process. The crucial novelty of the proposed project is to develop and integrate logical models of attention with the aim of 1) achieving better computational complexity and practical efficiency; 2) allow robots to reason about the limited attention of humans. The project will take inspiration from theories of human attention from philosophy and cognitive science and build models of dynamic epistemic logic that represent crucial aspects of attention, including its boundedness, its selectiveness, and the distinction between top-down and bottom-up attention. The main formalisms and algorithms developed will be based on dynamic epistemic logic that will both be investigated theoretically and implemented in a robotic setup. 

*Responsibilities and qualifications*

Your job will be to:

* assist in developing formal logical models of attention, using mainly dynamic epistemic logic (DEL);

* assist in developing theoretical frameworks, algorithms and implementations of attention-based epistemic planning, and investigate the (theoretical and practical) efficiency of the developed algorithms;

* be main responsible for implementing attention-based epistemic planning in multi-agent simulations and humanoid robots, and evaluate the dynamics and quality of agent interactions (including human-robot interaction).

Successfully completing these tasks requires skills in many areas of computer science (broadly conceived), including programming, algorithms, discrete mathematics, automated planning (symbolic AI), logic, machine learning and robotics. The ideal candidate has expertise in all of these. As the project has an interdisciplinary flavour – involving project members with expertise in philosophy, cognitive science, computer science, AI (symbolic AI and machine learning), logic, and robotics – having an interdisciplinary outlook will be an advantage.  

For more information and for applying (deadline 1 February 2025), please go to this page: https://efzu.fa.em2.oraclecloud.com/hcmUI/CandidateExperience/en/sites/CX_1/job/4447/?utm_medium=jobshare