Given a knowledge base describing the existing background constraints and assumptions about what is possible in the world as well as the prior experience of an autonomous agent on the one hand and probabilistic perception of the current state of the world of the autonomous agent, on the other hand, it is essential to devise and efficiently enumerate the most consistent world models that are likely to be valid under the prior knowledge in order to refine the agent’s up-to-date perception and take the most suitable actions. This project will exploit modern answer set programming (ASP), constrained ASP (CASP), core-guided maximum satisfiability (MaxSAT) reasoning, bounded model checking (BMC) and relate with Markov Logic Networks (MLN) in order to facilitate interaction between symbolic knowledge and a neural agent and succeed in overall neuro-symbolic tasks, i.e. those posed by Animal AI-Testbed. Students in this project will collaborate closely with those involved in the project on “Building consistent world states for an autonomous agent”.
The aim of this project is to develop techniques for reasoning about consistent world models in order to facilitate interaction between symbolic knowledge and a neural agent and succeed in overall neuro-symbolic tasks, i.e. those posed by Animal AI-Testbed.
Students applying for this project should have excellent programming skills (as evidenced, for example, by high marks in FIT2004). In addition, it would be desirable for them to have done the "Programing languages and paradigms" unit (FIT2102) or, if a masters student, the "Modelling discrete optimisation" and "Solving discrete optimisation problems" units (FIT5216 and FIT5220).