The increasing prevalence of autonomous systems in dynamic, human-centred environments, such as smart transportation networks and distributed IoT infrastructures, demands decision-making frameworks that can anticipate and adapt to the behaviour of humans and other autonomous agents. This PhD project aims to investigate the integration of Theory of Mind (ToM) reasoning into autonomous agents to improve both individual decision-making and multi-agent cooperation.
The student will investigate two complementary directions:
ToM-Enhanced Decision-Making for Autonomous Agents:
Develop decision-making algorithms that combine Reinforcement Learning techniques like Partially Observable Markov Decision Processes (POMDPs) with cognitive inference modules capable of modelling human beliefs, intentions, and goals.
Enable agents to simulate and adapt to human responses in real-time collaborative scenarios, enhancing performance in environments like smart transport systems, dynamic crowd management and other IoT-enabled infrastructures.
ToM-Enabled Multi-Agent Cooperation:
Equip autonomous agents to model the mental states of other agents, allowing prediction of behaviours, conflict resolution, and coordinated joint decision-making.
Evaluate performance in scenarios such as shared resource management, traffic coordination, and distributed IoT control, benchmarking ToM-enabled agents against traditional multi-agent systems.
Methodology:
Integrate ToM models with Reinforcement Learning based frameworks for single-agent and multi-agent decision-making.
Develop simulation environments capturing realistic human-agent and agent-agent interactions.
Evaluate performance using metrics such as task success rates, coordination efficiency, and conflict resolution effectiveness.
Expected Contributions:
A novel framework for ToM-enhanced autonomous decision-making.
Mechanisms for ToM-based multi-agent coordination and conflict resolution.
Benchmarked evaluation of ToM-enabled agents in simulated human-agent and multi-agent scenarios.
Opportunities for the Student:
Gain expertise in AI planning, cognitive modelling, and multi-agent systems.
Work on a real-world case study provided by the industrial partner British Telecom, bridging theory and practice.
Contribute to cutting-edge research in the domain of autonomous systems.
Please apply via the ‘Apply’ button above.