Background
The next generation of smart cities will be powered by fleets of autonomous AI agents that continuously monitor, analyse and respond to events occurring throughout the urban environment. These agents may support traffic optimisation, emergency response, public transport coordination, environmental monitoring, digital citizen services and public safety applications.
For these systems to operate effectively, AI agents must execute close to where data is generated. However, cities contain highly heterogeneous computing resources distributed across edge servers, cloud platforms and increasingly mobile infrastructure. Determining where AI agents should execute, when they should migrate, and how they should coordinate with one another represents a major research challenge.
Research Challenge
Future smart cities may rely on thousands of autonomous AI agents that continuously analyse data, make decisions and coordinate actions across transportation, healthcare, environmental monitoring and public services. These agents must operate close to where data is generated to provide low-latency, reliable and context-aware responses.
This project investigates how AI agents can be dynamically deployed across distributed edge computing infrastructures. Key challenges include determining where agents should execute, when they should migrate between edge nodes, how multiple agents should coordinate their actions, and how limited computing resources can be shared efficiently among competing services.
The project will explore intelligent deployment strategies based on techniques such as Multi-Agent Reinforcement Learning (MARL), enabling edge nodes to make autonomous placement decisions while adapting to changing workloads, user mobility and infrastructure conditions.
Example application domains include AI-powered traffic management, emergency response systems, autonomous transport networks, environmental monitoring and digital citizen services. The work will involve developing and evaluating novel deployment algorithms through simulation and experimentation.
Topics
Autonomous AI agents; Multi-Agent Reinforcement Learning (MARL); Edge computing; Smart city infrastructures; Dynamic service placement; Distributed decision-making
Impact
As society increasingly depends on AI systems to support transportation, healthcare and public services, ensuring that these systems remain responsive, trustworthy and scalable becomes critically important. This project contributes to the foundations of future AI-enabled cities.