Adaptive AI for Autonomous Drone Systems

About the project

About the Project Autonomous drone systems are expected to play an increasingly important role in applications such as environmental monitoring, infrastructure inspection, disaster response, precision agriculture, and search and rescue. As these systems become more capable and are deployed in increasingly complex environments, they must operate safely, efficiently, and collaboratively while adapting to uncertain, dynamic, and partially observable conditions. Developing intelligent decision-making algorithms that enable autonomous drones to perceive, reason, learn, and cooperate remains one of the key challenges in modern artificial intelligence and robotics. This PhD project will investigate adaptive AI techniques for autonomous drone systems, with a particular focus on multi-agent coordination, sequential decision-making, and learning under uncertainty. The research will explore a range of modern AI approaches, including reinforcement learning, transformer-based architectures, world models, diffusion policies, robotics foundation models, and vision-language-action models. Depending on the direction of the project, one or more of these methods will be developed individually or combined to enable more robust, scalable, and adaptable autonomous behaviour. A central aim of the research is to develop intelligent algorithms capable of coordinating multiple drones operating in cooperative, competitive, or mixed environments. Research topics include multi-agent learning, adaptive planning, robust decision-making, efficient coordination and communication, hierarchical control, sample-efficient learning, safe exploration, and learning from limited or imperfect observations. The project may also investigate human-in-the-loop autonomy, multimodal perception, or hybrid approaches that integrate model-based reasoning with data-driven learning. The research will primarily be conducted using high-fidelity simulation environments, enabling rigorous development and evaluation of novel AI algorithms under realistic operating conditions. Where appropriate, the project may also explore sim-to-real transfer techniques to improve the deployment of learned policies on physical robotic platforms and reduce the gap between simulation and real-world performance. The successful candidate will have the opportunity to contribute to the next generation of intelligent autonomous systems by developing next-generation AI methods for autonomous perception, planning, coordination, and decision-making. The project offers considerable flexibility, allowing the research direction to be tailored to the candidate’s interests while addressing important challenges at the intersection of artificial intelligence, robotics, machine learning, and autonomous systems.

Research areas