Safe and Socially Aware UAV Navigation in Co-Located Human Spaces

Unmanned Aerial Vehicles (UAVs) are increasingly being deployed in environments where people live, work and move, yet most navigation algorithms still treat humans as generic obstacles rather than social entities. This PhD project will investigate how UAVs can move safely and socially appropriately in co-located human spaces such as care facilities, campuses, warehouses, or public buildings. The core challenge is to design navigation and control strategies that respect human comfort and safety constraints (personal space, visibility, predictability) while remaining robust to uncertainty in human motion and perception, and still achieving mission goals such as inspection, delivery, or monitoring. Reinforcement learning and other data-driven approaches will be explored to enable UAVs to adapt their behaviour online to different users, layouts and crowd conditions.  

The project will combine methods from aerial robotics, motion planning, safety-constrained control and human-robot interaction. Depending on your interests, you may work on topics such as learning social navigation policies from human trajectory data, integrating safety guarantees (e.g. shields, formal constraints, or verified controllers) into learned policies, fusing onboard sensing with external localisation to track humans, or developing metrics and user studies to evaluate perceived safety and social acceptability of UAV behaviour. There will be opportunities to implement and test algorithms both in high-fidelity simulation and on real UAV platforms in controlled indoor environments, as well as to collaborate with domain experts (e.g. in safety, human factors, or application domains like healthcare). The ultimate goal is to develop UAV navigation strategies that people find safe, understandable and acceptable, enabling UAVs to become trusted collaborators in shared spaces.