10104210Smart Cities, Safe Crossings: AI-Enhanced Solutions for Data-Driven Pedestrian OptimisationClosedCollaborative R&DInnovate UKThe UK's commitment to Net Zero, sustainability and improving the nation's health is leading to a shift towards Smart Cities models which take advantage of new and emerging technologies. These priorities have highlighted the competing agendas of active travel and motorised traffic in the context of increasing driver numbers ([GB Driver Licence Numbers, DfT][0]). The interface between vehicles and pedestrians has never been more complicated, with crossing points a particular hotspot for accidents. In 2022, 29,804 people were killed/seriously injured on UK roads, of which 20% were pedestrians (UKGov/Admiral). Road-crossing sensors already play a key role in supporting the transition to Smart Cities, but they currently focus on traffic management with insufficient optimisation for other road and pavement users. There is a pressing need for tools to provide rich active travel data to Local Transport Authorities (LTAs) which would enable them to take account of pedestrian needs, provide proactive alerting to enable rapid traffic management interventions, and for the means to optimise pedestrian flows and minimise accidents without adversely affecting traffic flows. Starling Technologies is deploying, in the UK and internationally, its unique Starling Detector technology which has an exceptionally wide field of view and outstanding fidelity allowing us to 'see' significantly more than industry standard sensors. This technology enables a focus on pedestrian movement data including behavioural metrics, and pedestrian optimisation alongside traffic management. This redefines the current paradigm in which vehicles are always prioritised, pedestrians are incited to take risks and active travel is disincentivised. Starling Technologies' existing pedestrian optimisation uses algorithms to produce a measurable reduction in accumulated pedestrian waiting times without a commensurate delay to vehicular traffic. Building on this prior work, we will employ Large Language Models (LLMs), AI, and ML to unlock the potential of pedestrian data. By harnessing publicly available datasets, including weather and public transport information, we will transform our understanding of pedestrian behaviour and deliver significantly improved pedestrian optimisation. This project will provide LTAs with the data and tools to redress the balance of priorities in favour of pedestrians in accordance with local policies thus enhancing efficiency, productivity and safety. This project directly contributes to the nation's commitment to Net Zero, sustainability, and public health. [0]: https://www.data.gov.uk/dataset/d0be1ed2-9907-4ec4-b552-c048f6aec16a/gb-driving-licence-data