Project Summary

Railway stations struggle to manage disruptions, leading to overcrowding and inefficient recovery. Traditional methods fall short, necessitating real-time, data-driven solutions. OpenSpace proposes a proof-of-concept AI decision-support tool, ReCOVER, for real-time station disruption recovery, aligning with DfT’s strategic priorities and Network Rail’s needs. With a dataset of over 100 million data points from Euston station, the tool will develop algorithms for bespoke recovery strategies. The AI model will be adaptable, ethical, and transparent. Expected outputs include the proof-of-concept AI model and ReCOVER software, tested on one use case in a simulated environment.

Project Achievements

Data collection: A dataset comprising over a million data points from Euston station, focusing on passenger flow between the concourse and platforms, as well as timetables for arriving and departing train services. Simulation Model development: A data-driven pedestrian simulation model developed to replicate peak evening operating conditions. Optimisation: Deploying AI techniques such as Genetic Algorithms and combinatorial optimisation to identify optimal boarding plans – including, the timing of boarding announcements – to minimise conflicting flows at the ramps. Evaluation: The optimal solution resulted in a 65% reduction of passenger flow conflicts, while ensuring a minimum 10-minute boarding window for all departing services.

Conclusions

The ReCOVER project successfully combined a Digital Twin, people flow data, and AI to improve train station safety and operations. Open Space developed a first-of- a-kind AI-driven decision support tool that can reduce conflicting flows by two thirds in crowded areas, enhancing efficiency and safety during disruptions. The project outcomes validate the approach and tool’s effectiveness while highlighting its potential for broader use across mobility hubs, offering significant economic and social benefits. The project also provided key insights into integrating real-time data with scalable AI solutions for station management, despite challenges in merging simulation and optimization techniques. ReCOVER is now ready for live testing and further development.

Next Steps

Tool refinement & live demonstration: This includes integration with more real-time data sources and ensuring regulatory compliance in safety, security, and operational standards to effectively improve decision-making. Key milestones include pilot deployment at Euston Station to validate scalability and effectiveness across various operational settings, ensuring that ReCOVER meets the diverse needs of station management teams. Building strong partnerships & funding with railway operators and stakeholders is crucial for broader adoption, and the team is actively pursuing these collaborations. To support further development, Open Space is seeking additional funding and has already secured commitments from key partners. Commercialisation and wider adoption: The tool’s proven ability to optimize station operations, improve safety, and enhance management systems positions it for broader implementation. Through expanded pilot trials, securing funding, and deepening partnerships, ReCOVER aims to become a core component of modern station management.