COUNTER: Computing train Occupants Using Novel sensing Techniques to Enhance Rail services.

**CONTEXT**: Understanding realtime passenger rail service use is crucial, because overcrowded rail services (annually published as Passengers in excess of capacity, with maximum allowable PiXC of 4.5%) lead to more delays, increased risk of accidents, reduced passenger comfort (highest load measured was 184%, DFT 2012) and decreased passenger satisfaction (National Rail Passenger Survey (NPRS); Transport Focus, 2014) requiring train operators to invest in remedial measures, whilst underutilised services are not cost effective. Knowledge of passenger loading diversity by rail operators is not comprehensive, so services are provided on data collected: at route or network-level not station-level (insufficiently disaggregated); in discrete time episodes rather than continuously (incomplete); based on studies undertaken several years previously (out of date). Historically, the difficulties and expense of collecting and analysing high volumes of good quality data made such management decisions understandable. Recent developments in sensor and information communication technologies mean they are not justifiable today.

**NEED**: COUNTER has implications for all types of routine, incidental and disrupted rail travel: crowding penalties (MOIRA2; PiXC), delays (Public Performance Measure), optimising boarding/alighting times, journey suppression, insufficient capacity, customer satisfaction, on-board experience, revenues, congestion, timetabling, train/station design, operational cost reduction, health and safety, pollution, energy saving and facilities management. Accurate intelligence about passenger flows/demand is critical for Intelligent Trains. Despite being the primary challenge facing all train operators obtaining a precise and reliable measurement of real time passenger demand remains difficult based on current techniques, e.g. video imaging/weight sensors. Impact: train demand inefficiently managed.

**INNOVATION**: COUNTER offers a platform interoperable with existing Train Management Systems, combining a patent-pending sensor fusion method using wifi fingerprinting and lo-fi infrared sensors with machine learning algorithms to deliver an accurate assessment of real time and predictive passenger counting/flow.

**OUTCOME**: COUNTER will generate predicted revenue of £118M within 5 years of launch, along with new jobs and generation of new knowledge with wider applications including all mass transit systems and building management. UK-wide, it will cut delays due to boarding and alighting by 10% (worth ~£5.5m per annum); reduce station and platform accidents by 10% -- i.e. ~ 700 passenger injuries a year; cut rail customer dissatisfaction levels by 10%; attract 0.8% more passengers to UK rail network from higher service quality -- i.e. £75m per revenue per annum; increase revenue from station/platform and train advertising by better qualifying eyeballs/impressions and dynamically trade advertising space (worth ~£160m per annum). **Keywords**: _Sensors, Footfall, Network Capacity_.