How See.Sense Data Insights are Helping TfL Tackle Vision Zero

See.Sense was selected by TfL to participate in a ‘Vision Zero Proof of Concept (POC)’ which ran in 2022. This POC sought innovative solutions that can mitigate against incidents involving ‘vulnerable road users’ (pedestrians, cyclists, motorcyclists and more recently micro-mobility users). See.Sense supplied innovative sensor data that enabled TfL to better understand the risk profile for cyclists in the London area.

This project went on to be recognised with a Prince Michael International Road Safety Award for its innovative use of data. 

TfL and Vision Zero 

The Mayor’s Transport Strategy (2018) sets out a Vision Zero approach to road danger with the clear aim of eliminating all deaths and serious injuries on London’s transport system. 

“The Mayor’s aim is for no one to be killed in or by a London bus by 2030, and for all deaths and serious injuries from road collisions to be eliminated from London’s streets by 2041.” 

To help achieve Vision Zero, TfL are exploring how they can mitigate against incidents involving ‘vulnerable road users’ who currently make up 82% of all people killed or seriously injured on London’s roads. 

Problem statement for the Vision Zero POC 

To understand more about these events, TfL sought to explore opportunities (for example, using the Road Safety Risk Model) to understand more about the conditions that lead to near misses and potential safety concerns on the road network. 

It is known that there is under-reporting across injury collisions / incidents so it is not guaranteed to have a record in STATS19 even if something has happened in a location (estimates are around 66-70% of all cycling injuries are reported).

TfL had previously investigated the use of OEM (Original Equipment Manufacturer) car telematics data to enable them to model high risk locations on the network for near misses and potential safety concerns. This trial used data from a single vehicle manufacturer.  

To expand on this work, the Vision Zero Proof of Concept (PoC) sought to identify, capture and bring in additional data in order to explore insights on and model near misses on our network. The data needed to be London-wide data and provide a representative view of Londoners.  

What is the ‘Vehicle as a Sensor’ concept and how could it help Vision Zero?

TfL had previously investigated the use of OEM car telematics. This type of data includes insights into swerving and braking from the car data to help identify and model high risk locations on the network, for near misses and potential safety concerns. 

There was a gap in the data available to TfL in accessing this type of data from micromobility such as cycles and e-scooters.  With Vision Zero and the specific targets on reducing incidents involving vulnerable road users, there was a need to explore this data. 

See.Sense previously worked with RoSPA on a DfT-funded project that explored our cycling data to help understand more about cycling collisions. The research found a correlation between Stats19 collision events and See.Sense data and the report recommended that our data could be used to identify hazardous locations for cyclists.  

In this PoC, we wanted to replicate this work and explore use case with TfL who already had a risk model in place 

How See.Sense technology works 

See.Sense devices contain patented technology that monitors the rider's environment over 800 times a second, using AI and edge processing to gather highly granular, spatially located sensor data insights.

These include swerving, braking and road surface condition, collisions, speed and dwell time. When we see an unusual brake or swerve jerk event we will catalog the location and time. When they occur in the same area at the same time of day and/or day of week, we will aggregate them to give an indication of the frequency of these events at that location.

We also provided perception reports made by cyclists in our app, with reports categorised as close pass, collisons, potholes, obstructions, and other - along with up to 250 text field per report.  Data is collected in full GDPR compliance and aggregated and de-peronalised. 

See.Sense Data Insights in London gathered by a group of city commuters

How we worked with TfL

See.Sense provided aggregated and anonymised crowdsourced sensor data insights gathered from pedal cyclists who use our See.Sense sensor-enabled, connected bike lights in London.  

The data captured provides insights that are geospatially located and includes speed and dwell time as well as braking, swerving and road surface condition. This large dataset represented over 5.4 billion sensor readings gathered between 2019 - 2022.

Results 

TfL found that See.Sense data was useful to help inform holistic risk – the underlying risk across all cyclists using See.Sense devices on the London road network.

The PoC established that the See.Sense data would also be useful for risk modelling and for creating a ‘big picture’ view of road safety in London when used with similar data from other modes. It gives an idea of the volume of injuries TfL could expect if nothing changes, and could help identify trends across locations with similar characteristics (road layouts, distinct cyclists, speed limits etc) to determine which safety interventions work.

The data also helped to identify locations where cyclists face riskier journeys / conditions when travelling across London. This data would be useful for directing site-visits to understand why events are occurring and make an assessment of what could be done to make cycling journeys safer at those locations.

“Overall, See.Sense data was able to provide standardised sensor data insights on cycles across London.  These new insights will enhance the effectiveness of our risk model and therefore contribute to our understanding of the risk of collision, helping London to achieve its Vision Zero ambition” - Alex Drake, Data Scientist, TfL 

Our work was also featured in Transport For London's Open Innovation page. TFL stated that "To better understand conditions leading to near-misses and incidents with cyclists, we worked with See.Sense who provided over 5.4 billion crowdsourced sensor readings from their sensor-enabled connected bike lights. Insights on road condition, speed, swerving and braking allowed us to identify where cyclists face riskier journeys in London, helping us to plan improvements to our network and achieve London's Vision Zero ambition."

Where could this go in the future?

At the moment, TfL are considering their next steps in regards to making use of vehicle-as-a-sensor data for Vision Zero.  From a technology point of view, See.Sense solution is highly scalable on all micro-mobility modes, because in addition to collecting data via the bike lights on privately owned bikes, the tech can also be seamlessly integrated into bikes, e-scooters, e-cargo bikes using our new ‘SUMMIT’ technology.  See.Sense SUMMIT is a low power device which means it can run from the charge received from the bike’s existing dynamo  - making it compatible for use on dynamo or e-bikes as well as fleet e-bikes.  Our tech could even be scaled to other two wheelers, such as motorbikes.

E-scooter data in London with DOTT

After demonstration of our solution on a number of Dott e-scooters in Paris, our solution was procured by Dott for the e-scooter trial in London as PoC. This involved the design and installation of a bespoke device that fits and draws its power supply from the e-scooter, and is capable of sending the See.Sense data to our data lake.  This gives a standardised way for measuring risk, enabling integration into a wider risk model.

'LET'S TALK CITIES' NEWSLETTER