Soundbites and images from the third session of the 2024 National Road Safety Conference: the role of technology in casualty reduction.
- The role of technology in casualty reduction (14:30 to 15:30)
- Click here to view the agenda
This page does not automatically refresh, click here to see the latest updates.
14.30
Stewart Fowler, Road Safety Intelligence Team Leader, Kent CC
Stewart Fowler has been working in road safety with Kent County Council for nearly a decade.
As the Road Safety Intelligence and Innovations Team Leader he champions the use of data, analysis, and new technologies in supporting road safety delivery and safe system implementation.
Click here to read the full bio
Presentation: How Kent County Council improved road safety using connected vehicle data
What is CV data and how have we consumed it?
Working with Compass:
- In vehicle Sim Cards (from 2018)
- Telematic devices
- Approx 2% fleet coverage (1-5% road dependant)
- Representative sample from cars, LGVs, PSVs, HGVs
- Processed in the cloud
- Location, timestamp, heading, speed, braking, acceleration and swerving.
- Infer: journey times, origin and destination.
How have we used the data?
- Scheme Identification / Prioritisation
- Monitoring and Evaluation
Case Study: Running Horse Roundabout
Scheme delivered: Turbo Roundabout
Roundabout was redesigned for users to approach in the correct lane, then stay in lane whilst circulating and exit at the necessary exit.
Before and After study to understand changes in;
For impact on collision risk:
- lane discipline
- speeds through the junction
- G-forces through the junction
For performance of the roundabout:
- queue lengths on approaches
- journey times.
Initial findings:
- Fractionally less compliance with permitted routes: 74.67% before, 74.46% after (-0.22%).
- Improvement in road users making correct manoeuvres into the area of greatest concern.
- 5.22% reduction in vehicles crossing centreline from right to left.
- Small reduction in speeds.
- Speeds on the circulatory have reduced by up to 3% (mean speed) – from 38.06kmph (23.65mph) to 36.99kmph (22.99mph)
- Steering G-forces have reduced on the circulatory by up to 8% (mean g-force) – from 0.259g to 0.237g
- Harsh Steering events have reduced 30% (Pre-scheme period August 2023, post-scheme period August 2024)
- Reduction in injury collisions.
- Expected collisions is 3.11
- Actual collisions post scheme is 2
Application and learning:
- Initial scheme findings appear positive
- Second analysis period required to validate initial findings. Scheme needs time to settle.
- Some results are unexpected, further testing of methodology to ensure the data is as accurate as possible.
- Design implications; roundabout arrangement, signing provision or use of kerbs for future schemes.
What have we learned about using the data?
Considerations:
- Data availability and coverage (temporal, penetration into fleet)
- Granularity of the spatial accuracy
- Comparison to ground truth where possible
- Kmph not Mph!
- Understanding the limitations and using data accordingly.
Value:
- Speed data validates well to other sources
- Near real time – up to date and flexible time searches
- G-force data provides greater level of understanding of driver behaviour than previously available.
- Can monitor “after the event” – without preparation for E&M
- Reduction in time and expense of installing infrastructure
- Repeatable data collection – ongoing learning.
- Dashboard access provides simplistic and widespread access
- Pipeline of new data available – (e.g. activation of driver assist)
- Business Case and VfM – multiple use cases across organisation
14.45
Paul Cope, Managing Director, The Creative Lab & Director, Safe Roads Challenge UK
Paul Cope is Managing Director of branding and communications specialists, The Creative Lab.
He has spent his 25-year career making brands vital and creating campaigns that positively influence consumer behaviour.
Click here to read the full bio
Presentation: How tech-forward gamification can dramatically improve the mindset and performance of UK drivers
Despite a whole raft of awareness campaigns and punitive measures targeting poor driving, casualty numbers remain stubbornly high.
This presentation contends that smart tech that leverages behavioural science has never been a more important tool to help lower the amount and severity of road traffic collisions in the UK.
The Safe Roads Challenge aims to inspire a culture of positivity and reward for being a safer, more mindful driver and set new standards in driving across the UK. Key to this ambition is to make the programme fun, accessible and motivating for people to do on every trip, thus creating an enduring safe driving movement. This presentation explains why that’s important and how this is being achieved.
What is gamification?
The integration of game elements, such as points (XP), leaderboards, rewards, and challenges, into non-game environments to boost engagement and improve behaviour
- Telematics-based gamification can help modify drivers’ behaviours by providing ‘real-time’ feedback & rewards, incentivising safer driving practices
- Interventions of this nature have led to reductions in speeding and other risky behaviours by 30-50% (compared to non-gamified approaches)
- Use an interactive app, quizzes & point-based challenges to encourage residents learn traffic safety rules
- Participants who engaged were 40% more likely to retain and apply knowledge
The Safe Roads Challenge – rewarding mindful driving
5 big takeaways
- ‘Gamification’ has the potential to be a genuine game changer for UK read safety
- Rooted in behavioural science, it has proven success in many areas of public health
- The Safe Roads Challenge is rewards-driven and gamified to change driver behaviours for good
- It’s been successful in the North America for the past 2 years… now it’s time for the UK!
- We want to collaborate with UK the road safety community and socially conscious brands
15.00
Dr Christophe Bastien, Associate Professor in Transport Safety, Coventry University
Dr Christophe Bastien leads the Transport Safety and Simulation research group at Coventry University.
He has over 27 years’ industrial and academic experience in the field of vehicle safety, evidenced by 22 patents, 43 journal publications and 284 citations.
Click here to read the full bio
Presentation: The Role of Artificial Intelligence and Computer Modelling in Predicting Pedestrian Collision Parameters and Injuries
Research journey:
2015: UK Police Force challenge to Coventry University research team “Given a post-mortem report and the damaged vehicle, is it possible to calculate the vehicle impact speed?”
2017: Funding provided by The Road Safety Trust for a proof of concept (RoaD project)
2021: The findings from The Road Safety Trust project have been significant and to develop a new soft tissue organ trauma indicator (PMIRSA awarded)
2024: We are here:
- Patented Pedestrian Collision Tool
- SENTINEL Project (£409k) to predict brain injuries by the roadside (RST funded)
Challenges and AI Solutions
- Accurate impact speed key for legal proceeding and insurance claims.
- Impact speed and injuries are linked
- Current method: throw distance.
Current challenges:
- Required expertise,
- Time consuming & costly
- Limitations
- Hit-and-run (little evidence).
Opportunity 1: AI powered impact speed estimation
Create an intelligent tool to improve impact speed estimation by the roadside as well as provide more information.
Opportunity 2: AI Driven triage recommendations
Injury severity and speed are linked, hence if the impact speed is known, maybe triage suggestions can be derived.
Computation of vehicle impact speed prediction by the roadside
Using Vehicle Damage and Anthropometry
- Method has been validated using 3 West Midlands Police pedestrian real-life collisions and with collisions computer model (sedan).
- Requires some interpretation, required expertise (potential bias).
Innovation: PACE-AI
For a given vehicle profile, bumper and windscreen damage, pedestrian height and weight, we can calculate in seconds, by the roadside:
- Vehicle impact speed
- Pedestrian crossing speed
- Pedestrian gait
- Pedestrian crossing direction
PACE-AI: PedestriAn-vehicle Collision ForEnsics tool.
PACE AI can provide the Police Force with:
- Vital clues by the roadside to support live investigation,
- Support triage of forensic response.
- Complement standard investigation and provide suggestions
- Patented and available now.
- German Police purchased PACE-AI for 31 Police departments.
Brain injury severity prediction by the roadside (SENTINEL)
- “In-Situ Mobile Application for the Triage of Pedestrians in Vehicle Collision” (SENTINEL).
- Funder: Road Safety Trust (Large Grant).
- £409,000 (Start November 2024). 3-year project.
- Core Team: CU, UHCW, TAAS, and WMAS.
- www.sentinel-rtc.coventry.ac.uk
Project Summary:
- Create of a generic AI tool to compute white and grey matter injury severities
in pedestrian collisions - Create a West Midlands pedestrian collision database
- Validate AI tool against this new database
- Embed SENTINEL with PACE-AI.
15.15
Thierry Castermans, RoadTrace global lead
Thierry holds a PhD degree in Physics and has developed extensive experience in fundamental and applied research, in signal processing and data analysis.
Seven years ago, he joined AISIN, a tier-1 supplier in the automotive industry, and got involved as a data scientist in the development of AI projects with positive impact on society.
Click here to read the full bio
Presentation: Harsh braking KSI predictions
How can we use those technologies and data from connected vehicles to help road operators?
Connected vehicles:
- Any vehicle able to transmit data
- Most new vehicles since 2015
- Mandatory for all new cars since 2018
What kind of data?
- Accelerometers are required for airbags and ABS/ESC control deployed widely since the 90s
- Floating Car Data (FCDs) are designed for real-time traffic monitoring and congestion analysis
- Vehicle sensor data allow accurate detection of harsh braking (very high frequency signal).
Studies show a strong correlation between harsh braking events and real crashes
But is harsh braking better than looking at historical KSIs?
Our study in the UK:
- Comparing our unique approach, based on harsh braking clusters, to the usual historical collisions data approach (STATS19, UK)
- Validated in different regions (rural, urban, various traffic densities)
- Selected by jury of experts to be presented at ITS World Congress
3 months of harsh braking data to predict where crashes will occur; checked against real collision reports from the following 9 months.
3 months harsh braking data to identify clusters:
- 100 000 km road network
- 6 million vehicle traces
- 134 million hours of driving
- 2 million harsh braking events
Real collisions in following 9 months
- 38 738 crashes reported
Results:
Case studies in UK (Cambridgeshire, Kent, London, East & South East UK), using 3 months of harsh braking data to predict collisions:
On average, 21% of « RoadTrace » harsh braking clusters turned into real collisions in the following 9 months
Using 5 years of KSI data would have given maximum 6% conversion rate
Our technology is, on average, 3.23 times more efficient at predicting the exact location of future collisions compared to historical data
Comment on this story