Not being able to quickly predict collision risk for upcoming weather conditions, let alone traffic and population changes is leading to deaths and injuries which could be avoided. Auxi Brain has solved this challenge.

ARCS leverages cutting-edge AI technology to accurately model the risk of traffic collisions across the entire country, covering over 50,000 combinations of road layouts, weather conditions, population densities, and traffic volumes amongst a host of other data.

By ingesting vast amounts of data including STATS19, weather forecasts, population demographics, light levels and road network details, ARCS produces highly precise predictions of collision risks for any specific location in GB.

What makes ARCS truly exceptional is its deep-learning neural network, enabling it to continually learn and improve its predictions. Plus, with 130 million predictions and 98 billion recalculations overnight using hour-by-hour weather forecasts, ARCS provides real-time, up-to-date risk assessments.

ARCS isn't just about predicting risks; it's about taking proactive steps to shape the future for the better.

• 10-Day Collision Risk: Visualize live risks with unprecedented accuracy, helping authorities make informed decisions from upcoming weather, traffic, and network details.
• Optimal proactive interventions: ARCS identifies the most critical areas for speed reduction / enforcement, street lighting maintenance, road defect repairs and gritting.
• School Zone Risk Assessment: Get insights into collision risks outside schools at peak times, empowering local authorities to implement targeted safety measures.
• Roadworks Impact Analysis: Stay informed about the impact of roadworks on collision risks.
• Population and Traffic Volume Predictions: Understand how changes in population and traffic volumes will affect collision risks, allowing for proactive measures.

• the system is designed to help you manage road safety, and not replace human expertise
• we won't currently allow it to be incorporated into GPS systems to route vehicles directly
• on request we can extend weather risk past the 10 day current extent
• we're working on linking to live and scheduled roadworks and incident data
• the prediction points on the layers are the locations of collisions from 2017 to 2019. Whilst any point can be modelled, we've deliberately limited the pre-calculated points to these
• the model is trained on 11 years of data, and re-trained yearly