New company Predina is focusing on dynamically predicting and reducing the contextual/spatial risk of crashes using AI for the automotive tech and insurance telematics sector.
Predina has built a cloud-based spatial risk platform (STARI: Spatio-Temporal Accident Risk Intelligence) that learns from historical crashes and contextual/spatial variables to predict “where” and “when” a crash is likely to happen – dynamically and in real time.
STARI has been validated through a project with UK’s Digital Catapult and an enterprise fleet. In early results, the project demonstrated over 80% accuracy of crash predictions, reduced crashes by over 25% and increased awareness of risk among drivers by 100%. The company was also one of very few deep tech companies supported by Google through its Machine Learning for Good programme.
STARI provides dynamic hotspots, the real time and historic risk of crashes for a specific route, the safest route from A to B and driver risk alerts/warnings.
The cloud-based platform works dynamically, without hardware, installations or cameras, but rather it learns from a history of crashes and the external conditions (weather, traffic, time, visibility) to make a future prediction on “where” (spatio) and “when” (temporal) it is likely to happen again.
Longer term, the platform can be integrated with the decision-making engine within the vehicle to dynamically alert the driver to a high risk area and automatically reduce its speed in these areas in real time. Equally, the solution can be enhanced and customised through a history of autonomous vehicle incidents to provide a spatial understanding of risk on the roads and identify the lowest risk route.
Predina has developed the platform to easily communicate the information in a user-friendly and accessible way via audio, satnav, mobile applications, web and pdf.