Finding a CRItical SPeed function ahead of a road section for vehicles in motion

Heading: Making Roads Smarter and Safer with AI Every year, icy roads, sharp turns, and sudden weather changes cause countless accidents. What if vehicles could predict dangerous road conditions before they happen? That’s exactly what the CriSp project aims to achieve. CriSp (Finding a CRItical SPeed function ahead of a road section for vehicles in motion) is developing smart technologies to help vehicles sense road conditions, estimate safe driving speeds, and communicate warnings in real-time. This could make driving much safer, especially on curved, mountainous, and accident-prone roads. The project combines sensor technology, AI-driven data analysis, and wireless communication to help cars make better decisions. Sensors in vehicles detect changes in road grip due to rain, snow, or ice. Advanced AI models then analyse this data to estimate how slippery the road is. The vehicles don’t just gather data for them selves, but share it with other vehicles and road infrastructure using a wireless network called Vehicle-to-Everything (V2X). This means if a car ahead detects ice conditions, it can warn other cars behind it in real-time. AI-models predict the safest speed for any given road condition. This helps drivers (and in the future, self-driving cars) adjust their speed before reaching a dangerous section. CriSp tested its technology through simulations and real-world data collection, including field tests in winter conditions in Sweden. Many current safety systems only react after a vehicle loses grip. CriSp proactively prevents accidents by helping vehicles make better driving decisions before they happen. This innovation could lead to: Fewer accidents on dangerous roads; Better support for autonomous vehicles; and Safer driving in unpredictable weather.

Prosjektet har økt kunnskapen om hvordan maskinlæring og sensorbasert teknologi kan brukes til å forbedre trafikksikkerheten. Det har gitt kompetanse om samarbeid mellom kjøretøy og vegkantutstyr (RSU) gjennom V2X-kommunikasjon, og om hvordan slike data kan benyttes til sanntids friksjonsestimering. Et sentralt resultat er utviklingen av en metode for å estimere friksjon på vegoverflater basert på eksisterende kjøretøysensorer, med høyere nøyaktighet enn kommersielt tilgjengelige løsninger. Dette kan på sikt brukes av vegmyndigheter for å varsle om glatte partier og planlegge vintervedlikehold mer effektivt. Prosjektet har også bidratt til faglig utvikling gjennom en offentlig rapport om bruk av maskinlæring i kjøretøysikkerhet, utarbeidet av TØI, som gir både teknisk og samfunnsmessig innsikt i muligheter og utfordringer med slike systemer. Prosjektet har lagt til rette for tverrfaglig samarbeid mellom industri og forskning, og skapt ny innsikt som kan komme både trafikkforvaltning, kjøretøyprodusenter og fremtidig regulering til gode. På lengre sikt kan resultatene bidra til tryggere og mer intelligente transportsystemer.

Our overall approach is based on Leveson's STAMP methodology for safety system design, and hence our solution continuously monitors and responds to safety margins. The CriSp solution shall be implemented on a road section by: (i) installing a Road side unit (RSU) with computing resources for calculating the critical speed, (ii) reconfiguring and augmenting vehicular electronics for sensing, motion assessment and cruise control, and (iii) enabling wireless communications between vehicles and the RSU. We structure the Critical speed function with the following actions: (A1) The RSU detects approaching vehicles and sends them its current speed advisory, (A2) if the Autonomous speed control function is enabled on a vehicle, then this function regulates the speed, to obey the RSU's advisory, (A3) every vehicle transmits to the RSU a Motion assessment summary (MAS) as it leaves the section; the MAS records the speed, as well as signals that convey the extent of skidding and rollover, or the margins towards skidding and rollover, and (A4) the RSU updates its critical speed by feeding the received sequence of MASes into a data analytics engine. To implement the V2I and I2V communications of actions A1, A4 we adopt existing standards for short range communications (ITS-G5, WiFi), and Cellular communications if available to the RSU. To implement the sensing, motion assessment and cruise control parts of actions A2, A3 we use existing inertial sensors and ACC technology on board vehicles, and a promising new sensor for measuring tyre-road friction. To develop the CriSp solution we need to develop and test the whole solution via its parts: Sensing, Wireless communication, Data analytics, and Control components. Both the project's internal structure and the team structure match these components. Moreover, the project team brings expertise and research resources in Sensing technology, Vehicle dynamics, Vechicle testing, Transport safety science, ICT and Mechatronics.