New research reveals that certain road characteristics can predict the location of an accident. Problems such as sudden changes in limits and incomplete horizontal signaling between lanes are among the most significant factors that can predict traffic accidents. The study used a rich database from the Greek highways. Machine learning methods were used to assess the dangerous points on the roads.
Published in the journal Transportation Research Record, the study is the result of a collaboration between civil and environmental engineers from the University of Massachusetts Amherst and engineers from Egnatia Odos, a public utility in Greece.
The most influential features include road design issues. Some of the observed road characteristics are sudden changes in speed limits, problems with the guardrail, damage to the roadway. Incomplete markings on the road and poor signaling can often be a problem.
To identify these features, the researchers used a data set covering 9,300 miles of roads at 7,000 locations in Greece.
AI application
The scientists concluded that with this knowledge they could train AI models. Models could identify these features based on images and then predict accident risk as a first step towards an automated monitoring system. This knowledge base could also provide recommendations on what needs to be fixed and changed along the way.
How does street appearance affect safety?
According to Ohio State University research, city streets that look like freeways to drivers have a higher chance of serious accidents. You can read more about this topic at the following link.
