Smart transportation requires Smart solutions
For road-safety data analysis, road traffic mortality rate and injury data with GIS is the main data source. But do we recognize that one data point is a human life. Right now, by ignoring usage of the latest traffic analytics tools, we might be doing the costliest data collection ever.
Of course accident/injury data may be used but can we use analytic techniques which give similar information?
This post is about some novel and some established road safety metrics. We see the following as indicators of unsafe road:
1. Top Speed - Now if it takes 1 hour for a vehicle to commute 10 km, that doesn’t mean top speeds are low and the roads are safe. In fact, it’s the other way round. In our analyses, top speeds are higher when a traffic jam just clears or after a long wait at a traffic signal. While vehicles may accelerate from 30 km/h to only 50 km/h, chances of severe accidents increases from 15% to a whopping 80%.
2. Conflicts - An imminent collision unless one person or both abruptly change their movement. This is prevalent on roads of emerging countries where all vehicles, pedestrians bargain for road space and rights. Conflicts are the most available and related metric which can be used. Detection of illegal traffic behavior can also be used.
3. Time To Collision - How long was the emergency manoeuvre done by one person? This metric comes with the conflict itself and is used to filter out 'understanding' between road users and 'conflict' between them. Less-time to-collision means less safe.
4. Shared space percent - This applies when road spaces are not meant to be shared between pedestrians, cars and other road users. For eg. cars in cycling lanes, pedestrians not using pavement/footpath, unauthorized crowding indicate safety issues.
5. Pedestrian Speed - Yes. We found this interesting. Another simple yet effective safety metric. Simply put, if you have a road where you're observing people run or scramble in traffic, it’s unsafe for sure. A safe road would lead to calm behavior from people and they would walk a regular pace.
These are some of the metrics that we use to compute a road safety index of an intersection for our analysis. We definitely advocate intelligent transportation systems driven by deep data analysis but, at the same time, we recommend using road safety analyses processes that need less data. With advances in AI, we can create new road safety metrics that don’t need a human life to be a data point.
Please comment if you think something can be added here.