Cities are getting smarter, but are they getting any safer? Over the last decade, DfT data shows that road casualties have been on a gradual decline, with a 37% decrease in all types of casualties in 2021 compared with 2011. But is this enough to meet Vision Zero targets?
Lockdown created low casualty numbers with low traffic levels in 2020. But DfT’s latest data showed signs of a return to pre-pandemic trends (2019) in 2021 - there was a 7% increase in fatal incidents and an 11% increase in all casualties compared with 2020.
Progress with road safety has been delayed, but these blips in the direction of overall trends have also shone a harsh spotlight on the previous speed of progress not matching up with Vision Zero aims. It’s a welcome reminder.
As showcased by the graph below, the current rate of decline needs to double if we are to achieve Vision Zero by 2041. Clearly, a more proactive, technologically advanced and accelerated strategy is required to trigger a sharp downward shift to put this trend on track.
Near misses are a vital indicator of where dangers lie. By identifying near miss hotspots, councils can preempt and identify accidents before they happen - it’s one of the most effective ways road safety can be drastically improved.
But how can our roads be better equipped to spot and reduce near misses, casualties and transform road safety? To achieve this, it’s essential councils have technology that is able to provide an accurate understanding of how roads are being used and then use this insight to make them safer.
Why isn’t our infrastructure better at preventing road accidents?
Currently there are three main challenges that are preventing better understanding of road safety. Firstly, and fortunately, the sample size of accident data is small and only captures the most serious incidents that are reported. Secondly, road safety improvements are made after the accident has occurred. And, thirdly, the data is not detailed enough to accurately identify the cause of the accident. There is a need to fill these gaps with real-time, granular data to help cities proactively address hazards before accidents happen. However, defining what qualifies as a hazard or ‘near miss event’ is a challenge in itself. It can be very subjective, based on how ‘threatened’ or ‘in control’ a road user feels. For near miss reporting, universal metrics are needed, combined with the contextual information that surrounds that interaction.
How computer vision is overcoming these challenges
In order to reduce the number of near misses, we have to be able to identify the root cause. The most advanced computer vision technologies do this by tracking the movement path of road users and calculating interactions between them.
This requires an extremely high level of precision and technology that can track road users as 3D objects - not 2D images - thus achieving an accurate projection of an object’s footprint on the road and its true path. Utilising the shape and size of each vehicle is critical for automated detection. Without this, too many spurious cases will be raised.
Limitations of 2D detection: 2D detection boxes suggest interactions that do not take place, as the overlapping boxes illustrate.
With this in mind, it is useful to consider the following when identifying near miss events and their root cause:
- Proximity: how close two road users are in physical space.
- Encroachment time: the difference in time between two road users.
- Estimated time to collision: factoring in speed and direction of road users to calculate the time to collision if they continue on this path.
For the future of road management
By approaching near miss detection using the techniques above, it’s possible to identify not only how many near miss events take place but categorise them too. Combined with advanced AI-based detection capabilities, authorities and road safety organisations can identify accident hotspots, investigate their root cause and resolve situations - before further incidents occur.
From surveying customers in the industry, it's widely known that this proactive approach to road safety management is in high demand across the industry, not to mention its reputation as an essential asset on the journey to Vision Zero. If implemented effectively, intelligent AI technology could see Vision Zero reached significantly ahead of current targets. The road safety transformation is well underway.
Mark Nicholson CEO and Co-Founder at VivaCity
