Skip to content
Outline
- What exactly is a Near miss?
- Why are Near misses Important and how do they Complement Crash Data?
- Near Miss Technologies for Safe System Road Safety
- MICHELIN DDI’s Near Miss Solution for Proactive Road Safety
Implementing the Safe System approach and focusing on Vision Zero goals is all about embracing proactive road safety. And being proactive means looking at what happens before a car crash. A near miss, or a near-crash, is a car crash that doesn’t occur. What if it were possible to identify near-crashes on roads? To locate them and analyze them… like car crashes? Near-crashes, like car crashes, would help in showing where there are safety concerns and what kind of corrective actions can be taken, before any car crashes occur. New technology makes this entirely possible.
What exactly is a Near miss?
“Whew, that was close!” We have all had this thought at some point in our driving career. But what actually happened to make us say this?
The NHTSA defines a near-crash as,
“Any circumstance that requires a rapid, evasive maneuver by the participant vehicle, or any other vehicle, pedestrian, cyclist, or animal, to avoid a crash. A rapid, evasive maneuver is defined as steering, braking, accelerating, or any combination of control inputs that approaches the limits of the vehicle capabilities.”
Let’s pull this apart.
“Rapid, evasive maneuver”
In the NHTSA’s definition, “rapid, evasive maneuvers” are described as steering, braking, accelerating, or any combination thereof. These are clear actions on the part of the driver – or pedestrian, cyclist or animal – to change trajectories, to stop short, to get out of the way. Compared to a few seconds before, the driver has completely changed their behavior to stay safe. Near misses therefore reflect a sudden and atypical change in driver behavior at a given location and a given moment in time.
“Approaches the limits of the vehicle capabilities”
When the driver steers, brakes or accelerates, they do it very suddenly, and very strongly. This is not a smooth turn or gradual deceleration or acceleration. This is swerving, and harsh braking and accelerating. This is an event that has inserted itself into what previously was a drive down the road. Near misses therefore reflect a driving event at a given location and a given moment in time.
Near-crashes and crashes therefore have different outcomes, but share the same inputs. The combined information that they provide is highly complementary.
Why are Near misses Important and how do they Complement Crash Data?
Fatal car crashes in the first nine months of 2021 increased by 12% compared to the same time period in 2020. At the same time, to receive funding to fix a car crash location, it has to reach a threshold of crashes. It’s the number of crashes that decides. What if it were possible to identify potential crash locations before any lives are lost? The right resources for corrective actions could be budgeted and allocated.
Near-Crashes: Preventive Road Safety
Studying and analyzing near-crashes and near misses make this possible. As driving events, near-crashes and near misses can also happen repeatedly at the same location. They can therefore be aggregated and plotted on a map. With technology today, it’s even possible to contextualize near-crashes, including information about weather, the time of day, the day of the week…
When these clusters of driving events are combined with car crash reports, road experts have additional information that helps in assessing road safety within their road system. The near-crashes data provides a preventive perspective on creating safe roads for all users. Experts can prioritize locations for road safety action and make accurate decisions about how to introduce Safe System and Vision Zero safety measures.
Near Miss Technologies for Safe System Road Safety
How can road safety experts get near miss information? Different methods exist that record events from outside or inside the vehicle. Below is a quick, though not exhaustive, overview of some of the most common technologies currently in use.
Outside the Vehicle : Intersection Cameras Calculate the Scale of Danger
What?
Cameras positioned at an intersection, so that the intersection itself and the approach from each direction is clear. Vehicle identification possible. The cameras’ memory is finite, so images are relayed to a control room where an operator monitors and records them.
Contextual information?
Visual data on vehicle behavior, surrounding conditions, multimodal context.
Driving behavior?
Traffic light violations, behavior with pedestrians, excessive speeding, harsh braking, at that intersection.
How are near misses identified?
Video data is crunched into data points recording time and motion. The time-measured-to-collision parameter indicates danger over a scale, from “safe” to “collision”.
Feasability?
Major investment to equip all intersections with this equipment, and a time requirement to process all the information received.
Outside/Inside Hybrid: Connected Vehicle Communication
What?
Connected vehicles send and receive signals between each other and surrounding road infrastructure. Drivers receive information about road conditions: road works, accidents, congestion, weather, etc.
Contextual information?
Surrounding road infrastructure and other data sets provide road conditions and context.
Vehicle and GPS data provides information on type of journey (commute, business, leisure), time of day, location. This extra information is helpful for analyzing driver behavior in greater detail.
Driving behavior?
Traffic light violations, stop sign violations, behavior with pedestrians, excessive speeding, harsh braking, for the whole time the driver is in the car.
How are near misses identified?
Through spatial clustering, aggregated driver behavior pinpoints specific locations where and when certain driving behavior always occurs. Because the behavior is repeated across large numbers of drivers, it can be qualified as near miss driving behavior.
Feasability?
Only recent vehicles include the required technology. Major investment to fit all roads and infrastructure with equipment and sensors required for continuous communication with vehicles.
Inside the Vehicle: Dash Cameras are the Seeing Eye
What?
Inside cars, dash cams are typically known as ‘event recorders’.
Contextual information?
Visual data on vehicle behavior, surrounding conditions, multimodal context.
Type of journey (commute, business, leisure), time of day, location.
This extra information is helpful for analyzing driver behavior in greater detail.
Driving behavior?
Traffic light violations, stop sign violations, behavior with pedestrians, excessive speeding, for the whole time the driver is in the car.
How are near misses identified?
Event recorder data is processed using an algorithm that functions in three layers: time, movement across a grid, and presence or absence of a near miss.
> Time helps to identify details and features that provide the car’s status.
> The grid shows the vector of travel and whether there is a target.
> A multi-task layer then categorizes whether a near miss occurred with a target or not.
Feasability?
Unlikely that drivers will accept having all their actions recorded by a dash cam.
Inside the Vehicle: Recording Movement across GPS points
What?
GPS data (from a smartphone, onboard equipment, the vehicle itself) can provide information about travel routes, acceleration and phone handling. Aggregating GPS data can accurately identify and geographically locate critical situations. The technique has been tested on open roads and test tracks.
Contextual information?
Not provided by GPS data. Can be added in afterwards from other data sets.
Driving behavior?
Harsh braking, swerving, harsh acceleration, phone handling, from all drivers for the duration of the time they’re in the car.
How are near misses identified?
Through spatial clustering, aggregated driver behavior pinpoints specific locations where and when certain driving behavior always occurs. Because the behavior is repeated across large numbers of drivers, it can be qualified as near miss driving behavior.
Feasability?
All information can be plotted on a geographical information system. Because GPS data is aggregated, privacy is protected. Covers all geographic areas. No investment in infrastructure. Immediate implementation.
MICHELIN DDI’s Near Miss Solution for Proactive Road Safety
MICHELIN DDi’s Definition of a Near-crash
A near-crash is the result of a succession of unforeseeable factors and incidents. These lead to sudden and atypical driving behavior that causes a driving event, without resulting in a car crash. The driving event can be identified, measured, aggregated and contextualized.
MICHELIN DDi’s solution works with GPS and other data sets to produce actionable insights about contextualized and aggregated driving behavior. A road network becomes immediately legible in terms of ranking near miss hotspots and prioritizing actions.
DOTs have access to a powerful solution for proactive road safety:
- No additional infrastructure required (no cameras, sensors, etc. to install)
- Slips directly into geographic information systems
- Exhaustive coverage over an entire area
- Locations can be ranked in terms of potential risk due to a powerful new indicator
Incorporating near miss events, their identification and location, is essential to developing safer people and safer roads as part of the Safe System approach and meeting Vision Zero goals.
