According to the World Health Organization (WHO), more than 1.25 million people die each year due to car crashes. This costs most countries about 3 percent of their gross domestic product, and it is the leading cause of death for people between the ages of 15 and 29. The WHO has set a goal of halving the number of deaths and injuries from car crashes by 2020, which will be a monumental undertaking. Enter big data and predictive analytics.
Crash Maps and Predictive Analysis
By collecting data on car crashes, such as where, when, and why they happen, computers can produce a predictive crash map. In short, a computer analyzes historical data as well as any new incoming data, in order to predict the location of high-risk areas, be it roads or highways.
Tennessee
In 2013, Tennessee launched a crash prediction program at a cost of $263,000 for the first year alone. The program sifted through every crash report and traffic citation, combined with weather conditions and special events, to produce maps of 6-by-7 mile areas in four-hour increments that predict serious crashes and fatalities.
These maps are made available to state troopers on their patrol vehicle computers, allowing them to see where they should patrol or set up safety checkpoints. The data even allows for troopers to create enforcement plans for the month.
As a result, the agency’s crash response timed dropped by 33 percent between 2012 and 2016, from 37 minutes to 25 minutes. Traffic fatalities between 2013 and 2015 fell 3 percent, compared to a national average of 7 percent, but saw an 8 percent spike in 2016. However, officials were unsure whether the software needed reprogramming or if factors such as more people on the road and better gas prices could be the cause.
Commercial Applications
Big data also has commercial applications when it comes to transportation. Data can be used to monitor the safety of driver employees, predicting driver risk thanks to collecting information on speeding, seat belt usage, braking and acceleration habits, and after-hours vehicle usage. It can also be used to identify car models that experience higher amounts of accidents, allowing a fleet manager to retire a model if a design flaw is suspected.
There are other potential uses for the same data, as well. In a private setting, many of these same statistics could be used to show a parent whether their child is a safe driver, or at risk for an accident. In a government setting, data could identify whether a road needs to be redesigned based off of drivers’ habits.
Autonomous Vehicles
Big data is also being used to make autonomous cars safer, reducing the already small amount of crashes they have been part of. Tesla, for example, uses machine learning, another facet of big data, to create a sense-plan-act program, which uses high volumes of data to predict outcomes for an action the autopilot takes in any given scenario. While this could, in theory, lead to the infamous trolley problem of one vs. many, it is an attempt to allow a computer to make what it, based on data, believes is the smartest, safest decision.
Safety
With an estimated 10 million self-driving cars on the road by 2020, safety is a big concern. With about 81 percent of car crashes are the result of human error, and the person behind the wheel of an autonomous car when it crashes is still to blame, safety is the top concern. That means collecting as much data as possible. Google’s Waymo, for example, has driven a staggering 10 million miles in nine years of testing. They have simulated nearly 7 billion miles of travel. The Waymo vehicles use sensors to detect people, bicyclists, other vehicles, and even road work more than three football fields away in full 360 degrees. All this data is fed into machine learning. It can even detect a cyclist signaling with their arm which way they intend to turn, and the AI acts accordingly.
Increasing the safety of driving by going autonomous will also have a positive effect on the economy in general. In the U.S., for example, car crashes cost the economy an estimated $871 billion per year, but autonomous vehicles and other advanced driver assistance could cut car crash deaths by up to 90 percent. This would result in not only thousands of lives saved each year, but also saving about $190 billion in healthcare costs.
Big data is already helping keep drivers safer on the road, but this will only continue to get better. As semi- and fully autonomous cars spread, more data will be available, making prediction models better, and in turn make cars safer.
About the Author
Avery Phillips is a freelance human based out of the beautiful Treasure Valley. She loves all things in nature, especially humans. Leave a comment down below or tweet her @a_taylorian with any questions or comments.
Sign up for the free insideAI News newsletter.
