(Newswire.net — January 4, 2018) — It’s no secret that our current systems for dealing with traffic aren’t perfect. Despite our constantly increasing knowledge and ongoing commitment to safety, traffic deaths are actually increasing. There are more drivers on the road, more interconnected places to get to, and as a result, more traffic accidents to deal with. So what’s the solution?
Some scientists and engineers believe the secret could be turning to artificial intelligence (AI) to solve some of the problems for us.
Analysis and Optimization
Researchers from the Texas Advanced Computing Center (TACC), the University of Texas Center for Transportation Research and the City of Austin are working together on a new system of tools that allow for the in-depth analysis of traffic patterns. Relying on deep learning algorithms, which allow a system to self-improve over time by exposing itself to more data sets or more events, the team is set to unveil their newest tool at the IEEE International Conference on Big Data.
The tool is designed to capture data from raw footage gathered by traffic cameras throughout the city, recognizing the shapes of things like cars, trucks, bikes, and even pedestrians, and create data based on how those objects move and interact with one another. Once categorized, the machine can work with a searchable database, and analyze these patterns to draw significant conclusions.
For example, after collecting data for a period of weeks to months, the system may be able to zoom in on each accident within that period, analyze the sequence of events that led to that accident, and point to a definitive root cause or series of root causes. Usually, it’s clear to investigators that a car has hit the brakes too early, or failed to follow a traffic sign, but this algorithm can detect even more subtle underlying causes, such as improperly managed congestion or a consistent lack of visibility.
With more tools like this available, it will be easier for cities to review and make changes like:
- Adding safety features to dangerous areas. If there’s an area with a high rate of accidents, additional signage, buffer zones, and lighting could help to make it safer.
- Redrawing or readdressing intersections. Problematic intersections can be revised with different traffic light patterns and better design for traffic flow.
- Developing new roads or improving old ones. AI systems could recommend key changes, such as adding a lane or restructuring the shape of a road to minimize the chances of a collision.
It’s even easier when the tools are making suggestions themselves.
Autonomous Vehicles
AI is also emerging as a key feature of autonomous vehicles, capable of driving by themselves, without any human intervention. Though skeptics argue that the elimination of a human driver could introduce even more road vulnerabilities, the evidence points to the contrary; in the first 1.8 million miles of Google/Waymo’s self-driving car project, there were only 13 minor fender-benders, and all of them were the fault of another driver. The AI tech responsible for directing these vehicles has only gotten better in that time.
Considering the fact that more than 90 percent of car accidents are caused by human error, the addition of non-human machines, with more consistent reasoning and controls and faster reaction times, would almost undoubtedly make the streets safer. The trick is to adopt full vehicular autonomy; partial features, such as automatic parallel parking or “cruise” options for highway driving could lull human drivers into a false sense of security, driving the accident rates up.
If self-driving cars become more popular, roaming cities like cheaper, more popularly available taxis, it could also seriously eliminate congestion on the roads. Fewer people would own cars, more people would carpool, and there would be fewer cars on the road at any given time. This decongestion would make all driving faster and more convenient—and safer, too.
AI has a long way to go before it solves all our traffic woes, but it’s already making strides in two important directions: traffic analysis and self-driving vehicles. The more we understand about the roads we drive on and the better equipped our vehicles are, the safer we’re all going to be.