Network Traffic with Self-Driving Cars
Self-driving cars are the hot new trend, with Google and Uber leading the wave. This new and experimental form of transportation is coming at a critical time for road systems everywhere. The number of cars on roads globally is set to increases to 2 billion in the next 15 years. As more cars hit the road, traffic congestion also increases. Road congestion is both annoying to be in and expensive to deal with, costing the global economy $1.4 trillion a year. There is also the environmental drawback of having an excess 56 billion pounds of CO2 pumped into the atmosphere. As technology increases and urbanization proliferates, the need for cars is not going down anytime soon and neither, it seems, is the road congestion. But there is hope.
Autonomous cars, while flashy and cool, also provide an avenue for dealing with traffic congestion. The argument has been made that self-driving cars can make roads safer and commutes faster by decreasing the amount of accidents caused by human error. But is that true? The here.com article on traffic congestion takes the unique approach of breaking down the benefits of autonomous cars into 3 key timelines. 1 – the short term (0-5years), 2 – the midterm (5-20 years), and 3 – the long term (20+ years). Each of these periods can also be approximated as a network traffic model.
1 – Short Term
In the short term we see the roll out of semi-autonomous cars. These kinds of cars, equipped with cruise control, emergency breaking, etc. are distributed widely and will produce a positive impact on traffic congestion. The driver is still in complete control, but the machine corrects for behaviors that cause common accidents. By reducing the amount of accidents on the road, the total travel time for everyone would decrease.
If we try to display this behavior in a traffic network it could be characterized as such: For every non-autonomous car x, any semi-autonomous car x’ would decrease the total travel time of the path.
Example: For a path that takes 3 hours for every non-autonomous car, the addition of a semi-autonomous car reduces the travel time by an hour per semi-autonomous car.
2 – Midterm
The midterm see’s the rollout of a select number of fully autonomous cars. The combination of fully autonomous cars and man-controlled machines will cause a lot of confusion. This is the turbulent readjustment period where road accidents increase as people get used to the new system. Road accidents will increase as the number of autonomous cars increase until the point where there are more fully autonomous cars that semi or non-autonomous cars on the road. The increase in accidents with trigger an increase in commutes.
Modeled as a traffic network, this behavior can be summarized thusly: For every non or semi-autonomous car x or x’, any fully autonomous car x’’ would increase the total travel time of the path.
Example: For a path that takes 3 hours for every non-autonomous car, the addition of a fully autonomous car increases the travel time by 2 hours per fully autonomous car.
3 – Long Term
With the long term we finally see the long anticipated drastic decrease in traffic congestion. Within a system where most if not all the cars on the road are fully autonomous and traffic information is shared continuously to all cars on the network, there should be virtually no accidents or congestion. With the decrease in traffic congestion, the travel time to any destination would become negligible.
When this behavior is modeled as a traffic network: For every non or semi-autonomous car x or x’, any fully autonomous car x’’ would decrease the total travel time of the path by a large factor.
Example: For a path that would take 3 hours for a non-autonomous car, the existence of a fully autonomous car decreases the travel time by 10 hours per fully autonomous car.
