Intelligent Traffic Management

Rob Bell

Issue 16, October 2018

Could we eventually see a total abolishment of traffic signals?

I have been fascinated to learn recently about some IoT based transport initiatives right on our doorstep. Indeed in NSW where we are based, our state government is spending millions of dollars in smart technology to keep traffic flowing with existing infrastructure. Traffic lights that talk to each other, adjust for traffic loads in different directions, and other intelligent systems that can all get more from our existing lanes.

We’ve all been there - stuck in traffic for no apparent reason, or exiting a stadium carpark once the football has ended. While catastrophic events are sometimes the cause of major traffic disruptions, much of the peak hour traffic is indeed simply caused by the large volume of vehicles on the road.

Consider if your vehicle could talk to a centralised road management facility. Future planning (as in, the duration of your trip) could predict your arrival at an intersection, as well as the direction you’re going. Perhaps you even pay a surcharge (like Surge pricing) to drive during peak hour, but it’s free in off-peak times, as part of the load balancing system.


There are two distinct winners with squeezing efficiency from our existing infrastructure.

Less Infrastructure Spending

While our Governments like to promote infrastructure spending, the reality is still that if it’s not required, the money can be spent on health, education, and other vital resources which can no doubt use the cash. Infrastructure spending currently is critical to keep things flowing even with exploding populations and higher density housing.

However if we’re able to increase the efficiency of our road use, we can theoretically squeeze more vehicles onto the same roads, with better traffic flow than we currently have!

Lower Emissions

One unintended benefit of better traffic flow is lower emissions. If we’re all driving electric cars, perhaps this point is mute, but I still think we’ll be seeing the internal combustion engine (at least hybrids) on the roads for some years to come.

Naturally, the less time vehicles spend running for no reason (i.e. sitting in traffic) the less emissions they will produce. In turn they’ll consume less fuel and cost their operators less in travel costs too!


The human factor is undoubtedly one of the major barriers to this type of automated traffic management. People, by nature, are relatively unpredictable. Outside of distant memories of the Holden Precision Driving Team in a bygone era, people aren’t all that capable of timing things with repeatable accuracy. Humans can easily become frustrated, distracted, and impatient in traffic, and it all goes downhill from there. Even an incident on the side of the road can cause traffic havoc, even though there is actually zero direct impairment to the traffic flow or lanes in the direction you’re travelling.

In Australia we commonly call this “rubber necking”, because people love to gawk at the vehicle carnage and destruction, or have a chuckle at the person being booked by the police for whatever they did. This is human nature, and not something even a 20-lane highway can solve either.

With AI at the wheel, and indeed shared information between vehicles, autonomous cars can not only time things specifically to a microsecond, they can be aware of issues outside of their sight-lines too.

Consider when you’re turning across three lanes of traffic. Two lanes are banked up, but for some reason the third is clear. Do you cross? Or do you anticipate the vehicle still travelling at full speed which you’re going to collide with? If vehicles talk to each other, some simple vector math could predict a collision in this circumstance with a variety of options. Slow the vehicle in the third lane or delay the vehicle turning, for example.

The same way a game knows whether your virtual paddle was in the right spot to bounce the ball back, can prevent collisions by adjusting parameters ahead of time. This vector math increases safety, and could reduce vehicle collisions to zero.

This certainly won’t get us beyond all traffic incidents (though arguable an AI vehicle would spot a drunken pedestrian further in advance than a human could too). However if your autonomous electric vehicle is driving you to work, you can get that extra little bit of sleep, watch some Netflix, or check your emails so once you arrive at the office you get some real work done.

Once this becomes our reality, even though vehicles would be cruising at a slower top speed, you’ll probably still average faster.

Perhaps even traffic lights become obsolete because cars automatically request clearance to pass the intersection, with an intelligent AI engine making decisions to enhance traffic flow. When we move from human-based thinking and behaviour to a totally binary or AI process, all existing predispositions around what our roads look like goes out the window.

Perhaps best of all, unless your AI-driver has a lead foot, there’s no speeding tickets either! You may have to find something new to rubber-neck at though. But if you’re not driving, you can always spend your journey watching “fail” videos on YouTube anyway.