Here’s How Self-Driving Cars Need To Be Prepared To Deal With All Kinds Of Traffic Clogs

Dr. Lance Eliot, AI Insider

Image for post
Image for post

[Ed. Note: For reader’s interested in Dr. Eliot’s ongoing business analyses about the advent of self-driving cars, see his online Forbes column: https://forbes.com/sites/lanceeliot/]

When my children were young, we had a toy that they assembled consisting of seventy-five plastic interconnecting tunnel pieces, including having numerous tall ramps and winding paths, and when a marble was dropped into the topmost funnel it would be of great delight to all as we watched the marble roll throughout the structure.

The kids decided that if one marble is fun, certainly an entire bucket filled with marbles would create even greater joy, so they poured a bucket of marbles into the contraption.

At first, marbles began to flow into the tubes, but suddenly, the marbles at the top came to a halt.

Did something jam the funnel?

Upon inspection, the children discovered that with so many marbles sitting in the funnel, the marbles had collided with each other and did so in a manner that none of them was able to flow out of the funnel.

The kids had invented a clog.

Thinking Seriously About Clogs

There’s another kind of clog that you likely have to deal with every day.

If you drive to work on a freeway, which I do daily here in Southern California, there are inevitably clogs of one kind or another on the freeways here.

The first clog that I usually encounter involves navigating an on-ramp onto the freeway.

For many of our freeways, we are using a metered ramp system.

The way it works is that cars come from a street onto the lower part of an on-ramp and then come up to a point where there’s a traffic signal on the on-ramp.

When the traffic signal is green, you can proceed further onto the on-ramp and then onto the freeway. When the traffic signal is red, you are supposed to stay put until the light goes green.

In many cases, there are actually two lanes on the on-ramp, one that is for those that can use the HOV lanes and one lane for those that aren’t able to use the HOV lanes. Typically, the HOV lane is not constrained by the meter and can proceed ahead at will.

In the case of just a single lane for the on-ramp and with the meter being used, the traffic on the on-ramp is relatively orderly.

The reason why some people like this approach is that it turns what could otherwise be an ugly free-for-all into an orderly sequence series of events.

Rather than cars all jockeying for position, it is clear cut that you wait in line, in the lane, and you take your turn. It’s like being in kindergarten again.

A study done about ten years ago, around the year 2000, in Minnesota, claimed that ramp meters had a substantive positive impact on aiding freeway traffic, including that without the meters there was a 9% drop in available freeway capacity, a 22% increase in travel times, a drop in freeway speeds by 7%, and accidents increased by 26%. I’d vote you take that with a grain of salt and there are various studies that both support meters and refute the use of meters.

You might assume this metered approach eliminates any chances of a clog.

Not so.

There’s at least one loophole in this approach when there are two lanes available via the on-ramp and when one of the lanes is the HOV one that does not need to stop for the meter.

Clogs On The Freeway Onramp

Here’s what often happens.

A car at the front of the line and waiting for the meter to go green is eager to get moving, and the instant the light goes green, the car hits the gas to burst forward. Meanwhile, a car in the HOV on-ramp lane is eager to get ahead of the cars waiting for the meter, and so that driver has hit the gas to zip along past the standstill cars, doing so since they don’t need to wait for a green light.

You end-up with two cars both trying to rocket forward and yet the path just past the meter is often a slimming down of two lanes into one lane (the one lane that will lead onto the freeway).

A dangerous clog can occur.

Clogs In Nature

Clogs can happen in a multitude of circumstances.

A recent research study took a close look at fire ants and how they avoid creating clogs when they are developing their underground tunnel systems.

The study done by researchers from the Georgia Institute of Technology, the Department of Physics at the University of Colorado Boulder, and the Max Planck Institute for the Physics of Complex Systems, involved observing ants during a collective excavation effort.

The researchers placed the ants into transparent containers and rigged up a means to track them and analyze their movements. Using Lorenz curves, the researchers mathematically made various calculations about the work efforts.

They proceeded to also create a simulation involving a cellular automata model and wanted to compare biological behaviors to robo-physical behaviors. This work also pertains to swarm intelligence.

One of the key findings was that the ants seemed to be willing to undertake idle time for some of the ants that were outside the tunnel in order to avoid clogs in the tunnels.

It was a kind of wait your turn approach.

Anti-Clogging Techniques For Autonomous Cars

At the Cybernetic AI Self-Driving Car Institute, we are developing AI systems for self-driving cars. One aspect involves the traffic coordination of multitudes of AI self-driving cars.

There are some pundits of AI self-driving cars that seem to believe that the advent of AI self-driving cars will magically do away with any and all traffic jams.

This is a rather far-fetched assumption. It seems to be based on the notion that all AI self-driving cars will carefully orchestrate their collective movements and therefore they will overtly avoid any traffic jams or clogs.

If you live in some kind of Utopian world, I suppose you can imagine that all AI self-driving cars will politely and carefully communicate with each other in a flawless manner and somehow arise to the challenge of being able to ensure there aren’t any traffic jams.

I assure you this is a thorny problem and not so easily solved.

The reason that their driving is important takes us back to the anti-clogging topic.

The anti-clogging camp would assert that if you include human drivers onto the roadways then you are not going to achieve the full sense of anti-clogging.

Those darned human drivers will inextricably cause a clog.

A justification then of banning human driving would be that it would presumably then allow for no clogs.

Which has the greater weight in our society, people being able to drive or eliminating traffic clogs (of course, there are other reasons for restricting or preventing human drivers)?

You be the judge (for now).

For the moment, we’ll sidestep the question of human drivers in the mix.

Assume that we had an all and only AI self-driving car world.

Would we be able to avoid any and all clogs?

Let’s use the on-ramp circumstance as an exemplar.

You have AI self-driving cars trying to get onto the freeway.

We can assume that the principles of HOV lane use might still apply, and so we might have AI self-driving cars that have no human occupants or maybe one human occupant that are waiting in the metered line, meanwhile there are AI self-driving cars with two or more human occupants and thus considered HOV-permitted and able to speed-up the ramp and not abide by the meter.

The meter goes green and the AI self-driving car at the head of the pack starts to move ahead. The AI self-driving car in the unfettered HOV on-ramp lane comes up to the point where the two cars are going to meet-up and needs to decide which of them goes first. This is reminiscent of the human driver problem earlier described. Now, we have AI self-driving car getting caught up in the same predicament.

What happens?

There is the presumed availability of V2V or vehicle-to-vehicle communications available.

This means that the HOV on-ramp self-driving car, we’ll refer to it as car “X” might initiate a V2V conversation with the AI self-driving car that was waiting for the green light, we’ll refer to it as car “Q” and that X might inform Q that X is barreling ahead and please stay back.

This is somewhat akin to the fire ants.

One fire ant is proceeding into the tunnel, so to speak, and the other fire ant is remaining “idle” as it waits its turn.

Here’s a question for you, why should Q abide by the instructions or edict provided by X?

In other words, why can’t Q tell X that X should slow down and let Q proceed ahead?

Why should one of them be considered the commander of the other?

Indeed, some argue that our driving is based partially on a sense of “greed” or selfishness, whereby traffic generally flows because each car is doing what it can to maximize its own advantage.

But, if you have two “drivers” and each of which demands to go first, what kind of tie breaker do you have?

Who Decides To Avert A Clog

You could say that in this case it should not be up to the two cars and their respective AI’s to decide as to which goes ahead first.

Instead, it should be the infrastructure.

It is anticipated that our roadways will gradually be outfitted with high-tech sensors and other systems, and there will be the advent of V2I, vehicle-to-infrastructure communications.

Thus, in this example, perhaps the meter should be “smart enough” to realize that another AI self-driving car is coming up the on-ramp in the HOV lane, and so the meter then via V2I informs the Q to not proceed just yet (or, maybe keeps the red light a bit longer), and allows X to flow along through the ramp. Or, perhaps the V2I informs the X to slow down and allow the Q to proceed ahead.

How did the infrastructure determine which goes first?

It could be based on some algorithm that tries to ascertain which of the two is “best” suited to go first.

Or, maybe it randomly selects if otherwise everything else about the situation would be considered a tie. As an aside, if you are interested in algorithms for traffic jams solving, you might want to explore ALINEA, one of the more studied such algorithms for this purpose.

There are some that advocate we might consider implementing a points system.

A similar viewpoint is that maybe there would be electronic money exchanged.

Clogs Arise In Lots Of Driving Situations

So far, I’ve focused on the on-ramp example.

This serves as a simple means to look at the clog problem.

Enlarge the scope to the freeway overall.

The number of clogs and the emergence of clogs is many times the magnitude of the on-ramp example.

You’ve got multiple lanes.

Multiple on-ramps and off-ramps.

Hundreds or perhaps thousands of cars.

Each car is headed to its own desired location.

There are miles upon miles of freeway.

There are numerous freeway interchanges.

Maintenance and upkeep of the freeways is taking place and can mire the roadways while doing so. And so on.

The other day, I was driving on the freeway and a car became stalled in the middle of the freeway.

The traffic began to get snarled and it wasn’t at first apparent as to why.

As I got within a few cars of the stalled car, I could see the upcoming cars would come right up to the standstill car and then try to move into the lane to the left or right of the clogging car.

The cars in those lanes would sometimes allow the other cars to get into their lane, and in other cases they would not.

It’s a dog eat dog world.

A self-driving car is going suffer breakdowns.

It will happen.

Guaranteed.

Conclusion

How will the sudden stalling of an AI self-driving car while on the freeway be handled and done so in a manner that avoids any kind of clogging or traffic jam?

If the anti-clogging happens only for those AI self-driving cars near to the incident, by sharing with each other V2V, would this alleviate all clogging or would there still be some residual clogging?

And, it would seem like the residual clogging would have a cascading effect. Self-driving cars downstream of the disabled self-driving car are likely to experience some impact, even if minimal.

You might suggest that there would be a master control system that would oversee all traffic.

It would seek to prevent any traffic jams.

Therefore, rather than the envisioned more localized P2P of the V2V, we might have a “Big Brother” kind of system to optimize traffic flow and eliminate clogs. This would seem like a rather tall task computationally, and one even questions whether it is possible to achieve, given too the logistical vagaries involved.

There will also be some that find this notion somewhat repugnant as it might give the government excessive control and oversight.

It seems doubtful that a “perfect” world of no clogs is likely feasible.

The goal might be instead to focus on minimizing and mitigating clogs. Overall, the hope would be to limit the severity of clogs and the prevalence of clogs. This might combine both a global master control system along with localized P2P systems. It’s an interesting and challenging “edge” problem that will become more apparent as the advent of AI self-driving cars emerges.

Meanwhile, I guess we’ll all struggle with your day-to-day clogs, including that my kitchen sink has now clogged up and it looks like I’ll need to call a plumber.

For free podcast of this story, visit: http://ai-selfdriving-cars.libsyn.com/website

The podcasts are also available on Spotify, iTunes, iHeartRadio, etc.

More info about AI self-driving cars, see: www.ai-selfdriving-cars.guru

To follow Lance Eliot on Twitter: https://twitter.com/@LanceEliot

For his Forbes.com blog, see: https://forbes.com/sites/lanceeliot/

For his AI Trends blog, see: www.aitrends.com/ai-insider/

For his Medium blog, see: https://medium.com/@lance.eliot

For Dr. Eliot’s books, see: https://www.amazon.com/author/lanceeliot

Copyright © 2019 Dr. Lance B. Eliot

Written by

Dr. Lance B. Eliot is a renowned global expert on AI, Stanford Fellow at Stanford University, was a professor at USC, headed an AI Lab, top exec at a major VC.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store