Cop Car Chases And AI Autonomous Cars

An AI self-driving car has the potential to be involved in a car chase thus the AI needs to be ready with responses and action plans. (GETTY IMAGES)

By Lance Eliot, the AI Trends Insider

Most people have a New Year’s resolution involving going on a diet or exercising more.

Apparently, here in Los Angeles, making a resolution to lead a frantic car chase and be relentlessly pursued by the police is popular too.

At the start of this year, just a few short days after New Year’s Day celebrations including the famous and breathtaking Tournament of Roses Parade, a seemingly crazed driver led the cops on a two-hour wanton driving romp throughout the Southern California area.

In this highly dangerous venture, the driver managed within a half-hour of the chase to ram into a person riding a scooter and after doing so continued the driving rampage.

Notably, having later skidded into parked cars and after multiple attempts to be stopped by the police, the fleeing car began to look like something you would see at the end of a rather eviscerating demolition derby. The front windshield was shattered, and observers wondered whether the driver could now see the road ahead at all. The sides and front and rear of the car were all dented and various under-body parts of the vehicle were hanging loose or dragging or had fallen off.


You might not be aware that there is a well-known method often used by the authorities to try and curtail a chasing car, a method known as the PIT maneuver.

Most Californians have heard of the PIT maneuver since we are considered the capital of car chases in the United States.

We seem to have more car chases than any other state.

Last year, we had a reported 800 or so official car chases.

To count as a true car chase, there must be a police car or highway patrol car that undertakes the chase. Once a formal car chase gets underway, the news media usually provides nonstop coverage and sends their helicopters that do freeway traffic updates over to the top of the chase for breathless reporting. We all seem to tune into our radios or TVs or live streaming to keep up with the chase and where it is going.

The letters of “PIT” are said to mean Pursuit Intervention Technique, or some say it means Pursuit Immobilization Technique, or maybe it means Precision Intervention Tactic, or some believe it is Push It Tough.

This tricky maneuver involves a police car coming up alongside the fleeing car and then bumping into it, doing so to cause the target car to spin-out. It is hoped that the spin-out will then allow the police cars to surround and block the fleeing car from further movement.

When I have tourists here and a car chase ensues, I can potentially hoard over them the keen knowledge of what transpires in these car chases and watch as they often become mesmerized by the chase. They often falsely assume that the PIT maneuver will for sure end the car chase. It is not a guarantee of ending the car chase.

Indeed, what made this recent car chase especially notable was the fact that the police tried the PIT maneuver a near record of four times on this fleeing car. Each time, the car did a spin-out, but each time the driver managed to regain control and drove off. Some falsely assume that the spin-out will cause the target car to become disabled and unable to further move. That’s not the case. The main notion is that the driver is supposed to become disoriented, momentarily, and the spin-out will allow the police to then corner the suspect by then surrounding or encircling the stalled car with a circle of blocking police cars.

On the fourth PIT maneuver, the bumper came flying off and the rear window of the car became completely smashed up.

Undeterred, the driver continued the romp. You might wonder how did the car chase eventually end? He ran out of gas. Yes, the driver took to the freeways and headed down south toward San Diego. He got maybe half-way there and ran out of gas on the freeway. Once his car came to a halt, the police cars surrounded him. The driver still resisted and so the canine unit was used to help encourage the driver to give up, which he finally did so.

Not all of our car chases are quite so spectacular.

We have ones that are short-lived and by the time you hear about it taking place, the chase has ended. Enterprising entrepreneurs have kindly provided smartphone apps that will alert you when a car chase gets going and can point you to local online coverage, allowing you to hopefully not miss a moment of the car chase excitement.

There are some locals that say the only reason they care about the car chases is to make sure they don’t inadvertently get embroiled into one that perchance occurs when they are also on the road. By keeping tabs on where the car chase is going, you can try to reroute away from the path of the chase. Others are perhaps more forthcoming and openly state that they relish watching the car chases and believe it to be some of the best free entertainment around. For those of you that are historians, you’ll likely recall one of the most famous car chases ever, the O.J. Simpson chase, which was a slow chase that weaved throughout the SoCal environs and drew huge crowds in-person and was watched in awe by people globally via broadcast TV.

According to various statistics about police-pursued car chases, the average nationwide chase only goes for about 3 miles.

I’d wager that here in SoCal we tend toward longer car chases.

Part of the reason entails our fine weather and the sprawling roadway infrastructure. It is perhaps easier to draw out a car chase here than it might be in other parts of the country, and we offer year-round weather befitting a robust car chase.

We also are a car-centric location and tend to use our cars more so than might be occurring in other locales.

I often get asked why the police don’t just fully ram the fleeing cars. The PIT maneuver is a bumping procedure and it is not the same as actually ramming a car, and so this begs the question why not indeed do a full ram instead? Well, you need to consider the risks to the police in undertaking a true ramming action. Trying to ram directly a speeding car is a dicey proposition. Yes, you might stop the fleeing car, but you might also cause injury or maybe even death to the police undertaking the ramming.

In fact, one of the aspects about car chases that makes it so dangerous is the possibility that the fleeing car might ram into the police, or might ram into other cars, or ram into pedestrians, or as in this recent case ram into someone on a bike or scooter. There can be a variety of untoward results during a car chase.

You might be somewhat surprised or perplexed to know that sometimes a car driver that has nothing to do with the chase will intervene when it gets near them, trying to aid the police by blocking the fleeing car. The police say don’t do that. Leave the car chase to them and avoid getting involved in the car chase. These eager vigilantes think they are doing a grand thing by aiding the authorities. Though perhaps well-intended, it is not welcomed and in fact you could get in trouble for intervening.


There is an ongoing debate about whether the police should even pursue a fleeing car at all.

Maybe it is safer to let the crazed driver get away.

Why risk everyone else during the car chase?

Doesn’t the act of chasing the suspect lead to the driver being outrageously reckless and making dangerous driving maneuvers?

If you let the driver go, they presumably would resume normal everyday driving practices and thus the risks to others on-the-road is lessened.

The counter-argument involves the idea that you are letting get away a potentially violent criminal that might be willing to break the laws in other even worse ways. Suppose the wanted driver later on opts to harm someone, perhaps killing someone? If you were a loved one that had a close relative or friend harmed by that criminal, you would certainly bellow that had the police fully chased and caught the culprit, the subsequent crime would not have been committed. Don’t let dangerous people get away, would be the mantra there.

It is a delicate balance. Go after a driver and incur the risk that someone during the car chase will be harmed, versus letting the driver get away, but maybe they will later commit a dastardly crime. Some assert that if you routinely let a fleeing driver get away, you’ll encourage more car chases, since the crooks will know that you aren’t going to try and stop them. This seems to encourage illegal behavior, rather than discouraging it. If you always pursue the driver and make sure they get caught, it would seem to send a message that might actually curtail the number of car chases.

There are car chases that do not end in the driver actually being apprehended. I think that’s part of the excitement for those that watch the car chases. Besides seeing what kind of wild car driving the person does, there is also the fascination about whether the person will actually succeed in getting away. It is almost always the case that the police end-up with the getaway car, but once the car comes to a halt, and if the driver and its occupants jump out and scatter, they are sometimes able to avoid arrest.

We had another recent car chase in which the occupants leap out of the fleeing car, doing so during the car chase.

They hoped to get away by presumably making the police choose between continuing to follow the car or having to stop to try and get them. If that was their logic, it wasn’t very well thought out. These car chases often involve dozens of police cars in pursuit. Thus, the police usually have ample resources to then continue the car pursuit and also have some police that stop to go after the fleeing occupants.

Another question often posed asks whether the car chase could be undertaken entirely by helicopter.

Just have a police helicopter that follows the fleeing car. The car driver might not even realize the helicopter is tracking them. In that case, there isn’t a car chase per se underway and the driver might drive in a safer manner. The police helicopter can radio the police on the ground to keep them attuned to where the driver is, and once the car stops, the police can surprise the driver by suddenly appearing and apprehending them.

This is a nice idea, but it often falls apart in practical terms. There are places that you could drive into or under and for which the helicopter cannot readily follow. Once inside those places, like say a multi-level parking structure, the driver is likely to be able to get away on foot, doing so prior to the time that the helicopter can alert the ground forces to try and get the crook.

There are though many jurisdictions that require the police to weigh a multitude of factors before giving chase via cars. If the driver is not believed to be armed and dangerous, and if the car chase was sparked by something rather benign, such as expired tags on the license plate, the police might opt to follow only via helicopter, doing so knowing that it increases the risk that the driver will maybe get away scot free. If the expired tags turn out to show that the car was actually stolen, and since this is a more serious crime, the police might deduce that this is someone of an inherently dangerous nature and so the risks of a full car chase are warranted.

Most of the car chases usually involve male drivers.

It seems that the chances of a female driver undertaking a car chase is less common. I’m not making any gender-related remarks about why this is the case, and only pointing out that statistically the odds are higher that the fleeing driver is a male rather than a female. There was a police chase involving a female driver and for which she had originally been sought because she rolled through a stop sign, which is considered a low-level or relatively benign “criminal” act. The police then ran a background check via the plates of the car and discovered that she had no prior criminal record. They decided to abandon the car chase and instead waited at her home, which she eventually returned to, and she was arrested there.


There is more to car chases than meets the eye.

If the police are able to guess where the fleeing car is headed, there is a chance of setting up a roadblock to try and halt the car. Another alternative to a roadblock involves laying down a spike strip, which is a means of potentially ripping up the tires of the car as it runs over the metal spikes, which then will hopefully disable the car from driving much further. There are other inventive approaches such as one that casts a kind of metal mesh net over the fleeing car. These are techniques that have their own advantages and disadvantages, thus it is not necessarily the case that they are a foolproof way to successfully end a car chase.

A crucial part of the equation involves trying to ascertain the risks to humans. There is the risk of injury or death to innocents that have nothing to do with the car chase, other than perhaps being in the wrong place at the wrong time. There is the risk of injury or death to the police involved in the car chase. We need to also include the risk of injury or death to the driver of the fleeing car and its occupants, if any.

You cannot assume that the occupants in the fleeing car are necessarily criminals. They might be carjacked or kidnapped. I mention this because there have been car chases involving cars with multiple occupants, and I’ve had some tourists that said the police ought to just force the car off-the-road and maybe it crashes or rolls over. This is chancy since you don’t know the nature and status of the occupants, and you cannot risk tossing the baby out with the bathwater, so to speak.

Plus, as much as you might detest the driver of the fleeing car, you cannot just summarily decide that the driver should be killed to stop the car chase. I say this because I’ve had some tourists that ask why not just have the police car pull alongside the fleeing car and have a police officer shoot their gun at point blank range and kill the driver. Wouldn’t that end the car chase?

I suppose it would end the car chase.

Of course, the fleeing car is for sure going to do something even wilder once the suspect is dead and still behind the wheel, but putting that aside, I don’t think we can ask the police to render such outright killings. We do have a due process of law here.

The counter-argument by some is that the fleeing driver is putting others in danger and it is as though they have drawn a gun and are pointing it at the public, and so wouldn’t a criminal pointing such a gun by eligible to be shot by the police? The argument seems perhaps persuasive, since the multi-ton car is a kind of weapon, but no, our society has not made the logical leap that this is the equivalent to carrying around and threatening people akin to wielding a gun.

Other aspects that you might not notice about a car chase involve the police doing other subtle but important actions to try and reduce the chances of someone getting hurt.

For example, the police that are following the fleeing car are typically a few yards back of the car, rather than being directly at the bumper of the car. This is thought to reduce the sense of pressure on the wild driver and hopefully will calm them and avoid the high speed and chancier kinds of chases.

The police will usually have their sirens on and be trying to warn other traffic that something is afoot. If other car drivers are paying attention to the normal aspects of driving, and even if unaware of a chase taking place, they will hopefully hear or see the sirens and flashing lights of the police cars and will pull over as expected. The police will sometimes also try to get ahead of the fleeing car and block intersections, and otherwise try to ensure that other drivers and pedestrians stay away from the underway car chase.

There’s another tricky element you need to consider. Suppose the driver or the occupants have a gun or other weapons in the car with them. I mention this aspect because it also explains why the police want to stay a bit of distance away from the fleeing car. At any moment, the driver or its occupants could opt to open fire and shoot at the police. That’s dangerous for the police and obviously part of the goal of these chases is to try and reduce the chances of the police getting harmed, in addition to protecting the general public.


What does this have to do with AI self-driving driverless autonomous cars?

At the Cybernetic AI Self-Driving Car Institute, we are developing AI software for self-driving cars. One aspect to consider involves the potential of a car chase that in one means or another might involve an AI self-driving car and thus the AI ought to be ready for this to occur.

Allow me to elaborate.

I’d like to first clarify and introduce the notion that there are varying levels of AI self-driving cars.

The topmost level is considered Level 5. A Level 5 self-driving car is one that is being driven by the AI and there is no human driver involved. For the design of Level 5 self-driving cars, the automakers are even removing the gas pedal, the brake pedal, and steering wheel, since those are contraptions used by human drivers. The Level 5 self-driving car is not being driven by a human and nor is there an expectation that a human driver will be present in the self-driving car. It’s all on the shoulders of the AI to drive the car.

For self-driving cars less than a Level 5 and Level 4, there must be a human driver present in the car. The human driver is currently considered the responsible party for the acts of the car. The AI and the human driver are co-sharing the driving task. In spite of this co-sharing, the human is supposed to remain fully immersed into the driving task and be ready at all times to perform the driving task. I’ve repeatedly warned about the dangers of this co-sharing arrangement and predicted it will produce many untoward results.

For my overall framework about AI self-driving cars, see my article:

For the levels of self-driving cars, see my article:

For why AI Level 5 self-driving cars are like a moonshot, see my article:

For the dangers of co-sharing the driving task, see my article:

Let’s focus herein on the Level 4 and Level 5 self-driving cars. Much of the comments apply to the less than Level 4 self-driving cars too, but the fully autonomous AI self-driving car will receive the most attention in this discussion.

Here’s the usual steps involved in the AI driving task:

  • Sensor data collection and interpretation
  • Sensor fusion
  • Virtual world model updating
  • AI action planning
  • Car controls command issuance

Another key aspect of AI self-driving cars is that they will be driving on our roadways in the midst of human driven cars too. There are some pundits of AI self-driving cars that continually refer to a Utopian world in which there are only AI self-driving cars on public roads. Currently there are about 250+ million conventional cars in the United States alone, and those cars are not going to magically disappear or become true Level 5 AI self-driving cars overnight.

Indeed, the use of human driven cars will last for many years, likely many decades, and the advent of AI self-driving cars will occur while there are still human driven cars on the roads. This is a crucial point since this means that the AI of self-driving cars needs to be able to contend with not just other AI self-driving cars, but also contend with human driven cars. It is easy to envision a simplistic and rather unrealistic world in which all AI self-driving cars are politely interacting with each other and being civil about roadway interactions. That’s not what is going to be happening for the foreseeable future. AI self-driving cars and human driven cars will need to be able to cope with each other.

For my article about the grand convergence that has led us to this moment in time, see:

See my article about the ethical dilemmas facing AI self-driving cars:

For potential regulations about AI self-driving cars, see my article:

For my predictions about AI self-driving cars for the 2020s, 2030s, and 2040s, see my article:

Returning to the topic of police pursuits and car chases, let’s consider how an AI self-driving car might get involved.


Before I offer some real-world aspects, I’d like to tackle the Utopian crowd that says we’ll never have car chases once we have AI self-driving cars.

In their view, it is pointless to talk about car chases since they won’t ever happen.

Why bother to consider something that you know won’t occur?

Well, this presupposes that we have only AI self-driving cars on our roadways.

As I’ve already mentioned herein, the day in which we have only AI self-driving cars and no legacy cars is a long way off in the future. We don’t even know for sure that such a day will arrive. Perhaps society might not relent to having only AI self-driving cars on our roadways and insist that legacy cars can still exist. Or, maybe we aren’t able to achieve true AI self-driving cars and continue to muddle along with human drivers, albeit those human drivers being augmented by some pretty slick advanced automation driving capabilities.

So, I will concede that if we someday had only AI self-driving cars, true ones, and if you are saying that as a society we would have the auto makers and tech firms ensure that there was some kind of “virtual spike strip” that would prevent an AI self-driving car from engaging in a car chase, I suppose we would not have car chases.

This though also has some weak points in that you could indeed have a car chase that gets started, but for which it would then presumably be easier to curtail.

There is a chance of a carjacking, or some refer to it as a robojacking when involving an AI self-driving car. There is also the chance of a security breach that allows for someone to overtake the built-in controls of the AI and make it engage in a car chase. There is the chance too of a bug in the AI system that would allow it to undertake a car chase.

My point being that the Utopian world still has to consider that there are various “edge” cases in which a car chase might still take place.

I suppose if you are viewing the Utopian world as an all-perfect place, you might wave a magic wand and say that no human will ever do any carjackings, and that AI self-driving cars will never be vulnerable to a hack, etc. I wonder if this Utopian world also has candy and chocolates that flow out of the taps of our drinking fountains?

For my article about virtual spike strips, see:

For my article about carjacking of AI self-driving cars, see:

For my article about AI self-driving cars driving illegally, see:

For potential bugs in the AI of self-driving cars, see my article:

For back-door security hacks of AI self-driving cars, see my article:

I think that covers the make-believe Utopian world and so now let’s focus on the real-world.


We’ll start with the notion that an AI self-driving car might become an innocent that gets inadvertently enmeshed into a car chase.

Suppose we have a legacy car being driven by a human that is being pursued by human driven police cars. I had mentioned earlier that the police cars usually turn on their sirens and flashing lights, in hopes of warning other drivers to stay clear of the car chase.

Imagine that the AI self-driving car is driving along on some city street, quietly making its way to some destination, and all of a sudden it happens to come in contact with a police car chase that is underway. What will the AI do?

You might say that the AI should have already heard the siren or seen the flashing lights of the police cars. Yes, I agree that a well-designed and developed AI self-driving car should be able to detect the police car aspects. The visual sensors of the AI self-driving car will hopefully have detected the flashing lights and the sensor fusion will report to the virtual world model that there is a potential police car in an emergency mode. The AI Action Planner should then consider what to do, presumably pulling the self-driving car to the side of the road, or maybe rerouting the self-driving car away from the police car.

In terms of actually hearing the police siren, I’ve been advocating that AI self-driving cars need to have an audio listening capability, which most do not yet have.

As humans, we use our ears to listen for sounds that warn us about various driving related matters. You might hear the crunch of two cars hitting each other, being able to hear the sound before you can actually see where the accident has occurred. You might be listening and hear your own car make a sharp banging noise, which maybe means you’ve hit a piece of debris that you otherwise did not see on the roadway. Or, you might hear a police or fire department or ambulance siren, doing so before you can actually see it.

The omission of an audio listening capability is a significant detriment to driving a car, I assert. We humans use our sight and our ears to help us safely and successfully drive a car. The AI sensors of the self-driving car need to have a listening feature and need to incorporate this into the overall AI aspects of driving the car. I realize that some will say that by adding an audio feature, you are going to have AI self-driving cars capturing sounds all the time, and this could be a privacy invasion.

Though I certainly understand the aspect that this use of an audio feature can be a privacy issue, I usually point out that with the other sensors that are already on an AI self-driving car you have presumably accepted the potential for various kinds of privacy loses. The cameras on the AI self-driving car are capturing images and video of whatever is happening around the self-driving car, doing so pretty much all the time that the self-driving car is underway. That’s a huge privacy question right there.

I am not trying to minimize the privacy question about the audio capture, and only pointing out that we have a much larger overarching privacy debates about all of the sensory capabilities of the AI self-driving car, and thus the audio portion is but one added element.

For more about the audio aspects, see my article:

For privacy concerns about AI self-driving cars, see my article:

For olfactory sensors and AI self-driving cars, see my article:

For what happens when sensors fail, see my article:

For roadway debris detection, see my article:


Another means for the AI self-driving car to potentially become aware of the car chase would be via V2V (vehicle-to-vehicle) electronic communications.

The police cars might be electronically broadcasting a message via V2V to warn other nearby cars to stay away or be wary of the approaching car chase.

In that case, the AI of the self-driving car might receive the V2V, and without yet seeing the police cars or hearing them, be forewarned about taking protective action. I suppose that it might not even be the police cars alone that are broadcasting the V2V warning and it might be other AI self-driving cars doing so too. Imagine that for whatever reason the police cars are not yet broadcasting the V2V warning, and an AI self-driving car happens to detect the presence of the police car via its sirens or flashing lights, the AI might broadcast via V2V a warning to other nearby AI self-driving cars.

Here’s a question I sometimes get.

Suppose the car being chased opts to broadcast via V2V that all other cars should get out of the way?

Could that happen?

Yes, it could, and there might be legitimate reasons other than a car chase that such a message might be broadcast.

This also brings up the conundrum that just because an AI self-driving car receives a V2V does not ergo mean that it is valid and something to immediately act upon. Dealing with V2V is going to be trickier than that (unless you live in the all-rosy Utopian world).

Another question is whether the car being chased might exploit the AI self-driving cars by sending out a V2V that gets those AI systems to actually put their self-driving cars into the path of say the police and thus block the police on behalf of the car being chased.

Could that happen?

Once again, the answer is yes, and it further highlights that just because a message is received it does not mean that the AI should at face-value believe it or act on it without first trying to validate or verify the matter.

Besides V2V, a warning about a car chase could occur potentially via V2I (vehicle-to-infrastructure) electronic communication.

We’re heading toward an era in which our roadway infrastructure will be “smart” and outfitted with all kinds of electronic gadgets. Suppose an intersection has a “smart” camera that detects a car chase that is upcoming. It might send out a flash message via V2I to warn other cars that the intersection is about to become a dangerous spot and stay away.

Okay, one way or another, we have some means for an AI self-driving car to be tipped to the aspect that a car chase is underway.

This is handy and hopefully will allow the AI to try and prepare for the matter, either getting off the road or taking some other path entirely.

Should an AI self-driving car only attempt to stay out of the battle, or should it try to engage in the battle?

Right now, as mentioned earlier, the police take the stance that the innocents such as other cars and their drivers are supposed to say out of the matter.

Will this same policy continue when we have some smattering of AI self-driving cars?

Don’t know.

It could be that society opts to consider using AI self-driving cars as a helper in these car chases. It could be that the police via V2V are able to essentially “commandeer” your self-driving car and get the AI to use it as a pawn in the car chase. Maybe the police route a bunch of AI self-driving cars into a street to block the traffic and try and prevent the fleeing car from having any viable place to go.

Admittedly, it seems hard to imagine that as a society we would be willing to have “innocents” dragged into a car chase like this.

I think that people are going to already have suspicions that the government is going to act like Big Brother and potentially take over our cars. This kind of ability to allow the police or any such authority to takeover our cars, well, it seems hard to believe the public will stand for it. You might argue that if AI self-driving cars are mainly used for ride sharing, it could be that the government regulates that any ride sharing self-driving car must comply with the authority taking it over, but that’s also likely to cause some angst by the public.

For the nonstop use of AI self-driving cars, see my article:

For the ride sharing aspects, see my article:

For the conspiracy theories about AI self-driving cars, see my article:

For the concerns about groupthink, see my article:

For who is responsible and AI self-driving cars, see my article:


Here’s another twist for you – what about the PIT maneuvers?

I’ve mentioned that the police can try to spin-out a car by using the PIT maneuver.

Should all AI self-driving cars know how to do a PIT maneuver?

If so, perhaps the police or other authorities could commandeer an AI self-driving car and have it do the PIT maneuver on a human driven car that is being chased.

This is not overly far-fetched.

Suppose the police have an AI self-driving car and opt to make sure there are no humans riding inside of it.

They then direct the AI self-driving car to go ahead and execute a PIT maneuver on a fleeing car being driven by a human. This means that no police officer needs to actually do the PIT maneuver and spares the human police officer from possibly getting injured while trying to do the maneuver.

Notice that I asked the question by posing whether “all” AI self-driving cars should know how to do a PIT maneuver. You might argue that only some AI self-driving cars should have such a capability. Note that it is essentially a software-only aspect in that you do not necessarily need any special gadgetry on the car to do the PIT maneuver. It certainly helps if the car doing the PIT maneuver has good bumpers and some other stuff on it, but it is not a necessity.

Some would say that perhaps the PIT maneuver software routine would be something only provided on certain kinds of AI self-driving cars, or maybe downloaded only when needed, using an OTA (Over-The-Air) update to load the routine into the on-board AI system.

I’ll add a twist to the twist.

You can potentially defeat a PIT maneuver if you know how to cleverly drive when being approached by a car that wants to do a PIT maneuver on your car.

I doubt many criminals know how to try and prevent a PIT maneuver from working. Anyway, suppose the AI knew how to try and curtail the likelihood of the PIT maneuver from being successful.

Should the AI have that kind of capability?

These kinds of questions all raise various societal and ethics related kinds of points.

We are heading to a time when the AI can potentially be made to do these types of things, such as being able to perform a PIT maneuver and/or potentially avoid being impacted by a PIT maneuver. The technology is going to allow these kinds of capabilities. It won’t be a tech question and instead more of a societal question as to whether or not these kinds of capabilities are going to be allowed or not, and if not then how to prevent them from being put in place; while if they are allowed, under what circumstances will they be permitted to be employed.

For the potential use of ethics review boards, see my article:

For the potential of regulations about AI self-driving cars, see my article:

For my article about defensive driving and AI, see:


I’ve now covered some of the rudiments about a car chase involving a human driven fleeing car and it being chased by human driven police cars and considered how an AI self-driving car might get enmeshed into the matter.

There is the chance of a human driven police car that gives chase toward an AI self-driving car.

As I’ve suggested earlier, the Utopian world says this should never need to be the case.

There should never apparently be a possibility of an AI self-driving car that needs to be chased.

Suppose I instruct my AI self-driving car to rush me to the pier.

I’ll for the moment make the assumption that the AI is going to drive the car only at the proper speed limit and it will not try to drive radically or go above the speed limit.

I’m not saying that will be the case, and indeed have argued in my writing and speeches that we are going to have AI self-driving cars that don’t drive with such strict adherence. Anyway, let’s put that other argument aside for the moment.

For the illegal acts of AI self-driving cars, see my article:

The police find out that there’s a hostage in the self-driving car and so the police opt to follow.

We are now having a car chase.

If you think that all car chases involve going only at high speeds, you’d be wrong.

There are many car chases that we see here in Los Angeles that are low-speed or slow speed pursuits. Believe it or not, there have been times that the driver obeyed all traffic rules and stopped at the stop signs and never went faster than the posted speed limit, nonetheless it was a police pursuit.

You might argue that there’s no need for the police to follow the AI self-driving car.

For example, the police could maybe via V2V ask the AI where the self-driving car is going, and merely then have the police assemble there. Or, maybe the police use the OTA of the self-driving car and can find out where it is headed. Or, the police transmit a message to the self-driving car to tell the AI to stop the car, or maybe even have the AI drive the self-driving car to the police station, placing it right into the laps of the police.

Yes, those are all possibilities.

But, suppose the AI self-driving car is currently incommunicado. It might be out of the range of electronic communications, or it might have the electronic communications being blocked by the driver that is using the AI self-driving car to flee. I’m sure that once we have a prevalence of AI self-driving cars, there will be all kinds of tips posted online about how to try to block the electronic communications features of your AI self-driving car (getting it to become an autonomous car that is not connected to the cloud or any other device).

In the case of no viable means to electronically communicate with the AI self-driving car, it would seem that the car chase is still on.

There might be some kind of other signaling that is agreed to be allowed to perhaps stop an AI self-driving car. Maybe the police show a poster board with a special code on it, and this gets the cameras to realize that the AI self-driving car should come to a halt. You can imagine though that this method has holes in it and could allow for anyone to potentially do the same to your AI self-driving car.

The police might resort to a driving tactic based on known driving behaviors of the AI self-driving car.

For example, if the AI self-driving car is doing a simpleton follow-the-leader driving technique, perhaps the police car could get in front of the being chased car and gradually slow down the police car, which the AI might also then slow down as it would when following any car ahead of it that is slowing down.

There recently was the case of a Tesla driver that apparently was asleep at the wheel, and a highway patrol car got in front of the Tesla and was able to slow down and get the Tesla to also slow down. This was done until the car had been brought to a halt.

I am doubtful that this kind of pied piper approach is going to necessarily work on true AI self-driving cars. It is my hope that true AI self-driving cars will be much more capable. I’ve mentioned many times the dangers involved in people potentially “pranking” AI self-driving cars, doing so by knowing what the AI self-driving car is going to do, and my assumption is that future true AI self-driving cars will not permit this kind of pranking.

For the pranking of AI self-driving cars, see my article:

For my article about 5G, see:

For my article about the pied piper approach, see:

For NLP commands and AI self-driving cars, see my article:


There are various other twists and turns in the circumstance of an AI self-driving car that might be leading a car chase.

For example, perhaps the auto makers and tech firms will include a capability of the AI to realize that it is actually leading a car chase.

This is somewhat tricky and involves detecting other traffic, but in any case, it is something that a human driver would be able to determine and therefore I claim that the AI ought to be able to do the same (since presumably for a true AI self-driving car the AI can drive in whatever manner a human could).

The AI, once it ascertained that it was leading a car chase, might then engage the human occupants in a Natural Language Processing (NLP) dialogue to discuss the matter with them. The AI might try using the V2V, V2I, and OTA to connect with the outside world and find out what might be occurring. The AI might have some kind of fallback posture that if it detects it is involved in a car chase that maybe it automatically pulls to the side of the road or takes some other pre-staged precautionary act.

The other variant of the car chasing would involve the police car itself being an AI self-driving car. Will we be asking police to drive cars themselves, when we presumably have a world of AI self-driving cars, or will we instead have the police be occupants or passengers in an AI self-driving police car?

Some would say that the AI is not going to be able to be the designated driver for a police car because the AI would not be able to drive in the same manner as trained police officers can drive a car, such as being able to make radical maneuvers and drive urgently when needed.

I don’t quite get the logic of those that take such a position. Are they suggesting that the best the AI can do is the average everyday kind of driving? Though that’s the first point of entry for AI self-driving cars, I don’t know why things would stop there. It makes a great deal of sense that the AI would be able to drive in the same manner as a highly trained emergency driver could.

You might even make the case that we might be safer by having the AI drive those emergency vehicles.

If you believe that the reason that we have car crashes today involves various human foibles, those human foibles can likewise happen to a trained police officer or ambulance driver or fire truck driver. I’d say we ought to be aiming to have AI self-driving capabilities that can drive at the utmost of human driving. This would imply that then all AI self-driving cars have the “best” possible driving capabilities. Plus, the AI would presumably be connected with the V2V and V2I and be able to rapidly respond to real-time changes in the traffic and roadway, which a human driver might be less likely to be able to respond.

Using AI swarm techniques, an entire set of police cars could coordinate and act together when undertaking a car chase. On a more macroscopic scale, we presumably could have AI self-driving police cars, AI self-driving ambulances, AI self-driving fire trucks, and any other kind of emergency responding vehicles that would all make use of AI for doing the driving. Whether as a society we’ll accept this approach is a different question, and I’m merely suggesting that it would seem that we eventually would have the AI capability to do this, assuming we can achieve true AI self-driving cars.

For more about a swarm technique, see my article:

For the singularity, see my article:

For safety and AI self-driving cars, see my article:

For my article about the boundaries of AI and self-driving cars, see:


Car chases.

We watch them in rapt attention.

What will happen?

Will the driver get away?

Will the police capture the culprit?

Will there be problems during the car chase?

What is the driver thinking?

How will the police strive to stop the driver?

And so on.

We are going to continue having car chases, until we reach the magical Utopian world that some postulate will occur. There are going to be AI self-driving cars on our roadways that might get inadvertently dragged into a car chase that involves a human driver chase car and human driven police cars, and the AI needs to know how to detect it, avoid it if possible, or escape it if needed.

I’ve also claimed that we’ll potentially have car chases that are led by an AI self-driving car. This seems utterly counter-intuitive to most people and they shake their head in disbelief that it could ever happen. I hope that I’ve made a convincing case that it could happen. We need to be wary of it occurring and try to embody into the AI of the self-driving cars a capability to detect it and try to deal with it, if it comes to play.

We ultimately will likely have AI self-driving police cars, along with other emergency vehicles being driven by AI. I’m claiming that those AI self-driving police cars should be versed in undertaking a car chase, since they might be needed to do so when chasing a human driving a legacy car, or perhaps even when chasing an AI self-driving car.

For the foreseeable future, we are still going to have car chases. Whether you react to this assertion with fascination or horror, either way I’d say that we need to boost the AI of self-driving cars to have specific capabilities dealing with car chases. Putting our heads in the sand and pretending that the AI will never need to cope with a car chase is not a viable approach to the matter.

Hey, I just heard about an exciting car chase happening down the street, so excuse me while I go and take a look.

Can’t miss a really good car chase.

Copyright 2019 Dr. Lance Eliot

This content is originally posted on AI Trends.

[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:]