Traveling in Vehicle Caravans and the Advent of AI Self-Driving Cars


By Lance Eliot, the AI Trends Insider

I was stuck in traffic when a Presidential motorcade made its way from Los Angeles LAX airport to tony Beverly Hills in sunny Southern California. This caravan of cars included an impressive array of special stretch limos, armored SUV’s, and a slew of police vehicles that were making sure that nobody intervened into the VIP line of cars. The police kept ahead of the pact and made sure that intersections were open and ready for the motorcade to flow along without having to stop at any traffic signals.

It caused all of the surrounding or nearby traffic to come to a halt. Some of my colleagues groused at the aspect that we had to wait for the motorcade to pass past us. Why should the motorcade get a higher priority than the rest of us drivers, they bemoaned? I gently tried to point out that waiting a few extra minutes in SoCal traffic is pretty much a daily driving chore anyway, plus it certainly makes sense that to protect those involved in the motorcade it should proceed along quickly. Furthermore, if the motorcade was just part of normal traffic, I pointed out that it probably would inevitably have created even longer waits, especially since the motorcade cars were wanting to stick together as they drove on the roadways.

About a month earlier, there was a funeral procession of cars for a famous celebrity, and the procession or caravan of cars started at a funeral home in one location of town and proceeded on a lengthy driving journey to a cemetery in another city where the deceased was to be buried. This was a similar kind of “caravan” as akin to the Presidential motorcade and was guided by police cars to keep the group going along smoothly. In this case, the caravan moved at a much slower pace and the police did not keep all of the intersections open for the entire caravan. Nonetheless, it was another example of a series of cars that were aligned to a joint purpose of trying to drive loosely together.

At self-driving car conferences that I speak at, I’ve noticed that some people seem to inadvertently confuse the notion of car caravans with the notion of platooning. They are different aspects.

Platooning does include the idea that you have multiple vehicles trying to stay together while driving on a journey, and in that sense, it has apparent similarities to a caravan. With platooning, the goal is to have the cars or trucks be as close together as possible, high coupling of a virtual nature, doing so to potentially optimize fuel savings and reduce pollutant emissions from the vehicles. The vehicles draft off of each other and try to move in very tight unison with each other.

A car caravan usually has a much looser requirement of the caravan participants being so close to each other. Cars can vary their distances from each other, with some at times being close and at other times getting a bit distant of each other. It is low coupling.

The car caravan does not usually have as a goal the optimization of fuel or the reduction of emissions. Instead, it is more of a symbolic gesture of togetherness that the vehicles all travel together. In the case of an official caravan such as the Presidential motorcade, it can allow for better protection of the caravan participants and also serve a symbolic gesture too (somewhat like a parade of sorts). And, from a traffic perspective, it can be overall more optimal as to reaching their common end-point destination or even intermediary destinations along the way.

The lessons learned about how to best structure and operate a caravan can certainly transfer over to trying to best carryout a platooning operation. Likewise, aspects of doing platooning can be possibly reused when trying to manage a caravan. My main point is that caravans and platooning are not identical, and there are some distinctive differences. I’d say they are of the same species and thus each can glean insights from the other.

For platooning, there is usually heightened dangers and risks, since there are high-speeds desired and the vehicles are in tight formation, which means less time allowed to deal with any uncertainties and react to roadway conditions. If there is debris that can disrupt the platoon or one of the vehicles suffers a tire blowout, it can be dicey. Plus, there is a cascading impact that can cause harm to the entire set of vehicles involved in the platoon.

One must also consider that other cars that come near to the platoon are also at increased risk since those cars cannot maneuver as readily due to the platoon acting as a block. This also tends to reduce visibility for other cars that aren’t involved in the platoon. You need to also consider the ramifications of a car trying to be an interloper into a platoon and the consequent reactions that can cause wild actions or even collisions.

Some are suggesting that with V2V (vehicle-to-vehicle) electronic communications, the platoon can more likely coordinate their activities and also forewarn other V2V equipped cars about the ongoing presence of the platoon. Likewise, via V2I (vehicle-to-infrastructure) the platoon could alert the roadway infrastructure and in case the infrastructure could possibly aid the platoon such as a drawbridge that was raised and might be lowered in time for the arrival of the platoon.

In any case, platooning does have controversy associated with it. Though it seems like a no-brainer as a means of efficiently doing trucking, not everyone is convinced that the risks and qualms are worth the touted benefits.

According to recent comments made at CES in Las Vegas by Daimler’s head of trucks and buses, Freightliner’s Martin Daum, they have found that the fuel efficiencies are not sufficient when they factored in the added fuel consumed when a platoon has gotten semi-dispersed (the chain is broken, one might say) and the platoon members need to catch-up and reconnect together as a chain (see the CES coverage by the LA Times automotive industry specialist Russ Mitchell). As one of the world’s largest truckers, if Freightline is soured on platooning its a sure sign that platooning is not the silver bullet that some pundits proclaim.

I’ll cover more about platooning in a subsequent piece, so let’s herein focus on caravans.

Some Caravans Planned in Advance

Some caravans can be well-prepared in-advance and involve in-depth logistics for planning of the caravan.

I’m sure that the Presidential motorcade had included weeks or months of dialogue with local officials about the path to take through SoCal, along with coordinating with local police, highway patrol, city traffic engineers, etc. There were likely contingencies covered too. I noticed that the motorcade had ambulances and fire trucks that were part of the overall group. I’m assuming that if something had gone amiss while the motorcade was proceeding, those first responders were included as a just-in-case they might be needed right away.

Car caravans can also occur in a somewhat spontaneous fashion and be more ad hoc than formalized.

Allow me to share with you an example of an impromptu kind of car caravan.

I was at a meeting of a professional association that I am a member of, and all of a sudden someone suggested that we all go across town to a popular restaurant and continue our discussions there while catching dinner. Not everyone at the association meeting had driven to the event in their own cars and so we had a somewhat chaotic moment as people offered to give rides to others that also wanted to get to the restaurant. It took a few minutes of a bazaar-like trading effort to figure out who would go in which cars, and whether the cars so chosen could fit that number of people. It was also complicated due to the aspect that once the restaurant dinner was finished, we were trying to solve the secondary problem of whom would get driven home by whom, which added another layer of complexity to the impromptu matter.

Those of us driving were all in agreement as to the destination and the desire to get there in roughly the same amount of time. We all hoped to arrive at the restaurant at about the same time, thus no one would need to be waiting for anyone else to show-up. As you can imagine, there was some at times heated debate about which way was the fastest route to the restaurant. Some of the drivers thought that taking the freeway would be fastest, while others of the drivers tended to prefer using side streets since the freeway was likely to be clogged with commuter traffic.

After agreeing to go ahead and use the freeway (some grumbled about that, I assure you), we all got into our respective cars and started our informal car caravan on its way. Some of the drivers had obviously never tried to actually participate in a car caravan and they immediately zoomed ahead of everyone else, seemingly not having a care in the world about the rest of the cars in the caravan. Meanwhile, some of the drivers were admittedly the nervous Nellie kind of driver, and they began to fall way behind the rest of the pack.

Within just five to ten minutes of the caravan having gotten underway, we were no longer a “proper” caravan that could be considered of a line-of-sight nature. Some caravans try to keep all of the cars in a line-of-sight, meaning that each car that is a member of the caravan can see at least the car ahead of it that is also a part of the caravan. In our impromptu version, we had cars now that were stretched miles apart from each other and the caravan had become a somewhat disjointed and sporadic series of segments of cars.

Adding to the exasperation was the fact that when we each tried to get onto the freeway, by bad luck the specific on-ramp that we all were going to use was closed down due to an accident that had happened about an hour earlier. This completely disrupted the caravan. Some of the drivers opted to continue onward to the next freeway on-ramp and therefore kept to the approach of using the freeway. Other drivers decided that this was a sure sign that using side streets was the better idea (it certainly made them feel high-and-mighty about earlier stating that side streets would be best), and so those drivers decided to not use the freeway at all.

I guess it is a miracle that we all made it to the restaurant, though the timing of arrival was rather scattered. Fortunately, no one got lost and we all eventually arrived at the restaurant. Luckily, there wasn’t anyone that had a car problem like say a flat tire, and there wasn’t anyone that opted to abandon the quest (though I’m sure there were rather pointed discussions inside some of the cars about whether to just give up trying to get to the restaurant and instead call it a night).

A kind of funny thing happened during the caravan antics. I got a call on my cell phone from someone that had arrived late to the professional association meeting and he was perplexed as to why there was no one there. I explained that we had decided to adjourn the meeting and headed to the restaurant for dinner. He asked if it was Okay for him to head to the restaurant and he also mentioned that a few other late arrivers were also standing around trying to figure out what was going on. I told him they certainly were all welcome to come to the restaurant.

This then prompted another caravan!

I found out later on at dinner, once they arrived, they had done the same things we had done. They had first discussed whom was driving with whom. This involved some complicated deliberations as to car sizes and the number and sizes of the people involved. Though they had fewer participants, they nonetheless ended-up with several cars involved and it was a miniature version of our original caravan. They debated whether to use the freeway versus side streets. Once underway, their cars rather quickly lost line-of-sight of each other, just as had happened to us. They each also encountered the closed-off freeway on-ramp (I sheepishly realized that I probably should have mentioned that aspect). And so on.

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

At the Cybernetic AI Self-Driving Car Institute, we are developing AI software for self-driving cars. One so-called “edge” problem for self-driving cars involves their participation in a car caravan.

For my article about edge problems in AI self-driving cars, see:

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 auto makers are even removing the gas pedal, 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, 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 true Level 5 self-driving car. Much of the comments apply to the less than Level 5 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 the 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:

AI Self-Driving Cars in Car Caravans

Returning to the topic of car caravans, this notion of an AI self-driving car participating in a car caravan is considered by many auto makers and tech firms as an edge problem. An edge problem is a portion of a larger problem but an aspect that is considered at the edge or corner of what you are trying to solve. Right now, the auto makers and tech firms are focusing on getting an AI self-driving car to drive properly on our roadways. This is a priority focus devoted to having a single “independent” car doing driving and doing so without hitting or harming others. Dealing with packs of cars that are trying to caravan together is not a priority at this time.

We do care about car caravans and I offer herein some of the challenges and opportunities involved in adapting AI to be able to consider AI self-driving cars in a caravan mode.

First, there is the question of whether a human will instigate the utilization of a car caravan or whether the AI itself might do so.

In the case of the professional association meeting caravan, if we had AI self-driving cars, we could have presumably expressed to our AI self-driving cars that we wanted to get to the restaurant and that as much as possible we wanted to do so as a car caravan. In that case, assuming that the AI self-driving cars each had a module dealing with car caravans, the AI respective systems could have electronically communicated with each other and potentially determined the logistics of the caravan journey for us.

For conversing with an AI self-driving car to give driving commands, see my article:

For the socio-behavioral aspects of humans instructing AI self-driving cars, see my article:

For Machine Learning aspects about self-driving cars, see my article:

For more about how humans interact with AI self-driving cars, see my article:

The communication amongst the AI self-driving cars would likely occur via V2V (vehicle-to-vehicle communications). An AI self-driving car would attempt to open an electronic dialogue with another AI self-driving car and invoke the car caravan module. This specialized module would then take on the role of an overseer for figuring out the caravan details. This would be done in a federated manner and not require that there be one centralized system manager routine per se.

During the planning of the route to the desired destination, the AI self-driving cars could also make use of V2I (vehicle-to-infrastructure) electronic communications. AI self-driving cars are intended to operate in conjunction with “smart” roadway infrastructure, which might include that a bridge might electronically beam out a message that it is not passable or that a local street emits a signal that it is torn-up for road repair and should be avoided. In the case of my story about the professional meeting caravan, it is conceivable that if there was V2I that using it might have alerted us that the freeway on-ramp was closed. In which case, the argument we had about using the freeway versus using side streets might have been more easily resolved.

Once an AI self-driving car caravan gets underway, the AI self-driving cars can continue to keep tabs on each other by continuing to use V2V. If a lead self-driving car got to the freeway on-ramp first and discovered it was closed and hadn’t been alerted beforehand by any relevant V2I broadcasting about it, the lead self-driving car could alert the other AI’s of the self-driving caravan of cars. The AI’s would then in real-time rapidly deliberate about the rerouting of the caravan.

This form of AI self-driving cars sharing with each other about their surroundings during a driving journey is referred to as omnipresence.

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

The rerouting of a self-driving car caravan is a bit trickier than you might think. One aspect involves how close the self-driving cars in the caravan are supposed to stay with each other. This would be a parameter that could be set by those involved in the caravan. If you were doing a caravan with loved ones as part of a funeral procession, you might want the caravan to be very tightly woven together. If the caravan involves simply going to a restaurant for dinner, you might allow the AI to let the caravan stretch out.

For more about V2V and V2I, especially regarding 5G, see my article:

For AI self-driving cars having emergency vehicle awareness, see my article:

I had mentioned that the caravan of AI self-driving cars could be initiated by a human. Another way to initiate a car caravan might be via the AI itself.

AI Could Initiate the Caravan

Suppose that you are heading to work for your morning commute. You work at the ABC Building in downtown and live in a city that is about 20 miles away from the downtown area. If you are willing to share your destination with other AI self-driving cars, here’s what could happen.

Your AI self-driving car via V2V starts communicating with other AI self-driving cars. Specifically, the AI is trying to find out if there are other AI self-driving cars that are heading to the same downtown destination. If so, the AI of your self-driving car might suggest to other nearby AI self-driving cars that are also going to downtown that they proceed collectively as a car caravan. The AI self-driving cars would work in unison with each other, trying to generally aid each other in getting to the desired destination.

Please note that I am not saying that the AI self-driving cars would need to form a tightly woven pack. As mentioned earlier, a caravan does not necessarily need to consist of vehicles that are bumper to bumper with each other (that’s platooning). Instead, these AI self-driving cars in the virtual car caravan would arrange to at times to be close to one another and at other times perhaps be at some distance of each other. They could also be warning each other about road traffic status, street repairs, and other infrastructure related aspects that might or might not also be available via V2I.

It could be that every day, you go to work pretty much the same way, and as such the self-driving car caravan might become a learned and practiced effort for this particular route. The first time you engage it, the caravan might be an impromptu version. Later on, it might become a common place caravan that has established itself and provides value in being used over and again. This might include using a blockchain for purposes of keeping a sustainable version of the caravan aspects.

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

For my article about defensive driving tactics and AI self-driving cars, see:

For AI self-driving cars working together in swarms, see my article:

Overall, the approach is that a human can potentially initiate the use of a car caravan, or the AI itself can propose that a car caravan be used. This does bring up though some quite serious issues of privacy and other related concerns that need to be considered.

If your AI self-driving car starts broadcasting to other AI self-driving cars your destination, this could be a loss of privacy on your part, plus it could be a potential risk factor if you are someone that others might want to target. A variant then on the destination indication could be that rather than a specific address being communicated, the destination might be an overall location such as downtown. This would reduce somewhat the loss of privacy and the risk factors.

Another aspect to this car caravan is whether you as a human want to participate in it or not. If you initiate it, presumably you want to participate in it. If the AI initiates it, you might not want to be a participant, thus it would be crucial that the AI would let you know about the potential of joining a car caravan and allow you the choice of whether to join into it or not.

From a ridesharing perspective, the use of car caravans could be a substantive boost toward trying to optimize the number of people and the number of self-driving cars needed for particular journeys. Suppose there are other people in your neighborhood that need to also get to downtown for work each day. Rather than each of them taking a car to get there, the use of the car caravans might reveal a type of car pooling that could be undertaken.

There might be “private” car caravans and there might be “public” car caravans. In the case of the story about my colleagues going to dinner, we would have considered our car caravan to be a private one. Even if some other AI self-driving car made contact and requested to participate, we would have likely not wanted any other such cars to be in the caravan other than the ones we considered part of our group. In contrast, a public oriented car caravan might be open to any other AI self-driving car that wants to participate in the caravan effort.

Some AI self-driving cars might have a caravan module, and some might not. I say this because many people falsely seem to think that all AI systems for all of the different auto makers self-driving cars will be the same. They will not be. Each auto maker will have a distinct set of features available in their AI self-driving cars. Over time, we’ll likely see that most of the auto makers tend to gravitate toward having the same or similar features in their AI self-driving cars.

Even for an AI self-driving car that does not have the caravan module, if such a module is generally available and compatible with the rest of the AI system of that self-driving car, it could potentially be downloaded via the OTA (Over The Air) capability of the self-driving car. OTA is a feature that most auto makers are including in their AI self-driving cars and allows for the self-driving car to electronically communicate with the cloud of the auto maker or tech firm. Doing so allows the auto maker or tech firm to upload data from the self-driving car, along with being able to download and install new data and systems patches into the AI self-driving car. A module for being able to undertake car caravan processing could potentially be so installed via OTA.

For my article about OTA, see:

For potential concerns about privacy, see my article:

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

For aspects of responsibility and self-driving cars, see my article:

For family road trips with AI self-driving cars, see my article:

Car caravans can be used in a wide variety of settings. You might be heading on vacation with some friends, all of whom are driving their own cars, and you perhaps want to caravan together while visiting numerous wilderness sites across the country and other exciting destinations. Or, you might use a car caravan for getting to work or for undertaking dinners meetings with professional colleagues. Caravans also include the somber occasions of a funeral procession, along with the rather hectic motorcades of VIP’s.

AI self-driving cars can and should be able to participate in car caravans. If there is a mix of human driven cars and AI self-driving cars for a particular caravan instance, it will admittedly be harder to carry on the caravan. When the car caravan consists of solely AI self-driving cars, and assuming they are outfitted with the proper add-on module, the caravan can be a nearly seamless experience for any human occupants that are in the cars of the caravan.  I don’t want to though exclude human driven cars and so let’s not necessarily refer to this as self-driving car caravans and for the moment keep it to the broader notion of car caravans. Wouldn’t want to seem discriminatory towards human drivers.

Copyright 2018 Dr. Lance Eliot

This content is originally posted on AI Trends.