The Honda Element is a small SUV made by Honda. It’s designed to be practical and easy to use, with lots of usable space inside. It can be mentioned in self-driving discussions because it’s a normal, real vehicle that helps illustrate how automation works in everyday driving.
A self-driving car is a car that uses sensors and computer software to drive itself. It’s great at common driving situations, but it can struggle with weird or unexpected tasks.
A tow truck is the vehicle that transports cars that can’t drive themselves. Here it’s used as an example of a situation the self-driving system may not understand.
Sensors are the car’s “eyes and ears” that gather information about what’s around it. If they’re not working, the car’s computer can’t understand the scene well enough to drive.
This phrase means the car is mainly designed to get you from where you start to where you want to go. It may not be designed for special situations like being loaded onto a flatbed.
The Ford Edge is a mid-size SUV made by Ford. It’s built for regular driving and usually includes safety and driver-help features. It may be brought up in self-driving talks because automated systems have to handle not just easy situations, but also the tricky ones.
An edge case is an unusual situation the computer doesn’t see very often. If the car wasn’t trained for that exact scenario, it may not know what to do.
Sometimes the car’s computer can’t clearly identify what it’s looking at. It may see the tow setup as a generic blockage, so it won’t try to drive forward.
The car may decide it’s not safe or not understood, so it won’t attempt the maneuver. It’s not a human-like refusal—more like the computer saying “I can’t handle this.”
An obstruction detector is a safety sensor that stops a machine if it senses something blocking it. The point is that the AI can behave similarly—stopping because it detects a hazard.
OTA updates are wireless software updates. Instead of bringing the car to a shop, the car can receive new instructions through a cellular/Wi‑Fi connection.
A remote operator is a person who can control the car from far away. It’s like a human “backup driver,” but it can create extra safety and security concerns.
This means the car can be controlled from outside by someone with access. The worry is that if access isn’t tightly protected, the system could be misused.
LIVE
Hi, I'm Dr. Lance Elliott, and welcome to my podcast series about self-driving cars.
In this episode, I'll be discussing the topic of stubbornness in self-driving cars.
If you'd become interested in learning more about self-driving cars, please see my website
www.ai-self-driving-cars.gooroo for further information.
Okay, let's get started. You likely know the famous catchphrase of being as stubborn as a mule.
We all know people that are altogether stubborn and exercise mule-like behavior. Turns out,
we oftentimes consider things to be stubborn-like too. For example, a car can seem to be just about
as stubborn as a mule, perhaps being downright bullheaded. Here's an example of something I
witnessed the other day. A tow truck was getting ready to take a car for a tow. It was a flatbed
style tow truck consisting of a flat rear raised area that serves to hold and essentially piggy
back a car to be transported. The tail end of the flatbed portion tilts at an angle to allow for
driving a car up onto the riding platform. This forms a ramp for the car to traverse upward
into the empty and awaiting flatbed area. The driver began to inch up the ramp and had a very
hard time doing so. You could assert that the car was stubborn and did not want to go up that ramp.
I doubt any of us would seriously, though, contend that the car knew that it was going up a ramp.
All that happened was that the physics were working against the cars at trying to climb the ramp.
You might say that the stubbornness was actually rooted in the car driver.
Since we're discussing the topic of cars, the future of cars and tail self-driving cars,
the stubbornness element in the flatbed truck tail brings up an interesting facet about self-driving
cars. First, be aware that true self-driving cars are driven by an AI-based driving system,
and not by a human driver. Thus, in the case of this flatbed truck scenario, if the car had
been a self-driving car, the AI driving system would have been trying to get the car up that ramp
and onto the flatbed. Secondly, there are going to be instances wherein a human wants a self-driving
car to go someplace, but the AI driving system will refuse to do so. Now, I want to clarify,
the AI is not somehow sentient since the type of AI being devised today is not in any manner
whatsoever approaching sentence. Today's intriguing question then gets us to this.
How can you convince an AI-based self-driving car to go where you wanted to go when it won't go there?
We'll use the loading of the car onto the flatbed as a handy example.
Invasion envision a self-driving car that is supposed to get itself onto the flatbed of
a tow truck. We'll assume that the self-driving car is fully operational. For example, if the
sensors needed for driving weren't working properly, the car isn't likely going to be able to drive
itself up the ramp. Okay, you might be tempted to assume that the AI driving system will simply
zip up that ramp and park itself onto the flatbed. Indeed, perhaps the AI driving system
could undertake this flatbed tow truck driving challenge and maybe even better than a car driver
could. In theory, if the AI driving system has been programmed for this kind of driving task,
one supposes it could do a job as good as a human potentially could do. The problem is that the AI
driving system probably has not been programmed for this specific kind of driving task. Keep in
mind that the automakers are focusing their energies on getting AI driving systems to drive a car
safely from point A to point B, such as going from someone's home to the grocery store.
Programming an AI driving system to climb up a ramp onto a flatbed is not very high on the list
of top priority efforts. At best, it's an edge or corner case. Unless an AI driving system has been
programmed or trained to cope with the tow truck matter, the odds are the AI driving system would
be ill-prepared to drive up that ramp. In addition, the sensors would likely be detecting the tow
truck and the flatbed and not be able to discern what it's all about. In short, the tow truck
and its elements would appear to be a blob that is simply blocking the path forward. The AI driving
system would likely balk at being commanded to drive ahead. You could say that the AI driving
system was refusing to try and drive up the ramp. Again, though, this is not a conscious effort.
This is a refusal in the same manner that you might have a garage door with a built-in obstruction
detector that won't finish closing the garage door until that object is out of the way.
How can you convince an AI driving system to get up that ramp? Well, you could try to trick it.
Perhaps mask the appearance of the tow truck and its ramp, make it look like a normal road is ahead
of the self-driving car. But this is a very bad idea. Another approach would be to contact the
fleet operator. Here's what that means. Unlike conventional cars, it's anticipated that self-driving
cars would be part of a fleet. The fleet operator will be able to use OTA over the air electronic
communications to download software updates into the onboard AI driving system. It could be that
the AI driving system has a special software-enabled mode or added software component that could be
devised to cope with this kind of tow truck flatbed driving scenario. Another thought you
might have would be to ask a remote operator of the fleet to take over the driving controls
and manually drive the self-driving car up that ramp, though this does open a entire can of worms.
Some self-driving cars are going to be outfitted with remote accessible driving controls,
and others will not. Those that are opposed to remote accessible driving controls point out
that such a capacity could be used by bad actors in rather insidious ways.
A self-driving car is not a mule. That being said, you could assert that a times an AI driving
system will be stubborn in terms of refusing to drive where you want it to go. The beauty
of human driving is that a human driver can pretty much drive however they want,
meaning they can drive in places that you wouldn't normally think a car could go.
Perhaps someday AI systems will be programmed well enough to navigate all kinds of driving
situations, even ones that perhaps it hasn't even seen before. Until then, it won't make
sense to get mad at the mule. Definitely don't kick one, since it will only hurt your own foot,
and won't especially bother the AI. Thanks for listening, and again, I'm Dr. Lance Elliott.
I hope that you found today's episode informative. If you're interested in learning more about
self-driving cars, see my website, www.ai-self-driving-cars.google, for further information.
About this episode
A tow-truck ramp becomes a stand-in for why self-driving cars can seem “stubborn.” The host notes that AI systems are typically focused on safe point A to point B driving, so unusual tasks—like climbing onto a flatbed—may be treated as an obstacle or an edge case the system isn’t trained for. When the AI won’t comply, fleet operators may rely on OTA updates or remote operator takeover, though critics warn about security risks.
