By Lance Eliot, the AI Trends Insider
There must be a better way, some lament.
It is taking too long, some say, and we need to try a different alternative.
What are those comments referring to?
They are referring to the efforts underway for the development of AI-based self-driving driverless autonomous cars.
There are currently billions upon billions of dollars being expended towards trying to design, develop, build, and field a true self-driving car.
For true self-driving cars, the AI drives the car entirely on its own without any human assistance during the driving task. These driverless cars are considered a Level 4 and Level 5, while a car that requires a human driver to co-share the driving effort is usually considered Level 2 and Level 3.
There is not as yet a true self-driving car at Level 5, which we don’t yet even know if this will be possible to achieve, and nor how long it will take to get there.
Meanwhile, the Level 4 efforts are gradually trying to get some traction by undergoing very narrow and selective public roadway trials, though there is controversy over whether this testing should be allowed per se (we are all life-or-death guinea pigs in an experiment taking place on our highways and byways, some point out).
So far, thousands of automotive engineers and AI developers have been toiling away at trying to invent a true self-driving car.
Earlier claims that progress would be fast and sweet have shown to be over-hyped and unattainable.
If you consider this to be a vexing problem, and if you have a smarmy person that you know, they might ponder the matter and offer a seemingly out-of-the-box proposition.
Here’s the bold idea: Rather than trying to build a self-driving car, why not instead just make a robot that can drive?
Well, by gosh, why didn’t somebody think of that already, you might be wondering.
The answer is that it has been considered, and indeed there are some efforts trying to construct such a robot, but overall the belief is that we’ll be more likely to sooner achieve self-driving cars via building driverless cars rather than trying to craft robots that can do the driving for us.
For my framework explaining the nature of AI autonomous cars, see the link here: https://aitrends.com/ai-insider/framework-ai-self-driving-driverless-cars-big-picture/
For my indication about how achieving self-driving cars is akin to a moonshot, see this link: https://aitrends.com/ai-insider/self-driving-car-mother-ai-projects-moonshot/
It is useful to consider a Linear Non-Threshold (LNT) model in the case of autonomous cars, see this link: https://www.aitrends.com/ai-insider/linear-no-threshold-lnt-and-the-lives-saved-lost-debate-of-ai-self-driving-cars/
A grand convergence of technologies is enabling the possibility of true self-driving cars, see my explanation: https://aitrends.com/ai-insider/grand-convergence-explains-rise-self-driving-cars/
Many Benefits Of A Driving Robot
Imagine if we had robots, the walking and talking type, and they could drive a car.
These are some of the benefits we’d derive:
- Conventional Cars Become Self-Driving. If you could have a robot sit in the driver’s seat of any conventional car and be able to drive the vehicle, you’d be able to turn any and all conventional cars into being “self-driving” (well, they wouldn’t need a human driver). That would be a huge plus. Right now, conventional cars generally need to be redesigned and built anew to be self-driving, leaving in the U.S. alone the 250 million conventional cars out-of-the-loop and ultimately headed to the scrapyard if people decide they’d rather get themselves a self-driving car (once such driverless cars arrive).
- Easily Switch Self-Driving From Car-To-Car. Presumably, you could have a robot driver that would readily be switched from driving one car to the next day driving a completely different car, merely by walking or carrying the robot to the driver’s seat of the other car. With the design of the emerging self-driving cars, everything is built into the specific car and you can’t somehow share it to suddenly make another car become self-driving.
- The Driver Would Be Seen. One of the qualms some have about the emerging self-driving cars is that there isn’t a driver in the driver’s seat, which is kind of eerie and worrisome since we are used to seeing someone sitting in that crucial position. The driver sitting there is reassuring in the sense that you know whether the car can possibly be driven, plus the driver can move their head and make eye contact to convey their driving intentions. Self-driving cars might have some kind of LED’s or other displays to do similar signaling, but a robot with a robotic head would be even more familiar to us.
- Could Use V2V, V2I, Etc. Self-driving cars are being outfitted with V2V (vehicle-to-vehicle) electronic communications, allowing the AI of nearby cars to communicate with each other, and there will also be V2I (vehicle-to-infrastructure) involving the roadway signs and structures to electronically interact with driverless cars. If a human driver wanted to do V2V and V2I, it would be problematic since we humans aren’t geared for direct electronic communications, but a robot driver would readily be able to do so.
- Adjustable As Cars Advance. Semi-autonomous cars are increasingly being loaded with ADAS (Advanced Driver-Assistance Systems), enabling automation to do more of the co-sharing driving effort with human drivers. In theory, a robot driver that was properly designed could easily be adjusted or updated via OTA (Over-The-Air) electronic downloads to be able to accommodate whatever new advances occur in the Level 2 and Level 3 cars. Your robot driver would then ascertain how much of the driving it will do versus how much it would let the ADAS do.
- Invest In A Robot, Not In A Car. There is an ongoing debate about the ownership of true self-driving cars, whereby some believe that only large corporations will own driverless cars and proffer them in fleets for ridesharing purposes. I am a contrarian and claim that we’ll still have individuals owning such cars, but in any case, if you have a robot driver then potentially you wouldn’t need to invest in a car per se. You might instead buy a robot driver and use it whenever you need to go for a ride, perhaps borrowing a friend’s car, or using a ridesharing car that isn’t yet driverless, and so on.
- Reusable For Other Purposes. A true self-driving car has pretty much one purpose, it is a car and used for transportation. A robot driver could be designed and built for doing lots of things, more so than merely driving a car. One of the worrisome issues about self-driving cars is that the “last mile” of doing actions like say delivering a package to someone’s door is not feasible for the driverless car to do by itself. Potentially, a robot driver could get out of the car, on its own, and walk up to the door to deliver a package (these kinds of walking delivery robots are being built and tested today).
I’ve listed some of the handy advantages of pursuing a robot driver.
There are more aspects that arise as benefits, but let’s not ignore the other side of the coin, namely the potential disadvantages or drawbacks.
No free lunch when it comes to the robot driver idea.
Downsides of Driving Robots
A robot driver is not necessarily a cure-all.
Consider this list of downsides about robot drivers:
- Might Be Prone To Disrepair. For a self-driving car, the guts of the tech are hidden within the car and hopefully going to work reliably. A robot driver that you are pulling into or out of a car is bound to get a lot of wear-and-tear. Do you really want a robot driver that is maybe in a state of disrepair driving your car? Don’t think so.
- Forces Cars To Remain As Designed Today. Some believe that self-driving cars will have an entirely new interior and allow for human passengers to sleep, play, or work inside a car. This is partially possible due to the aspect that you can remove the driver’s seat entirely, which is a fixed-in-place position that today that limits whatever else you might want to do with the design of the car interior. A robot driver would need to sit where today’s human drivers sit; therefore, the car interior would still be burdened with a driver’s seat.
- That Frightening Feeling. A robot sitting in the driver’s seat is going to be quite chilling to see. Movie after movie has forewarned us of the day that robots take over our world. Even though having no visible driver might be eerie for truly self-driving cars, I’d bet that having a robot driving our cars will make people really bug out. Does society have the stomach for this or will humans’ rebel once they see robots driving around town.
- Hackers Delight. There is worry that self-driving cars might get hacked, perhaps an evil programmer might plant a computer virus via using the OTA of the driverless car. It would seem perhaps even likelier that hacks would happen to a robot driver, more so than for true self-driving cars. The hacker can readily get ahold of a robot driver and try endlessly to crack into it. Doing the same for self-driving cars is going to be harder (though not impossible).
- Robot Maintenance Looms. When a car has troubles, you usually take it to an auto repair shop or a dealership. If your robot driver is having trouble, maybe the arms aren’t working right or the robotic feet are slow to respond, where will you take it? We aren’t yet ready to deal with thousands upon thousands of robots that need maintenance (maybe millions of them!), along with the repairs and replacement parts needed. It would be a significant undertaking to put in place such an infrastructure for robot driver upkeep.
- Safety Limitation Worries. If you assume that the robot head is going to be somewhat akin to a human head, presumably the robot would have cameras as eyes and the visual component of the robot driver would be the mainstay of how it drives. Self-driving cars are being outfitted with cameras, along with radar, ultrasonic sensors, thermal imaging, LIDAR, and so on. Some believe that those other devices aren’t needed since humans only use their eyes (mainly) to drive, but no one can say for sure that a robot driver using only visual elements could drive as well as a human. It might be worse.
- Other Concerns. Maybe the robot driver weighs several hundred pounds, in which case it is not going to be so easy to switch it from car-to-car. The robot driver would presumably wear a seat belt, but how much movement would it be prepared to have? Might different seat belts and different kinds of driver’s seats impact its ability to drive? Suppose the car takes a tight turn, will the robot driver stay in proper position to seamlessly continue driving the car? Lots of questions arise.
Those are just some of the issues that ensue when you consider the robot driver approach.
For my warnings about cyber-security and hacking of AI, see this link here: https://aitrends.com/ai-insider/ai-deep-learning-backdoor-security-holes-self-driving-cars-detection-prevention/
For more about safety and autonomous cars, see my coverage here: https://www.aitrends.com/ai-insider/chief-safety-officers-needed-in-ai-the-case-of-ai-self-driving-cars/
For my comments about how software neglect is going to raise risks of AI autonomous cars, see this link: https://aitrends.com/ai-insider/software-neglect-will-impede-ai-self-driving-cars/
For an indication about bifurcating the levels of self-driving, see my indication here: https://aitrends.com/ai-insider/reframing-ai-levels-for-self-driving-cars-bifurcation-of-autonomy/
The consensus among self-driving car aficionados is that a robot driver is a long way away from being practical. A robot driver is considered generally to be more futuristic than trying to develop a self-driving car instead.
I’m not saying that we’ll never have robot drivers.
There are some companies working on them today, along with research taking place in universities and labs.
Yet, one supposes that if we do achieve true self-driving cars, there might not be much of a need or value in having robot drivers, in which case we’ll likely see robots that do other things but aren’t particularly versed at driving.
Of course, as mentioned earlier, robot drivers might take up the slack of being able to drive conventional cars, until eventually and presumably all conventional cars are weaned out of the stock of cars and self-driving cars become the nature of all cars.
You could also portray this as a moonshot-like race between the self-driving car makers and the robot driver makers.
If you could perfect robot drivers sooner than the perfection of self-driving cars, it would obviously put robot drivers into, well, the driver’s seat.
Suppose too that true self-driving cars turned out to be impossible or infeasible, maybe the robot driver would provide an alternative that could be feasible.
No one knows.
Which then are you cheering for, the advent of self-driving cars or the emergence of robot drivers?
You might as well tell your smarmy friend that the notion of robot drivers is already underway, thus if the friend is irritatingly smarmy it’s time to come up with another idea for solving the autonomous car problem.
Good luck with that.
Copyright 2020 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: https://forbes.com/sites/lanceeliot/]