A recent video online shows a humanoid robot sorting packages, mostly just flipping them label side up so they can be scanned. It was painful to watch because the conveyor was slowed down to at least half normal speed and the "robot" was horrifically inefficient at its job.
Irony Alert: Before I could play the linked video above, it asked me to "sign in to prove you're not a bot." Rules for me, but not for thee, I guess.
Warehouse robots exist, but they don't look like ersatz humans. Rather, they are machines optimized for handling, sorting, and stacking packages and retreiving and packing goods. They look nothing like the "row-bots" of science fiction because doing so would be pointless. Why give a package sorting robot legs and a torso and a head and human-like arms, when history has shown that human anatomy is ill-suited to repetitive motions. Do you want these robots to have tendon and joint problems? Backaches and arthritis?
But seriously, a humanoid robot might have some specialized uses, particularly when interfacing directly with humans, but for other purposes, it makes no sense. Imagine an automated car wash that, instead of having rotating brushes and spray wands, had dozens of humanoid robots sloshing around in the suds, hand-wiping your car. Maybe there is a kernel of an idea there, but why would you give them legs to walk around with, when stationary or wheeled robots would make more sense. And why make them look like humans?
I suppose the idea of a "general purpose" humanoid robot might make some sense. It could mow your lawn, wax your car, and then shake up a martini for you at five o'clock. That seems to be a common fantasy with robot fans - that "Jeeves" the family robot will mix up that martini and light your cigar (and empty the ashtray) while you sit in your smoking jacket in the den and listen to Jazz music.
But as history has shown, it is a lot more effective to make a number of domestic robots, each specialized to a purpose. Having a humanoid robot operate a vacuum cleaner is an expensive waste of machinery. Having an automated vacuum cleaner makes more sense, as does an automated lawn mower.
With regard to the latter, such things exist, but never became popular. I recall a few years back seeing a display at Lowe's of a robotic lawn mower. The display was broken. A neighbor actually bought one and I guess it worked for a while. One day I saw them out on the lawn, looking at the inert machine and then poking it with a stick. It sat there, motionless, for about a week and then disappeared never to be seen again. It never pays to be an early adapter.
The problem with the automated lawn mower is obvious - suppose your pet is run over by it? Or maybe a crawling infant? The liability is huge and no matter how many sensors you put on it, it may be likely to fail. For that reason, most automated lawn mowers used string trimmers, not whirling blades, so little junior won't be sliced to death, just slowly whipped to oblivion. But I digress.
Another problem with humanoid robots is the power they require. Humans are a fascinating machine - converting calories of food into energy every day for nearly a century before they wear out. It is just a shame it takes 25 years or so to program their neural networks, and often that goes horribly wrong.
Companies like Boston Dynamics like to show off their robots dancing to music and doing amazing flips and stunts. In most cases, they are tethered to a power cord, or in some cases, an IC engine powered electrical generator. The actual ambient sound is rarely shown, as instead of music, you would hear the scream of a lawn mower engine or the sound of electrical or hydraulic actuators. Lithium-Ion batteries are one solution, but of course add weight to the design and have a limited capacity. I suspect many of these short videos are short because that is the extent of the battery life.
While Boston Dynamics claims that their robots are for sale, at the present time they are limited to enterprise customers and there is little or no word as to how many have been sold or are in use. Have you seen one being used? Let me know.
Meanwhile, other more plebeian robots are already out on the streets - quite literally. No doubt you've seen the food delivery robots on some city streets. But again, they are not humanoid, but small boxes on wheels. A humanoid delivery robot could do things like climb stairs and get around obstacles better, but the energy required would no doubt limit their range and the cost would be prohibitive. Boxes on wheels are cheap - and get the job done.
And of course, robot taxis are a thing, at least on a trial basis in some cities - with mixed results. When error-prone humans run over a small child, we chalk that up to the cost of doing business - and the driver gets a slap on the wrist. If you want to get away with killing someone, use your car. Ask Bruce Caitlyn Jenner. She got away with it!
Sam's Club and Walmart have a floor cleaning robot that seems to work "OK" - although it has a seat and manual controls. Not only does it wash floors, but it also takes inventory with a side-mounted scanner. Again, it is on wheels, not legs. No one seriously suggested using a humanoid robot with a mop-and-bucket - it would make no sense whatsoever.
Getting back to UPS and other shipping companies, when I worked there as a nascent Teamster in the 1980s, we were basically human robots. Bar codes were a thing of the future, and boxes came down a chute in pre-sort where four or five guys would sort each package by zip code. These guys had the entire zip code structure of the greater Syracuse area memorized and I could not keep up at all. I lasted a day in pre-sort.
The solution wasn't robots to "read" hand-written addresses and sort by zip code, but to put standardized labels and bar codes on each package. Back in 1986, few had home computers and printers. Today everyone does, and if they don't, the local UPS store will print the label for you. Automation didn't mean creating mechanical men to mimic manual human labor, but to automate the process instead.
So bar codes are read by lasers and mechanical flaps divert packages to different conveyor belts - a much more efficient and cost-effective process than having humanoid robots pick up packages and look at them. Humanoid robots are not an advancement, but a step backwards.
Similarly, when I worked at Domino's, we took orders over the phone and then hand-addressed each box of pizza using carbon-copy strips. One went to the line cook to make the pizza (and make a record of the transaction), while two went on the box. The driver would tear off the pink copy upon delivery. It was a primitive system, but it worked. I told my boss there that someday a computer would print out these labels and he laughed at me. At the time, I had to deal with his shitty handwriting, which almost got me mugged.
Now, I suppose you could use a humanoid robot to pick up the receiver on a landline phone and take orders and write them down on slips of paper, but that would be dumb. Instead of automating human actions, they automated the process. Customers order online using an app and the pizzas are made after the labels were printed automatically. Addresses can be verified (and customers tracked for loyalty rewards) and phone numbers confirmed through automated call-backs. Automating the process makes much more sense than using a humanoid robot to mimic what was done before by humans.
In warfare, the same is true. Much has been made of humanoid "soldiers" replacing men in the field. A bipedal robot can walk through ditches and climb over fences. But what we are seeing in Ukraine is that flying robots (actually, remote controlled drones) are far more effective, efficient, and less costly than some clanking monstrosity from Boston Dynamics. Seaborne and landborne drones also exist, the latter tending to be of the wheeled variety.
The idea of autonomous mechanical warriors, however, should give everyone pause. Some drones are deemed "semi-autonomous" in that they can continue their mission even if cut off from their operator by jamming signals. Even these are controversial, as without a human-in-the-loop, it would be all-too-easy to bomb a school or create a friendly fire situation.
But again, with drones, the idea of a walking humanoid fighting drone seems dead in the water, at least for now. Masses of cheap and lightweight drones - often "suicide drones" - makes more sense than an expensive walking-talking doll.
Perhaps I am wrong about this, and instead, in a decade's time, my personal robot will clear the dishes after cooking my meal, mow the lawn and drive my car. I kind of doubt it, only because I can't afford such nonsense. I also wonder whether humanoid robots are really solving a problem that exists - is it really cost-effective to replace low-wage jobs with robots? Or are we just chasing a tech dream?'
Already we are seeing incidents where "AI" ends up being more costly than replacing human workers. And as a "tool" it seems highly flawed. At least from a consumer standpoint, it is of limited use - and if you rely on AI answers to basic questions, you may end up in trouble.
For example, our van has a problem in that the switch or sensor that detects whether the sliding door is open, isn't working. Not only doesn't the dome light come on (grammar?) but the "auto relock" feature kicks in, if you unlock the van with the remote and open the sliding door. The system detects if a door has been opened after unlocking. If no detection occurs within 30 seconds, it assumed you hit the button in error and re-locks the van. Since the sliding door sensor is broken, it doesn't detect this as an opening and re-locks the van. No big deal, unless you set the keys on the counter and then re-shut the door. Congratulations, you just locked the keys in the van!
And no, there is no way to turn off auto re-lock. :(
Anyway, Google AI keeps cheerfully informing me that a 2015 Mercedes Sprinter 3500 high-top has a simple plunger switch "on the B-pillar" and it costs only $4.95 on eBay. No, it does not. Mercedes chose instead to put a switch on the door latches (all doors) including the sliding door. It is a PITA to get at and guess what? It costs a lot more than $4.95!
I've tried rephrasing the question (the "prompt" I guess) many times, with the same result. The information provided is clearly wrong, every time. And sadly, this seems to be true of many technical questions, regarding cars, computers, or whatever. AI merely skims the surface and provides the most facile responses. Did you try unplugging it and then plugging it back in? Gee, thanks, Sanjay!
I guess I take some comfort in this - that our robot overlords are as clueless as we are (indeed, we "trained" them, no?). In the future, there will be robot counterparts of all the inept and clueless people in the world, including robot Karens and robot BMW drivers who refuse to use turn signals.
Or, God forbid, Robo-Trump. Maybe we are already there - after all, he died of rabies!
