Posted on 05/26/2014 3:15:22 PM PDT by Kaslin
In August of 2013 I wrote Message to 5.7 Million Truck Drivers "No Drivers Needed" Your Job is About to Vanish.
The key word in that sentence is "about". I did not mean immediately, but I did mean a lot sooner than truck drivers and the general public expect. Most protested. I received many emails saying this would not happen for decades.
Many truck drivers thought it would never happen. Most mentioned insurance issues. Yes, there are problems, but time has marched on even quicker than I thought.
TechCrunch reports California Will Start Granting Licenses For Driverless Cars In September.
Come September, the California Department of Motor Vehicles will begin granting licenses to select driverless cars and their human co-pilots, which will make it a bit less legally iffy as to whether or not they’re actually allowed to be on a public road.
The good news: The license will only cost $150 a pop, and that covers 10 vehicles and up to 20 test drivers.
The bad (but probably actually good) news: You probably can’t get one, so don’t go trying to make your own Googlecar just yet.
Stiff License Terms
Yes, the terms of the license are stiff including $5,000,000 insurance against personal injury, death, or property damage. And a test driver has to be able to take immediate control of the car at all times.
Nonetheless, the licensing is a big step forward. Totally driverless cars are but a single step away. All that needs to happen is for California to eliminate the requirement that someone has to be in the car at all times to take control.
A big issue is that radar can detect size and shape of objects, but it does not have human judgement regarding danger. For example, a balloon blowing across the road is a much different thing from a hunk of metal the same size sitting in the road.
Such difficulties will be overcome.
Incentives and Implications
The implications on the shipping business are staggering. A full-time truck driver might cost as much as $100,000 a year. The incentive to get rid of millions of full-time drivers is massive.
A July 2013 Truckers Report headline reads ATA: Self-Driving Trucks Are “Close To Inevitable”
However, the article itself dismissed the idea totally.
“People come up with these grandiose ideas,” says Bob Esler, a commercial trucker for almost 50 years. “How are you going to get the truck into a dock or fuel it?”
And then there’s loading and unloading. Pre-trip inspections. Signing for drop-offs and pickups. Making sure cargo is properly secured. Making sure the cargo that’s being loaded actually gets loaded. The list just keeps going on and on.
The Last Mile
Many of the objections in the above article have to do with the last mile. Let's assume someone has to load the truck. Let's also assume an actual skilled driver has to dock the truck and make the final delivery (arguably a bad assumption).
Yet, even if those assumptions are true, nothing stops a trucking company from having distribution facilities right off an interstate near major cities, where local drivers deliver the goods the last mile.
Why can't all but the last few miles be driverless even if a skilled driver is needed some step of the way for safety reasons?
Technology marches on at a breathtaking pace. We might actually see commercial driverless vehicles on the roads within a few years.
Humans are better at analyzing the situation and predicting problems. For example, if you see a car that is barely able to stay within the lane, and this happens not just once but for the last thirty minutes, chances are that the driver is drunk or falling asleep. A computer may not make that connection, and it will not assign a "red flag" to that car. When that car finally ends up in your way, the robot will only be able to react. Perhaps it will react fast enough - but an attentive human would not be driving near the suspicious vehicle in the first place.
I do not expect to see a complete, functioning AI as a car's driver. However it may be possible to add software modules to mathematically analyze behavior of other drivers. That weaving within the lane, or leaving it, may be measured and noted. The same would apply to poor speed control, tailgating, frequent lane changes, and so on - these are not complicated facts. Then different vehicles may be assigned different danger levels, and the algorithm will be slightly influenced to stay away from those.
These cars are not preprogrammed; they look at signs and road markings. This is necessary to drive in construction zones and on new roads. As result, these cars always obey the law. As an owner, you will only select the destination - and, if desired, the route. (The car may know which route is shorter, but it may not know or care that the green arrow in one specific place is on only for ten seconds every five minutes. But you do care about that.)
With so many idle, where are the taxes going to come from? Who is going to pay those?
I really do not see a significant further reduction in workforce without starving millions. The government can't keep creating "money", taking it out of one pocket and putting it in the other, less the overhead for that. Someone has to create wealth, and if not the people at the bottom of the economic food chain, where does the wealth come from?
If you tax the ones who will just add the tax to the price of goods, to feed the multitude, you create a spiral of increasing prices to cover increasing taxes to pay for feeding the increasing idle multitude, which means higher taxes, etc. A vicious cycle.
Nope, there will always be a need for someone to feed the machine.
When you come up with a computer that can chain up, let us know...
They can drop a tail hook. They already handle things on trucks like ABS breaking. An automated system to deploy the equivalent of chains should not take Einstein.
When driving defensively, one of the first things you learn is to never assume the best case scenario. You see a corner with restricted visibility, you slow down, cover the controls, just in case some idiot comes barreling through the intersection. You live longer because you can recognize what you do not know.
How will you be able to program a machine to take into account the absence of data in such situations?
At least on a motorcycle I have the advantage when it comes to maneuverability, horsepower to weight ratio, and braking efficiency, as well as the ability to judge tractive efficiency by noting the road surface, on everything form dry pavement to wet leaves, mud, and pea gravel--and drive accordingly, without having to nail the brakes to assess that. When you start discussing a semi, those factors change considerably, and without some means to assess them, the computer will likely respond more quickly, but improperly to prevent a wreck.
There is something a computer doesn't have that I do.
The computer does not feel pain, it does not concern itself with dying and leaving a family without its main breadwinner, and is incapable of caring about damage to the vehicle, another vehicle, a human, the cargo, or anything else.
It is incapable of assessing priorities when presented with a situation where it can hit that 45 pound kid in the road or a solid fixed object like a bridge abutment, and even if it did a predicted damage assessment, would likely run over the kid.
There are decisions made by humans behind the wheel that I don't think you can program for.
What would you call “the equivalent of chains”? Keep in mind I live in North Dakota and work in the oilfield.
Oh come on. The computer can monitor the behavior of its surroundings thousands of times a minute. It can easily be programmed to detect erratic driving. It’s reaction time is a fraction of a humans. It doesn’t get bored or distracted or sleepy. I have zero doubt that before too much longer we will have automated trucks, cars and buses.
“...and the truck will use the GPS of and supplied on and around a given property.”
So what’s the course of action during solar flares when GPS is unavailable or skewed hundreds of feet?
Something like this....
Something similar could easily be designed to self deploy.
Most of the roads leading in and out of oil drilling/production locations in these parts are surfaced with crushed natural brick, locally called "scoria". Essentially, this material makes for some of the finest gravel roads I have driven, but it's still a gravel road, and the heavier the vehicle, the more likely it (the road surface) will be damaged.
Between cold, foreign objects, and the variety of road surfaces, whoever designs a self-deploying system to maintain traction will have their work cut out for them. Reliability will be an issue, and by the time a vehicle is in trouble, it will be many miles from the nearest town.
Good. Make my automated car a flying car that runs on the free electricity from the nuke plant.
Precisely.
If the country wasn’t full of Idiot Greens and Luddites we would have a deal.
I have had numerous experiences with roadways where the speed limit is not accurately marked, or is confusing, where it switches back and forth or where it is not marked at all, and you have to know the default speed. I assumed that the car would be able to detect signs, traffic lights, etc. But that is not always enough to escape the attention of speed traps.
If they’re anything like when we lane switched here in Tucson it’s all weekdays. One of my “hobbies” was to use the suicide lane on Thankgiving, giving people the horn for being in the wrong lane. But again, holiday schedules are easily handled, DOT schedules are generally pretty easy to find.
You’ve got another piece of human error there, not computer error. All the car computer would need is stuff it already has access to thanks to GPS. It knows the time and date, it knows the map, the map data includes the schedule, all it’s gotta do is put them together. Easy peasy, no human input necessary.
Sorry but you’re wrong. The accident avoidance systems have a LOT to do with driverless cars. It’s the vehicle detecting the situation AND controlling the vehicle when the driver screws up. They ALL allow the driver to close his eyes, none of them are looking at him, they’re looking at the world getting ready to override him. Truthfully they’re built under the assumption he’s got his eyes closed.
Sure is. Because in the end the “boundless” stupidity of other drivers falls into a handful of predictable categories: cutting you off, lane impinging, wrong way, tailgating, a few others. And all of these categories have a correct method to avoid collision. Really it’s pretty straightforward. I see fewer dead people, because every computer controlled car is one less boundlessly stupid human on the road making hazards.
As long as they come with an inflatable auto-pilot like in the movie Airplane I’m okay with it.
Well hopefully, you’d be smart enough to program it to pull into a rest area, and say “Wakey wakey!”
“Easy peasy, no human input necessary”
I have to say, you have a much more simplified view of the world, and engineering problems, than I do.
That’s important to note, because, well....I am an engineer. Degreed, licensed, work in the field every day.
I’ve got some bad news for you: There will always be human input in a situation as simple as reversible lanes. No database is going to continuously be re-populated with lane switch information, in real time, to account for changes due to weather, the football game went into overtime, or the rock concert was cancelled.
I am reminded of a product I once had, called the ‘Road Wizard’. It pre-dated the commercial internet, and was a handheld device that would tell you where hotels and gas stations were, based on the interstate and mile marker you entered in. Of course gas stations and hotels change, so you were supposed to mail it in every year, to get re-loaded with new information. A driverless car system would have to get re-loaded on a continuous basis, in real time, even in far off places with no cell coverage, and during bad weather when GPS doesn’t work well. Again, just one of many obstacles.
I will repeat - there is a vast difference between technically possible, and repeatable with an extraordinarily high rate of reliability.
And I believe you are over-estimating the capabilities of accident avoidance systems and/or under-estimating what is truly needed to control a driverless car. No, you can’t close your eyes for 10 seconds for any of the automated features out there. Even something as simple as the parking feature - a driver has to look in his mirror, identify the traffic approaching him from behind, and make a judgment call on whether or not there is enough time to safely park. And then the driver has to watch out and make sure nothing happens with the adjacent cars (I.E. if he see’s the brake lights and reverse bulbs come on, he starts hammering the horn). Again, just one small example.
But let’s talk accident avoidance, lane departure warning, etc. All of these systems use fairly fixed situations, and the computer reacts very simply when an input exceeds a certain criteria. This technology, as far as the computer is concerned, is not a lot different than the technology that retards timing based on the knock sensor. If it were described in terms of video gaming, its Pong.
A few of these systems involve rudimentary logic - they will steer you away from an obstacle, unless a sensor knows there is an object also next to you, for example. One step logic. Think about all the logic that you process in your head, when you do a simple driving task. Say there’s a dead skunk in the road. I make a determination as to whether or not it will damage my vehicle to go over it...if it would, I make a determination whether or not I can stop in time...if not I look at the shoulder...I see there is a bunch of broken glass on the shoulder...so I consider crossing into oncoming traffic. It may be a double yellow, and there may be oncoming traffic way too close to even consider a traditional passing maneuver...but I make a judgment call that I can momentarily swerve out into the oncoming lane, to avoid the skunk. And btw, I was keyed to this problem early, because I observed several cars in front of me going around it also. This is multi-step logic. Think Halo 3 vs pong. We’ve got a long way to go, before accident avoidance leaps from Pong to Halo.
And that’s what Google is doing, with $150k worth of LIDAR and computers in the car.
Now we haven’t even talked about weather. If this is ever to be viable in trucking, it must work all the time. Even in a freezing drizzle that covers the sensors. I mentioned I have experience with LIDAR - guess hat we do when it rains or is snows? We have a cover we put over the LIDAR head.
And what happens when the fleet gets older? What happens to a new car, when it gets older? The power locks start to get erratic...some of the window switches don’t always work, and the power seat is stuck in one position until the end of time. All this equipment, in order to justify its up front costs, would have to be robust enough to last a few years. We are constantly sending laser equipment to the manufacturer for calibration and repair. So I’m skeptical.
And frankly, when you start bolting hundreds of thousands of dollars of equipment to an unmanned vehicle...you are inviting theft.
The list goes on. I will make you an iron clad guarantee. There will not be a large scale use of driverless cars for at least the next 50 years.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.
