|
Comrade Gorbash posted:The problem here is conflating two different kinds of error, systematic and random. Humans tend to make random errors - if you run someone through the same situation a hundred times, they'll probably screw up at some point, but you can only predict it as a statistical chance. Programmed systems already do far better than humans with random errors. I'm not conflating anything, what I said is correct and comparable regardless of the classification of the error, or the ability of it to be recreated under given conditions. The fact that you can introduce mitigation steps, in this case by pushing software updates, among several others, to those sort of systematic errors is a bonus from an engineering and widespread acceptance perspective.
|
#
?
Mar 22, 2017 02:38
|
|
|
|
| # ? May 27, 2024 04:41 |
|
|
CarForumPoster posted:I'm not conflating anything, what I said is correct and comparable regardless of the classification of the error, or the ability of it to be recreated under given conditions. The fact that you can introduce mitigation steps, in this case by pushing software updates, among several others, to those sort of systematic errors is a bonus from an engineering and widespread acceptance perspective. Just loving lol at the notion of fiat-chrysler turning out avionics quality software for their poorly built poo poo boxes; it will be garbage suitable only for a narrow window of road conditions, and only when the car is working perfectly. I am curious to see what happens with multiple autonomous car systems mixed with human drivers start interacting oddly with each other en-mass perhaps generating weird oscillations or mass stoppage scenarios. Edit: Try to sound less excited about autonomous mayhem! Blitter fucked around with this message at 03:27 on Mar 22, 2017 |
|
#
?
Mar 22, 2017 03:16
|
|
|
Chrysler has been using the same defective clockspring on Wranglers for a decade, and extending the warranty on the drat things each year as time passes rather than fix the problem. I wouldn't trust them to design and program an automatic egg timer.
|
|
#
?
Mar 22, 2017 03:29
|
|
|
Mortabis posted:Yes, but people connect through those hubs on those airlines to other final destinations, which competes with US routes. Yeah, to back this up, there is a big (and growing) spat between the ME3 airlines (Emirates, Qatar and Etihad) and the US3 (Delta, United and American). The US3 assert that because of their ownership by their respective state to varying degrees, and through massive direct or indirect subsidies, the ME3 do not compete on a level playing field with any other airline, and that they operate less like airlines and more like a marketing tool for their respective states. All of this is true to some degree. On the other side of the argument, the ME3 allege that the US3 are colluding to prevent them from even getting the opportunity to compete with the US3, and are openly calling them out by saying that instead of resorting to legal/political action (what seems to be the favourite tactic of American business these days), they should compete on service, amenities as well as price. Again, all of this is true to some degree. Since this ruling disproportionately affects the ME3's ability to do business in the US, I can see the US3 framing this as a victory for themselves, and also the ME3 taking whatever legal action is available to them to stop it.
|
|
#
?
Mar 22, 2017 03:39
|
|
|
Blitter posted:Just loving lol at the notion of fiat-chrysler turning out avionics quality software for their poorly built poo poo boxes; it will be garbage suitable only for a narrow window of road conditions, and only when the car is working perfectly. Self-driving cars are just around the corner! [takes a few steps back from the curb]
|
|
#
?
Mar 22, 2017 03:47
|
|
|
It's ironic that the US3 are fighting alleged government intervention that benefits the ME3 with....government intervention. The US3 know they can't compete with their bullshit product and are having to hide behind the specter of terrorism to force people on to their aging airplanes with sub standard service and beaten down crews just so they can keep their iPads and laptops.
|
|
#
?
Mar 22, 2017 03:49
|
|
|
Aren't most car manufacturers finding major software companies to do the automated legwork part? I highly doubt Chrysler is writing the software on this one, it makes no sense for them to take on something that difficult when they can just partner with someone else. Self driving cars, but more importantly trucks, are going to be a thing in the very near future. I know thats been said forever but there is a lot of very clear evidence that this poo poo is actually going to happen and large companies are working hard on getting it right because companies like Uber will literally throw billions of dollars at it. The major trucking/shipping companies alone will likely end up pushing it along all on their own once the financial benefits start becoming real. Mazz fucked around with this message at 03:55 on Mar 22, 2017 |
|
#
?
Mar 22, 2017 03:51
|
|
|
Mortabis posted:Yes, but people connect through those hubs on those airlines to other final destinations, which competes with US routes. Let them fly American Airlines. 
|
|
#
?
Mar 22, 2017 04:09
|
|
|
FuturePastNow posted:Wouldn't the best way to make a vision-based airplane nav system be to just point a camera at the cockpit guages Yes and no. I'm specifically interested in the ability to create something like an autoland system that doesn't rely on expensive radio navaids that aren't ever going to be installed at most airports. Automating flights between major, well-equipped airports is one thing, but what are you going to do about critical flights to fly-in reserves that don't even have paved runways, for example? En-route navigation seems like a comparatively small problem, and landing drones on modern aircraft carriers (as an example) doesn't seem like a result that can be generalized to other situations. But I'd be interested in being proven wrong. Would a sufficiently precise GPS system be enough for those situations? EDIT: In other news, the drunk Slovakian pilot who got arrested in Calgary apparently arrived at 12:48AM, and was scheduled to report for duty at 6AM. They found an empty fifth of vodka in his hotel room. I don't doubt a robot could do better than that  I'm fairly sure I couldn't even drink a fifth in 5 hours, much less do anything other than puke and curl up in the fetal position afterward.
PT6A fucked around with this message at 04:16 on Mar 22, 2017 |
|
#
?
Mar 22, 2017 04:10
|
|
|
CarForumPoster posted:I'm not conflating anything, what I said is correct and comparable regardless of the classification of the error, or the ability of it to be recreated under given conditions. The fact that you can introduce mitigation steps, in this case by pushing software updates, among several others, to those sort of systematic errors is a bonus from an engineering and widespread acceptance perspective. You are correct that there are "self-driving cars" today that use CV for navigation, but you're ignoring the reality of what contemporary "self-driving" means and the vast distance still to cover between NHTSA Level 2 automation (what Tesla, M-B, Audi, and most other high-end manufacturers have today) and NHTSA Level 4 (what people think of when they think of a self-driving car.) A Tesla can keep itself between the lane markers (most of the time) (if there are lane markers available) (and they aren't covered up) (and there's nothing strange like a pylon or jersey barrier in the way) and it can maintain distance to the car ahead of it and automatically brake (most of the time) (if the car ahead is producing a good return) (and it's not too sunny) (and the target surface isn't white). That's technically self-driving. It's much better to think of it as a very good cruise control, given the list of situations where it's acceptable to use it. And to your point that "Teslas are safer than humans already" -- there isn't enough data to make statistically valid comparisons yet, but when that guy using Autopilot got decapitated, it was one death in ~100 million autopilot miles, which is nearly identical to the normal rate of automotive fatalities per mile. And the human-driver statistic is for all types of vehicles on all types of roads and in all situations, not "extremely safe luxury cars with automatic braking and collision avoidance, cruising at a constant speed on an open freeway in good weather." By that metric, you can assume that Tesla autopilot is currently less safe than a human driver. Mazz posted:Self driving cars, but more importantly trucks, are going to be a thing in the very near future. I know thats been said forever but there is a lot of very clear evidence that this poo poo is actually going to happen and large companies are working hard on getting it right because companies like Uber will literally throw billions of dollars at it. The major trucking/shipping companies alone will likely end up pushing it along all on their own once the financial benefits start becoming real. Self-driving NHTSA Level 2 cars (can maintain lane position and distance from obstacles automatically in certain situations, requires the human driver to be alert at all times and ready to take control instantly) already exist. Self-driving NHTSA Level 4 cars (car can do anything a human driver can do within the car's normal operational envelope, i.e. no offroading or driving through a blinding snowstorm but otherwise you will never need to touch the wheel) are not going to be a thing for a decade at least. Probably longer. Self-driving trucks are an idiotic idea that produces no benefit or savings for anyone at all and doesn't make the remotest bit of sense for the way trucks are used in logistics networks. Uber is going to collapse within the year specifically because they are throwing billions of dollars down the toilet on impossible goals. e: for the record, NHTSA Level 3 is essentially "level 2, but for longer periods of time." There is still no expectation that the car can drive itself in any situation, but the car can handle "most of the trip" and only require transferring control to the human in certain circumstances. This means that the car has to handle emergencies on its own, because the human driver may be distracted and it may take a few seconds for them to wake up and react. Since being able to handle that reliably gets you most of the way to level 4, most of the manufacturers plan to skip level 3 and simply sell level 2 cars until level 4 is ready. Tesla claims they're at level 3 already, but they obviously aren't because the cars crash when you aren't paying attention. Level 1 is speed-holding cruise control, Level 0 is fully-manual driving, and level 5 is a car that can drive itself and also plan a route through an unmarked field and figure out how to get itself unstuck from the mud and other stuff that probably requires hard AI. Sagebrush fucked around with this message at 04:59 on Mar 22, 2017 |
|
#
?
Mar 22, 2017 04:49
|
|
|
Sagebrush posted:Uber is going to collapse within the year specifically because they are throwing billions of dollars down the toilet on impossible goals. That would be a fitting reward for their hubris, which means there is no way on earth that it will happen.
|
|
#
?
Mar 22, 2017 04:52
|
|
|
PT6A posted:Yes and no. I'm specifically interested in the ability to create something like an autoland system that doesn't rely on expensive radio navaids that aren't ever going to be installed at most airports. Automating flights between major, well-equipped airports is one thing, but what are you going to do about critical flights to fly-in reserves that don't even have paved runways, for example? There are a subset of tasks for autonomous vehicles that are going to be infeasibly expensive at the least or just not currently possible. I think we'll see manned control for lots of those tasks for a good while yet. Probably includes lots of winter conditions, and most anything unpaved. 
|
|
#
?
Mar 22, 2017 05:03
|
|
|
 
|
|
#
?
Mar 22, 2017 05:20
|
|
|
Is that...is that a bow-wave of dirt in front of the main gear? That's  as hell.
|
|
#
?
Mar 22, 2017 06:05
|
|
|
It's the strip at Sainte-Mère-Église shortly after D-day, which was made with Sommerfeld tracking--basically wire mesh with steel rods woven through to try to stiffen it. Jack Lieb, the war correspondent who filmed that says in narration to the video I got the gif from that the vibrations were strong enough to uncouple 500lb bombs from the wings of departing P-47s and that there were incidents of that bulge tangling into props  https://youtu.be/a4kmRTZrgMQ?t=1297
|
|
#
?
Mar 22, 2017 06:53
|
|
|
Sagebrush posted:Level 1 is speed-holding cruise control, Level 0 is fully-manual driving, and level 5 is a car that can drive itself and also plan a route through an unmarked field and figure out how to get itself unstuck from the mud and other stuff that probably requires hard AI. True level 4 really requires hard AI.
|
|
#
?
Mar 22, 2017 06:54
|
|
|
PT6A posted:Yes and no. I'm specifically interested in the ability to create something like an autoland system that doesn't rely on expensive radio navaids that aren't ever going to be installed at most airports. Automating flights between major, well-equipped airports is one thing, but what are you going to do about critical flights to fly-in reserves that don't even have paved runways, for example? It's pretty easy to do a computer vision autoland. Edge detection is really easy when your edge is a perfectly straight runway and you know roughly where it should be, and only slightly harder when its an empty field (but then there's nothing to hit, right? ) Certifying it might be a problem, but the tech is straight forward. The question is, since it's VFR only, why bother?
|
|
#
?
Mar 22, 2017 06:58
|
|
|
HookedOnChthonics posted:It's the strip at Sainte-Mère-Église shortly after D-day, which was made with Sommerfeld tracking--basically wire mesh with steel rods woven through to try to stiffen it. Jack Lieb, the war correspondent who filmed that says in narration to the video I got the gif from that the vibrations were strong enough to uncouple 500lb bombs from the wings of departing P-47s and that there were incidents of that bulge tangling into props Wow the collision in the next scene is brutal
|
|
#
?
Mar 22, 2017 08:41
|
|
|
BurgerQuest posted:Wow. One of my worst fears is a propeller breaking off and coming right through the fuselage at me. http://www.airlineratings.com/news/1110/more-rex-aircraft-grounded-after-prop-incident has some new information on this: quote:A police helicopter has found a propeller that flew off a Regional Express Saab passenger aircraft as it approached Sydney on Friday. quote:The Australian Transport Safety Bureau said the propeller was found broadly in an area matching a trajectory calculated using information from the plane's flight data recorder. quote:The crew had feathered the propeller to reduce drag and it was at this point that the first officer saw it separate from its shaft. quote:The airline said the propeller was found to have sheared off at the shaft and all the fittings of the main assembly were intact. http://www.abc.net.au/news/2017-03-21/rex-flight-detached-propeller-found-in-bushland/8373504 says it landed in bushland in suburban Sydney, metres from houses, and weighs 100kg! I think that as a child I would have wished something like that would land in my back yard. Now I'm not so keen.
|
|
#
?
Mar 22, 2017 09:37
|
|
|
The audio recording of the pilot calling it in was hilarious. You'd never know he'd just lost half his available thrust. Talked about it as if it was at best, a mild annoyance.
|
|
#
?
Mar 22, 2017 09:49
|
|
|
HookedOnChthonics posted:It's the strip at Sainte-Mère-Église shortly after D-day, which was made with Sommerfeld tracking--basically wire mesh with steel rods woven through to try to stiffen it. Jack Lieb, the war correspondent who filmed that says in narration to the video I got the gif from that the vibrations were strong enough to uncouple 500lb bombs from the wings of departing P-47s and that there were incidents of that bulge tangling into props
|
|
#
?
Mar 22, 2017 14:11
|
|
|
david_a posted:The P-47s seem to have something white in the gun barrels. What's that? Wax plugs to keep dust out? Condoms, generally.
|
|
#
?
Mar 22, 2017 14:28
|
|
|
On the laptop ban, there appears to be intelligence about Al Qaeda bombs, see here: http://www.cnn.com/2017/03/21/politics/electronics-ban-devices-explosives-intelligence/index.html The UK and Canada joined in the ban as well. As for Uber, I'm waiting for that company to collapse. It runs on people who are willing to lose money as drivers. As long as there's a supply of poor suckers, Uber could exist. Sperglord fucked around with this message at 14:47 on Mar 22, 2017 |
|
#
?
Mar 22, 2017 14:43
|
|
|
Sperglord posted:....I'm waiting for that company to collapse... As long as there's a supply of poor suckers, Uber could exist. So don't hold your breath then.
|
|
#
?
Mar 22, 2017 15:33
|
|
|
Carth Dookie posted:The audio recording of the pilot calling it in was hilarious. You'd never know he'd just lost half his available thrust. Talked about it as if it was at best, a mild annoyance. "We've had uncommanded engine operations." How droll.
|
|
#
?
Mar 22, 2017 15:39
|
|
|
Uber's basic business model is brazenly ignoring government-sponsored taxi cartels. A mentally retarded gorilla could make money doing that. Eventually it may have to give up its research or something but undercutting taxi prices in cities where supply is artificially restricted by medallions is trivial if you can get away with it, and Uber evidently can. Someone, maybe Uber, maybe someone else, will continue the "ridesharing" business even if Uber goes bankrupt. Also, lol: https://electrek.co/2017/03/22/electric-plane-startup-150-seat-battery-powered-plane/ seeing as Jet-A has at least 40 times the specific energy of even the best batteries this is hilariously stupid.
|
|
#
?
Mar 22, 2017 15:54
|
|
|
Blitter posted:Just loving lol at the notion of fiat-chrysler turning out avionics quality software for their poorly built poo poo boxes; it will be garbage suitable only for a narrow window of road conditions, and only when the car is working perfectly. Sagebrush posted:You are correct that there are "self-driving cars" today that use CV for navigation, but you're ignoring the reality of what contemporary "self-driving" means and the vast distance still to cover between NHTSA Level 2 automation (what Tesla, M-B, Audi, and most other high-end manufacturers have today) and NHTSA Level 4 (what people think of when they think of a self-driving car.) I am talking about the ability to use computer vision based system to drive a car, and made the point that doing so is safer by some commonly accepted metrics than having a human driver. Yes it is at NHTSA 3 Levels. I don't agree that Level 4+ is the only definition of a self driving car. The idea that only you get to define what the comparative basis is is silly. There are enough miles driven with autopilot on that the number of events can be compared. That they have to only be compared to NHTSA Level 2 cars is silly. Related: http://fortune.com/2017/03/14/tesla-autopilot-insurance/ Related to this thread, I can compare the safety record of huge, super expensive, more safety feature having 787/A380/etc. to regional jet liner operating a combo of MD-80s and Dash 8s. (I am guessing the former have fewer incidents but haven't checked)
|
|
#
?
Mar 22, 2017 16:29
|
|
|
Ola posted:So don't hold your breath then. Touche... Mortabis posted:Uber's basic business model is brazenly ignoring government-sponsored taxi cartels ... and paying drivers less than minimum wage after gas prices, maintenance, and automotive depreciation are taken into account. Anyone can make money that way... On electric airplanes, I read an AIAA Magazine article to the effect that NASA is giving up on pure electric airplanes, as battery technology is not advancing as well as expected.
|
|
#
?
Mar 22, 2017 16:37
|
|
|
Mortabis posted:Also, lol: https://electrek.co/2017/03/22/electric-plane-startup-150-seat-battery-powered-plane/ I think an airship might work? The V tail is comical on that, but I guess that goes with the entire "doesn't know physics" thing
|
|
#
?
Mar 22, 2017 17:01
|
|
|
Mortabis posted:Also, lol: https://electrek.co/2017/03/22/electric-plane-startup-150-seat-battery-powered-plane/ Hell, why not instead of speed brakes and high-drag flap configurations, the plane's have turbine generators that will regenerate energy as they descend from 30k feet while slowing them down. There's no need to jeopardize reserve amounts by weighing airline profit margins either, the battery doesn't weigh any more when it's full than empty. And when you have to turn around because of an unruly drunk passenger, you don't have to spray all that fuel out into the sky before you land. On top of that they can tie the plane's energy use to the grid, work out large scale deals with the power companies and on-site energy storage at airports and now nothing moves except the aircraft, no fuel shipping, fuel ferrying, fuel trucks. Just wires to transfer energy that will last for decades because electricity never changes. Instead of huge fuel storage facilities, there are just energy storage facilities that would also provide a benefit for not only the airport but the city its located in. I'm not saying it will be easy, but it will happen. Remember, we're not talking about replacing an A380 that flies for 10,000 miles. A lot of these regional planes aren't even in the air as long as they are boarding, taxiing, takeoff, landing, and deboarding.
|
|
#
?
Mar 22, 2017 17:57
|
|
|
Vitamin J posted:Electric flight is coming.
|
|
#
?
Mar 22, 2017 18:12
|
|
|
Vitamin J posted:Electric flight is coming. Tesla just doubled the energy density of their cells in the past 3 months. Double it one more time and suddenly an electric 737 could work. Sure fuel is more energy dense, but electric propulsion is more energy-efficient. For comparison my electric car can go 90 miles, but the energy equivalent of the battery pack is less than 1 gallon of gasoline so at the end of the journey I'm using less than 3x the energy of even a super efficient car. Same holds true for airplanes, if Jet A costs $$$, then filling a plane up with electricity for the same journey would only cost $. Also electric airplanes can use electric traction motors in the landing gear for taxiing so there's no need to run the main thrust engines until take off, think of the fuel savings in that alone. The traction motors can run as regenerative braking and the thrust fans can as well when landing for recover energy. You would have to double it at least four times and jet engines are already not that heavy. Electric motors are more energy efficient than the engine of your car, but is it more efficient than a jet engine when flying >FL300 and mach 0.8? Suddenly it looks a lot less appealing. There is no reason to believe that electric aircraft will ever replace fossil fuel jets, especially if you think electric cars and trucks are inevitable. Fuel prices, as we have seen, do not just increase monotonically, and if you cut out the demand for gas and diesel from ground vehicles you're left with plenty for aircraft. Mortabis fucked around with this message at 18:23 on Mar 22, 2017 |
|
#
?
Mar 22, 2017 18:18
|
|
|
There's also the chance of replacing fossil fuel jet aircraft with biofuel jets, although that's a different can of worms, figuratively speaking. Electric recreational flying seems pretty viable though. Give it five years and you could probably get something like the Airbus E-fan which crossed the Channel two years ago with three hours of flying time. Or you can just drive your EV up a hill and fly your paraglider from there.
|
|
#
?
Mar 22, 2017 18:31
|
|
|
Even when every other form of transport is electric, we will probably make artificial hydrocarbons for planes. They’re just too good at storing energy, and jet engines are just too good at converting hydrocarbons to thrust.
|
|
#
?
Mar 22, 2017 18:40
|
|
|
Mortabis posted:Uber's basic business model is brazenly ignoring government-sponsored taxi cartels. A mentally retarded gorilla could make money doing that. Uber's business model is undercutting taxi services by subsidizing around 50% of each ride out of VC funds. They lost around 3 billion dollars in 2016 alone. They're hideously unprofitable both in the short and long term; the only way they could turn it around is to eradicate literally every other competitor and jack up the rates, or replace every single Uber car with a level 4 self-driving car that works perfectly in all situations and also costs half as much to purchase, maintain, fuel and clean as they were paying their drivers (who cover all those costs themselves). Maybe they should have hired the retarded gorilla. CarForumPoster posted:I am talking about the ability to use computer vision based system to drive a car, and made the point that doing so is safer by some commonly accepted metrics than having a human driver. Yes it is at NHTSA 3 Levels. I don't agree that Level 4+ is the only definition of a self driving car. It's semantics. Level 4 automation is what people think of when they think "self-driving car," as in "a car that I can get into and sit back and do spreadsheets or crush code or go to sleep." A level 1-3 automated car is also "self-driving" to some extent, but because they still require human attention for many/most daily driving situations and cannot reliably handle emergencies, it's important to think of those levels as advanced driver assistance and not as a limited form of self-driving. Tesla would love for you to believe otherwise, of course, at least while you're buying the car; when the car goes off the road because the lines were being repainted and you were digging around in the back for a snack, they'll be sure to emphasize that you were supposed to be ready to take control of the car at a moment's notice at all times. Your point about computer vision doesn't really mean anything. The US Army had an experimental computer-vision system to drive APCs in the 1980s that worked by looking for parallel lines (the edges of the road) and attempting to keep the truck in the middle of them. Turned out that trees also had parallel lines, oops, and the technology wasn't ready yet. It's better today but this isn't a yes/no thing, it's gradual continuous improvement.
|
|
#
?
Mar 22, 2017 18:52
|
|
|
Sperglord posted:On the laptop ban, there appears to be intelligence about Al Qaeda bombs, see here:
|
|
#
?
Mar 22, 2017 18:56
|
|
|
Sagebrush posted:Self-driving trucks are an idiotic idea that produces no benefit or savings for anyone at all and doesn't make the remotest bit of sense for the way trucks are used in logistics networks. While I'll concede most of those points, the idiotic part of self driving trucks is questionable to me. I'm not referring to replacing Amazon delivery vans when I mention this. While it's not applicable for every situation, there are countless examples where trucks are required to travel long distance by highway to set shipping hubs, situations like the business I'm in where trucks are driven daily from Mexico/California/Florida to Chicago, and 99% of that driving is either highway or immediately off it to very truck friendly shipping centers. While I completely agree it's still probably a decade away (I should've been more clear that's really the reasonable time frame for the automation we need), there seems to be plenty of situations where trucking could benefit from reducing the numbers of drivers needed to perform the portions of work, merely congregating at offloading points to handle the final steps of backing into tight confines or offloading/loading for the next step. And yeah the decade away argument has been a thing forever but besides the last 5 years when did major tech and truck manufacturers actually dump the kind of money into this that they have been? The potential is finally there to make real progress. Refueling is an obvious problem with full automation but not an insurmountable one, nor am I talking full automation, just a lot more robust driver assistance in most situations. Most of this work is very set routes where the actual road data can be preloaded and the main work is obstacle management (which I agree is still a hurdle). Additionally to this, solid automation technology removes a lot of the issues faced with time restrictions on drivers and the accidents that coincide with this, many of which tend to garner a lot of news time. I've seen several articles where major trucking companies expect large problems with staffing in the next decade, and many see AI as a good investment on this if it shows any progress. Mazz fucked around with this message at 19:27 on Mar 22, 2017 |
|
#
?
Mar 22, 2017 18:58
|
|
|
Platystemon posted:Even when every other form of transport is electric, we will probably make artificial hydrocarbons for planes. They’re just too good at storing energy, and jet engines are just too good at converting hydrocarbons to thrust. What about high-endurance solar-electric drones, like Helios? Do they have a niche or can Global Hawk-type gas drones still do better?
|
|
#
?
Mar 22, 2017 19:00
|
|
|
Didn't Helios get torn to pieces by high winds?
|
|
#
?
Mar 22, 2017 19:02
|
|
|
|
| # ? May 27, 2024 04:41 |
|
|
Mazz posted:While I'll concede most of those points, the idiotic part of self driving trucks is questionable to me. While it's not applicable for every situation, there are tons of where trucks are required to travel long distance by highway to set shipping hubs, situations like the business I'm in where trucks are driven daily from Mexico/California/Florida to Chicago, and 99% of that driving is either highway or immediately off it to very truck friendly shipping centers. While I completely agree it's still probably a decade away (I should've been more clear that's really the reasonable time frame for the automation we need), there seems to be plenty of situations where trucking could benefit from reducing the numbers of drivers needed to perform the long haul portions of work, merely congregating at offloading points to handle the final steps of backing into tight confines or offloading/loading for the next step. Most of this work is very set routes where the actual road data can be preloaded and the main work is obstacle management (which I agree is still a hurdle). Keeping with the technology readiness level thing, testing in the environment at a system level is at TRL 7, so we are closeish in terms of technology. Otto completed a beer run with a semi autonomous truck and engineer/driver on board. I think the next step would be highway only trucks that go from city to city and depart the highway near the destination city to be picked up and have the last 20 miles driven by humans. I am with you though, the problems solved by semi autonomous trucking that stops at the city limits: -Enables truck drivers to go from driving all over the country to being city experts. This greatly improves quality of life for the drivers and has the potential to reduce incidents. Work normal hours. -Increase the rate at which trucks can traverse the country, basically 24/7. Greatly reduces fixed cost of trucks due to higher utilization. -Potentially reduces the cost of shipping, no need to pay for people to drive it the whole way. (Drivers salary is ~1/3 of cost per mile) -This lower cost could take business from rail shipping depending on other market forces with regard to cost/mile so you'd now get products faster than via rail at comparable rates.
|
|
#
?
Mar 22, 2017 19:19
|
|