[video=youtube;PUw_DMaQ264]https://www.youtube.com/watch?v=PUw_DMaQ264[/video]
pretty bloody impressive.
pretty bloody impressive.
Not coming to motorcycles anytime soon
- Thread starter MacDoc
- Start date
bigpoppa
Well-known member
And why should it? Removes all the fun from riding....
Reminds me of the electric bike, no noise, no shifting...
https://www.youtube.com/watch?v=Lz1aTLBKIoQ&t=765s
Reminds me of the electric bike, no noise, no shifting...
https://www.youtube.com/watch?v=Lz1aTLBKIoQ&t=765s
Neat indeed.
Biggest issue I foresee with automatic self-driving cars is operator impatience. They drive the speed limits, they drive cautiously, and they operate within traffic laws. They don't aggressively weave through traffic on the highway to get 20 feet ahead of where they'd otherwise be. They won't blast to the very last 12 inches of a a highway onramp and then force their way in (or maybe drive down the shoulder a little) just go get ahead of everyone else. They won't pass when the car in front of them is doing the speed limit. They won't rocket through a red light at 50 over the speed limit a moment after it's changed to "get there faster". They don't do rolling stops at stop signs.
Basically, they drive like a professional, but they might take a few minutes extra to get to point B. For some people who are so bloody impatient (we all see them every day) that all the things they perceive as saving them time (see examples above) are no longer happening, I'm left wondering how many will have the patience to just let it do it's thing vs getting all flustered, disabling it, and driving in their usual idiotic fashion.
Biggest issue I foresee with automatic self-driving cars is operator impatience. They drive the speed limits, they drive cautiously, and they operate within traffic laws. They don't aggressively weave through traffic on the highway to get 20 feet ahead of where they'd otherwise be. They won't blast to the very last 12 inches of a a highway onramp and then force their way in (or maybe drive down the shoulder a little) just go get ahead of everyone else. They won't pass when the car in front of them is doing the speed limit. They won't rocket through a red light at 50 over the speed limit a moment after it's changed to "get there faster". They don't do rolling stops at stop signs.
Basically, they drive like a professional, but they might take a few minutes extra to get to point B. For some people who are so bloody impatient (we all see them every day) that all the things they perceive as saving them time (see examples above) are no longer happening, I'm left wondering how many will have the patience to just let it do it's thing vs getting all flustered, disabling it, and driving in their usual idiotic fashion.
The technology is much further along than I think many people expect. It relies heavily, from the look of the path planners and object recognition blocks on the right, on lane markings -- painted lines on the road etc. When roads look like this:
how will it do?
Will be interesting to see how it performs in tricky lighting too: fog? Heavy rains at night with the glare of oncoming headlights or where road features such as lines aren't as well lit or or high contrast to the road surface...
For a while it may be a "fair weather" friend to highway cruisers but I think there's a ways to go before it's ready for fleet deployment for all-weather, all-condition use.
But it's pretty damn impressive.
how will it do?
Will be interesting to see how it performs in tricky lighting too: fog? Heavy rains at night with the glare of oncoming headlights or where road features such as lines aren't as well lit or or high contrast to the road surface...
For a while it may be a "fair weather" friend to highway cruisers but I think there's a ways to go before it's ready for fleet deployment for all-weather, all-condition use.
But it's pretty damn impressive.
Hopefully, it doesn't drive slowly in the wrong lane either which is an even bigger factor.
Had this discussion in another thread a while back. It's still a pipe dream.
Yes, there was a proof of concept a while back, but reality is there was a driver sitting in the truck the entire time and it's unlikely regulators are going to waive that requirement anytime soon as the reality is that they're basically setting loose 80,000 pound "what iffs" at that point and the public is a LONG way away from accepting that.
And the system basically managed to drive the truck on the highway - as soon as it exited or got into any sort of traffic of city streets the system wasn't capable of driving it anymore, not to mention actually getting into a customers facility, opening doors, and backing into a dock. Add in Blackfins observation above about what happens when suddenly all the reference points it needs are no longer visible (road lines, anyone?) and you realize even more how unlikely it is to happen anytime soon.
Will it happen? Sure, it's inevitable, but I think it's 30-40 years away, and even then there's no way it's ever going to completely remove the person from the equation as there will be things it can't do.
Already a problem, at least so far as the driving slowly part. Going back to my earlier comments, a lot of people out there think that anyone driving the speed limit is an idiot that needs to GTFO of their way.
statsman
Well-known member
One of the great questions of our time I ask myself when riding or driving on the mean streets of Toronto.
If we're 10 feet from the corner and you want to turn right, what are you doing in the left lane?
I agree that it may not happen anytime "soon" but it is coming and perhaps that was Roadghost's point; the sooner they can do this, the happier they'll be.
For seemingly insurmountable problems like road lines, as one example, research into ground penetrating radar is showing promise. As for backing into a dock, again, research is ongoing and early product is out there (e.g. trailer assist) that is clearly heading in that direction. Sensors, cameras, LIDARs, radars etc are coming down in price as computing power is heading up. It's only a matter of time.
Not sure it's 30 or 40 years away for road vehicles. The current pace of automation is way faster than that.
Large aircraft are technically capable of take off, climb-out, cruise, descent, landing and roll-out without pilot intervention. (And, it bears noting, even at these relatively early stages of the technology when an aircraft crashes it's usually because of the sack of meat behind the yoke, not in spite of him.) I suspect you'd see a guy behind some sort of "wheel" or "brake" for a number of years though I don't know if it's going to be measured in decades. We'll see I guess.
Motorcycle Mike
Well-known member
I think the biggest hurdle for autonomous vehicles is going to be the unpredictability of people, and the fact we can be complete jerks.
Imagine a time when autonomous cars are the majority of vehicles on the road, and you want to cross the road 200m from the nearest crosswalk. On a busy street of human drivers, you likely wouldn't dare because you don't know if that human driver will see you and will stop. Once automated cars are the norm, and you can be confident the cars will stop, why walk to the crosswalk?... just cross the road, the automated cars will stop. Kids will have awesome new games to play... throw a ball in the street and see how many cars come to a stand-still.
Same thing goes for trucks. Automated trucks will stop for obstacles. Lets say you and your gang wants to steal a trailer of beer leaving the Sleeman brewery... you know the route, all you need to do is get the truck to stop, pull the pin on the trailer, hook up to your own tractor and leave. No witnesses to shoot, no worry of the driver ramming into you. Sure the truck will likely have cameras and an emergency beacon... so wear masks and use a signal jammer and get out of there fast. (ok, I've likely been watching too many gangster shows lately, but you get the idea).
I suspect that until these issues are figured out, people will become incredibly frustrated with self-driving cars because they will be slow, completely non-aggressive, and too easy to game. Will self-driving cars be made to be aggressive so they aren't easily gamed by jerks? What liability will the manufacturer face if they program a car to potentially kill? Mercedes recently announced their self-driving cars will favour the driver to pedestrians: http://blog.caranddriver.com/self-driving-mercedes-will-prioritize-occupant-safety-over-pedestrians/ ... which they have to do, because who will buy a car programmed to kill the driver?... but how many legal issues are they going to run into when a self-driving car plows into a group of pedestrians?
Anyway, interesting times coming for sure...
Imagine a time when autonomous cars are the majority of vehicles on the road, and you want to cross the road 200m from the nearest crosswalk. On a busy street of human drivers, you likely wouldn't dare because you don't know if that human driver will see you and will stop. Once automated cars are the norm, and you can be confident the cars will stop, why walk to the crosswalk?... just cross the road, the automated cars will stop. Kids will have awesome new games to play... throw a ball in the street and see how many cars come to a stand-still.
Same thing goes for trucks. Automated trucks will stop for obstacles. Lets say you and your gang wants to steal a trailer of beer leaving the Sleeman brewery... you know the route, all you need to do is get the truck to stop, pull the pin on the trailer, hook up to your own tractor and leave. No witnesses to shoot, no worry of the driver ramming into you. Sure the truck will likely have cameras and an emergency beacon... so wear masks and use a signal jammer and get out of there fast. (ok, I've likely been watching too many gangster shows lately, but you get the idea).
I suspect that until these issues are figured out, people will become incredibly frustrated with self-driving cars because they will be slow, completely non-aggressive, and too easy to game. Will self-driving cars be made to be aggressive so they aren't easily gamed by jerks? What liability will the manufacturer face if they program a car to potentially kill? Mercedes recently announced their self-driving cars will favour the driver to pedestrians: http://blog.caranddriver.com/self-driving-mercedes-will-prioritize-occupant-safety-over-pedestrians/ ... which they have to do, because who will buy a car programmed to kill the driver?... but how many legal issues are they going to run into when a self-driving car plows into a group of pedestrians?
Anyway, interesting times coming for sure...
For some perspective....remember that the first "flying car" was in the 1940's, it was the 60's when it was predicted we'd all be driving one by the year 2000, a generous 40 year prediction at the time. And hey, it's 16 years later and we are nowhere close. Not even fathomable at this point.
So yes, I'm highly skeptical about the whole situation of self driving cars, and particularly trucks. It's been proven that it's doable, but it's also been proven that it still requires a lot of perfect-world scenarios and things can and do fail, and go wrong.
Planes are highly predictable due to standardized practice. A sky full of the same is reasonably easy to automate...but throw one guy with an ultralight wandering into Pearsons airspace (for example) and watch the **** hit the fan - now things are no longer predictable and actual thinking needs to happen. This is the same thing that driverless vehicles struggle with when they're forced to share the road with unpredictable and often aggressive other drivers. If every other vehicle on the roads were also automated things would be easier of course, but that's even further into the future...and heck, it may never happen - any driving purist is never going to give up the experience of operating their vehicle themselves.
Pilotless planes are also a long way off. People trust humans, people won't place the same trust in computers - would you get on a plane that you know has no humans in the cockpit?
There's also the thing about not dealing with unexpected occurrences very well. A computer may be able to handle a "normal" emergency like an engine failure or whatnot, but would a computer have handled something like the Gimli Glider like the pilots did, or would the computer have just dorked the plane into whatever was underneath it when it ran out of gliding distance?
It's easy to suggest that these sorts of things don't happen very often...until you're the one on the plane when it does, or you're in the house underneath it.
The people suggesting that driverless cars can't properly see motorcyclists is also probably being downplayed by many..until you're the motorcyclist it doesn't see.
Except that we just saw a video of a car driving autonomously on public roads. The level of research into autonomy at academic and corporate levels -- especially on the part of automakers -- is astounding. I design electronics for a robotics firm doing autonomous vehicles for materials handling; even the things we learn are rolled into the whole autonomy ecosystem and vice versa.
I think it's a little regressive and short-sighted to assume that problems (such as the road lines issue mentioned above) to which those outside the research and development loop can see no solution make the notion of full autonomy half a century away. We went from Sputnik beep-beeping in LEO to men walking on the moon in just 12 years, with slide rules and wire-wrapped 1MHz computers.
(And BTW, I count myself among those outside the R&D loop; although I design the electronics used in our UGVs the autonomy guys in our organization operate in a different realm and on a different system level than I...)
A good point. But the velocities involved (and # of dimensions in play) means that it's likely that the worst that will happen in such a scenario is little different than that happens now; yes, someone may get hurt in those scenarios where they would have been hurt with human drivers at the controls. I also suspect that in pretty much all of those cases fault will not be lain at the foot of the autonomous vehicles but rather the aggressive/incompetent/suicidal person. No machine or system will be 100% safe against those that ignore rules or purposely misuse resources: If you step into traffic without looking there's a chance you could be seriously hurt. Doesn't matter if the cars are autonomous or meat-driven.
I suspect that if you factor in human emotions and potential for distraction, reaction times etc, most would welcome increasingly competent autonomous control of vehicles. As a motorcyclist I look forward to autonomy measures that will help prevent left-turn errors or lane-incursions. Commuters should look forward to a time in the not too distant future when a trip to Toronto on the 401 is orderly and efficient and free from every-day wrecks caused by inattentive or incompetent sacks of meat in control now.
Yup.
Every day we get on planes that were, the night before, maintained by guys with what amounts to high-school and college diplomas. Why not trust proven systems that are in use on probably 90% of most commercial flights already? When your 757 from Antigua hits the tarmac hard at YYZ it's probably because the pilot took control away from the ILS at the DH and landed the plane himself, just to stay in practise...
Again, we face that problem now with meatsack drivers; I would never make any assumptions about my visibility to a human nor to a computer. I simply feel that the computer won't be distracted by screaming kids in the back, gabbing on the CB radio or phone, texting, eating, applying makeup or any of myriad other sources of distraction humans subject themselves too every time they get behind the wheel.
Same video at normal speed - at Min 3:55 notice the car got confused by two joggers and the angle of approach of the road and it stopped in the middle of the road for no reason, that would have caused an accident as the driver behind would have no reason to suspect the car in front would just stop, yea no thanks.
https://www.youtube.com/watch?v=VG68SKoG7vE
https://www.youtube.com/watch?v=VG68SKoG7vE
Last edited:
The ILS system doesn't land a plane, it's a ground based system. It just provides the raw data that the planes pilot (or autopilot) uses. I've flown an ILS approach before, it's not a particularly complicated system, many 2-4 seat general aviation planes even have the avionics equipment for it.
As for the (plane/autopilot based) autoland feature, I suggest you look into its limitations – even a very light headwind or an even lighter tail wind can disqualify it from use, and it is notoriously bad at handling things like wind shear or severe cross winds to the point where many airline SOP's limit or do not allow its use in many challenging weather conditions.
Planes are landing themselves far less than most people think.
Emphasis mine. You're picking the flyshit out of the pepper: It's a system with components on the ground and the aircraft. Beacons on the ground supply glideslope and localizer signals to the APP- & G/S-equipped AP on the aircraft through antennae on the fuselage. It is a requirement for autoland.
Anyway, like arguing we'll basically never have autonomous cars and comparing the situation to not having flying cars yet while seemingly oblivious to the video of a Tesla autonomously driving on public roads, I think you missed the point: That the capability to autoland large commercial aircraft already exists and is used in challenging weather conditions suggests that the pilotless aircraft is not as far off as you may think or hope.
Similarly, backing a tractor trailer into a dock -- which you make out to be an onerous and tricky thing only a bearded guy with a beer belly can figure out -- is a comparatively simple physics problem that can be solved quite easily with localization, cameras, LIDARs and a physical model of the rig for an algorithm to work with. Even opening a trailer door; air, hydraulic or electric actuators are relatively easy additions to any trailer.
I guess I sense resistance on your part because (a) yours is one of the jobs that would be at risk from such automation so naturally you don't want it to thrive, but also (b) you don't fully grasp or appreciate the true state of the art as it applies to autonomous vehicles. I think your own perceptions about how challenging certain scenarios are doesn't allow you think that such problems have solutions already validated, under test or being designed.
It's a pretty rapidly evolving and developing field. Automakers have been rolling the required sensors and hardware into their cars for some time now (think about things like DBW throttle, electrically-assisted power steering, ABS pumps, blind-spot sensing, rear cameras, radar-based cruise-control) as well as adding functionality that is inching closer to autonomy (e.g. parallel parking, lane-keeping, auto-braking/collision avoidance, trailer backing assists and, in more advanced cases like Tesla, full autonomy suites that are very close to -- if not already -- the "real deal.") The Tesla & beer-truck vids may appear to be stunts to the disinclined, to the naysayer, to the "never gonna happen" crowd but the fact is it's already happening.
I've never said that any of it isn't going to eventually happen, I think that's inevitable – what I do disagree with is the timelines for it reaching mass adoption and regulatory acceptance.
To return to the flying car example once again, they do actually exist and have in various forms for decades, however when was the last time you saw one at your local Walmart?
Like somebody else mentioned, when the systems are so easily fooled by a couple of joggers on the sidewalk it becomes evident that although the capabilities exist, they're still in their infancy and far from reaching mainstream so far as the vaunted "just have a nap in the back seat while the car gets you to point B" dream.
The tractor that I am sitting in right now as I type this has a state of the art collision avoidance system in it, so I'm not exactly clueless about the topic, I have first-hand experience – I've also had lots of first-hand experiences where it false triggers at a variety of objects (reflective corner markers in curves are its favorite) that it shouldn't. Just a few weeks ago it slammed on the brakes unnecessarily because it detected a car that was turning right in front of me (in its own turn lane) as apparently being in my lane and at risk of imminent collision. It's first-hand experience with automation like this that makes me realize that although it's a useful tool, it's certainly not anything to be 100% relied upon anytime soon...i.e. the "the driver/pilot/engineer" serves no practical purpose anymore.
To return to the flying car example once again, they do actually exist and have in various forms for decades, however when was the last time you saw one at your local Walmart?
Like somebody else mentioned, when the systems are so easily fooled by a couple of joggers on the sidewalk it becomes evident that although the capabilities exist, they're still in their infancy and far from reaching mainstream so far as the vaunted "just have a nap in the back seat while the car gets you to point B" dream.
The tractor that I am sitting in right now as I type this has a state of the art collision avoidance system in it, so I'm not exactly clueless about the topic, I have first-hand experience – I've also had lots of first-hand experiences where it false triggers at a variety of objects (reflective corner markers in curves are its favorite) that it shouldn't. Just a few weeks ago it slammed on the brakes unnecessarily because it detected a car that was turning right in front of me (in its own turn lane) as apparently being in my lane and at risk of imminent collision. It's first-hand experience with automation like this that makes me realize that although it's a useful tool, it's certainly not anything to be 100% relied upon anytime soon...i.e. the "the driver/pilot/engineer" serves no practical purpose anymore.
Well, if you prefer often somewhat unintelligible posts with random links and unrelated videos mixed in for no apparent reason, there's someone here for that too.
God forbid somebody has a debate around here that involves more than 10 word responses, however.
If its TLDR for you, feel free to scroll on past.
God forbid somebody has a debate around here that involves more than 10 word responses, however.
If its TLDR for you, feel free to scroll on past.
Last edited:
