Rossi Vs Riderless Bike

[peter-may]

https://youtu.be/ISz8Mvs7GcE

The future's here.....is that possible?

First attempt at adding link via telling bone......anything could happen.

[dablik]

Jeebus wonder if they do a swmbo version impressive though.

[Studley Ramrod]

It's more a bike ridden by a bot, rather than a riderless bike, fookin' impressive though. 

[dandywarhol]

Anyone watch Guy Martin take on the AI car on Ch4? Really interesting programme (even better using subtitles  ). Still a way to go , especially when AI hadn't a seat of the pants feel when the tyres cooled off and it launched into the kitty litter.....................

[TKH]

Would expect the AI to beat the man - AI has no fears and should be able to replicate the perfect lap. 

[fixitsan]

Would expect the AI to beat the man - AI has no fears and should be able to replicate the perfect lap. 

 

I tend to agree but AI has no feel. It is programmed to avoid collisions, so like the adaptive cruise control on cars, if you pull in front of it then it must take avoiding action. i don't know how they'll negate that for track use !

 

Impressive though it seems, is it really much better than the self driving car of ten years ago ?

 

https://www.youtube.com/watch?v=VGUWz-29gEs

 

 

These vehicles are being tested in strictly controlled and enclosed situations. The biggest problem is yet to be overcome, ordinary road conditions. A track is a relatively safe place to play with this tech and we shouldn't just be taken in by the idea that a high  speed on the track equates to a well developed product, because it is 'nearly as good as a professional rider or driver'.....The driverless Golf from ten years ago was nearly as good. IMHO

 

I'm a big tech fan, but carefully controlled stunts don't seem to me to be relevant to the real world for much of the time

 

There are some other interesting problems which at the moment seem to be dealt with only in a very simple manner - why they should be programmed to kill.... https://www.technologyreview.com/s/54262...d-to-kill/

 

 

 

 

When I was studying  to be clevererer than wot I was before i did have to do some research which looked at control systems, and what i found was a 1950's research paper by a big US university which developed road trains made from radio interconnected cars, for highway use. The driver still steered but the cars maintained a gap of about 12 inches between themselves at highway speeds automatically  - The first car punched a hole in the air and the following cars made quite a significant fuel saving....AFAIR they had about 20 cars going around a racing oval for hours without trouble.

 

It's a great principle but driving still remains a very personal experience, and not many want to give up full control just yet, but I can see the benefits of saving fuel

This post was last modified: 01-12-2017, 07:24 PM by fixitsan.

[TKH]

I totally agree. 

 

 

I tend to agree but AI has no feel. It is programmed to avoid collisions, so like the adaptive cruise control on cars, if you pull in front of it then it must take avoiding action. i don't know how they'll negate that for track use !

 

Impressive though it seems, is it really much better than the self driving car of ten years ago ?

 

https://www.youtube.com/watch?v=VGUWz-29gEs

 

 

These vehicles are being tested in strictly controlled and enclosed situations. The biggest problem is yet to be overcome, ordinary road conditions. A track is a relatively safe place to play with this tech and we shouldn't just be taken in by the idea that a high  speed on the track equates to a well developed product, because it is 'nearly as good as a professional rider or driver'.....The driverless Golf from ten years ago was nearly as good. IMHO

 

I'm a big tech fan, but carefully controlled stunts don't seem to me to be relevant to the real world for much of the time

 

There are some other interesting problems which at the moment seem to be dealt with only in a very simple manner - why they should be programmed to kill.... https://www.technologyreview.com/s/54262...d-to-kill/

 

 

 

 

When I was studying  to be clevererer than wot I was before i did have to do some research which looked at control systems, and what i found was a 1950's research paper by a big US university which developed road trains made from radio interconnected cars, for highway use. The driver still steered but the cars maintained a gap of about 12 inches between themselves at highway speeds automatically  - The first car punched a hole in the air and the following cars made quite a significant fuel saving....AFAIR they had about 20 cars going around a racing oval for hours without trouble.

 

It's a great principle but driving still remains a very personal experience, and not many want to give up full control just yet, but I can see the benefits of saving fuel

 

I totally agree. 

[Studley Ramrod]

Anyone watch Guy Martin take on the AI car on Ch4? Really interesting programme (even better using subtitles  ). Still a way to go , especially when AI hadn't a seat of the pants feel when the tyres cooled off and it launched into the kitty litter.....................

 

Kitty litter on the first bend too.

 

Despite it being a controlled stunt it was hugely entertaining.

[peter-may]

No takers for these raffle tickets for first pillion ride round the IoM circuit then!

[Scurve]

How long before the police starting testing them !!!!!!!!!!

[marko67]

Fascinating technology. I imagine some bizarre headlines when AI vehicles go mainstream.

 

The fuel saving "train" is being discussed in Parliament for some UK motorways I believe. The aim is to minimise congestion and save fuel but the opposing safety arguments are very valid.

AI software still has a way to go... Volvo recently tested in Australia and found out that their autonomous vehicles did not recognise Kangaroos. Nobody had thought of programming which recognised things that hop.......

 

Honda, Yamaha and BMW have bikes that ride themselves. The uncrashable, lol, bike is just around the corner. I'm looking forward to autonomous cars but for bikes I think it defeats the object of riding. I'm a poor pillion at the best of times.

[ChrisG]

These vehicles are being tested in strictly controlled and enclosed situations. The biggest problem is yet to be overcome, ordinary road conditions. A track is a relatively safe place to play with this tech and we shouldn't just be taken in by the idea that a high speed on the track equates to a well developed product, because it is 'nearly as good as a professional rider or driver'.....The driverless Golf from ten years ago was nearly as good. IMHO

Its not just controlled and enclosed conditions. Google's system has done well over 3 million fully autonomous road miles now, and Tesla claim their owners have done over 100 million miles on autopilot, although that's a more basic system rather than fully autonomous.



 

The fuel saving "train" is being discussed in Parliament for some UK motorways I believe. The aim is to minimise congestion and save fuel but the opposing safety arguments are very valid.

Must admit I'm not so sure about platooning trucks compared to proper autonomous driving, it saves fuel by bunching them up really close, which could make entering or exiting a motorway a bit tricky if there's a long line of trucks all nose to tail



 

Volvo recently tested in Australia and found out that their autonomous vehicles did not recognise Kangaroos. Nobody had thought of programming which recognised things that hop.......

Kangeroos and computer can be a bit of an issue...https://www-users.cs.york.ac.uk/susan/joke/kanga.htm

 

[fixitsan]

Its not just controlled and enclosed conditions. Google's system has done well over 3 million fully autonomous road miles now, and Tesla claim their owners have done over 100 million miles on autopilot, although that's a more basic system rather than fully autonomous.

 

 

Tesla's nearly decapitated an owner when it ran underneath an articulated trailer.-> https://www.theregister.co.uk/2016/06/30...rist_dead/

 

Every year in California car makers have to report the number of human interventions needed with their systems. The numbers are interesting -> https://spectrum.ieee.org/image/Mjg2MDYzNw.jpeg

 

The respectable IEEE make a fair analysis here ->  https://spectrum.ieee.org/cars-that-thin...iving-cars

 

 

Business wants autonomous vehicles so that they can cut their labour costs down, where transport is concerned, and no doubt there will be entrepreneurs looking to offer cut price transport.

 

I don't think this is something we can rush into. There are so many potential problems that a happy go lucky approach (which seems to prevail) isn't going to work. Perhaps on motorways where there are no roundabouts or right of way issues to speak of then it might be acceptable, but on ordinary streets.....no thanks !

This post was last modified: 05-12-2017, 12:03 PM by fixitsan.

[dandywarhol]

http://www.suffolkgazette.com/news/motorhome-crash/

[fixitsan]

http://www.suffolkgazette.com/news/motorhome-crash/

 

She's not the first either !

 

You can see more people getting confused this way though, they have 'intelligent cruise control' and 'active assist' and a gps, and they might think that autonomous control is the same thing....

 

What we need is a public information advert, lets dust off Sir Robert Mark   https://www.youtube.com/watch?v=BE15EtuA6Z8 

[Studley Ramrod]

Well, if electric cars are anything to go by, the technobods won't have a fully autonomous vehicle on public highways for at least another 50 years. TF.

[ChrisG]

Tesla's nearly decapitated an owner when it ran underneath an articulated trailer.-> https://www.theregister.co.uk/2016/06/30...rist_dead/
 
Every year in California car makers have to report the number of human interventions needed with their systems. The numbers are interesting -> https://spectrum.ieee.org/image/Mjg2MDYzNw.jpeg
 
The respectable IEEE make a fair analysis here ->  https://spectrum.ieee.org/cars-that-thin...iving-cars
 
 
Business wants autonomous vehicles so that they can cut their labour costs down, where transport is concerned, and no doubt there will be entrepreneurs looking to offer cut price transport.
 
I don't think this is something we can rush into. There are so many potential problems that a happy go lucky approach (which seems to prevail) isn't going to work. Perhaps on motorways where there are no roundabouts or right of way issues to speak of then it might be acceptable, but on ordinary streets.....no thanks !

 
 
It's an interesting field certainly.  Looking at those IEEE numbers Google's managing 1 disengagement per 5000 miles on average, it'd be interesting to see how human drivers compare in terms of accident rates per mileage, a quick google is only finding data for road deaths at the moment.
 
I agree it's not there yet, but it's not far off.  The Tesla crash you mention was something to do with the system not being able to tell the difference between the sky and the white side of a truck wasn't it?  However that wasn't a fully autonomous system and in theory the driver should have still been in control, but probably trusting the system too much.  there was a really interesting TED talk from one of the Google car guys a few years ago, and I can see his point that full autonomy is safer than over reliance on driver aids. https://www.ted.com/talks/chris_urmson_h...d#t-554768
 
The first big crash of a fully autonomous car will cause a bit of a media storm, but a lot of people tend to forget that a system needs to be better than a human rather than perfect.  Loads of people say they'd never get in one, yet happily hop in an Uber with a human they know nothng about behind the wheel.
 
I think one of the biggest problems though is going to be other people, drivers and pedestrians will know that an automated car is going to yield to them so are probably going to be more inclined to step out in front of one.

 

Well, if electric cars are anything to go by, the technobods won't have a fully autonomous vehicle on public highways for at least another 50 years. TF.

I see quite a lot of electric cars daily. Loads of Tesla model S's around here, with 200+ mile range and a reasonable amount of self driving capability. Although a mate has recently bought a second hand Zoe and we do take the piss a bit that my fuel light comes on with about the same remaining range as his on a full charge



Edit:
Thinking about it, if Tesla have had one death in 100 million miles, that's about 6.25 deaths per billion km, according Wikipedia the US average is 7.1 deaths per billion km

This post was last modified: 05-12-2017, 03:41 PM by ChrisG.

[Studley Ramrod]

Which is where the technology may have to decide to either kill the pedestrian or kill the occupants of the vehicle.  And who's responsible in such a situation  ?  The notion that the vehicle would select the option of causing the least amount of death is a scary one.  Hasta la vista baby !

[ChrisG]

The "trolly problem" is one that people often bring up, but as far as I'm aware none of the current cars make any attempt at that kind of logic and will swerve if it feels it's safe to do so and if not just stand on the brakes.
https://www.theregister.co.uk/2017/11/16..._aev_bill/

Nothing to say someone won't end up adding it leglisation though, however I think that's fundamentally misunderstanding how these things work.

[fixitsan]

ChrisG that's an interesting point regarding autonomy being safer than a bunch of electronic aids. Maybe if we had 4 decades of using the same driver aids as each other we would learn an adaptive behaviour to master them ?

 

Technology moves so quickly these days, I think a lot of people who are unsure of their driver aids will just look forward to the day they can shed all responsibility

 

I have a car with adaptive cruise control which I absolutely enjoy using, but have found that I rely on the fact that it will brake for me if i happen to be looking at other road features, when the car in front brakes heavily.

 

The IEEE article raises an interesting scenario - what will happen when the autonomous car reaches a red traffic light, but is being waved through by a police officer ? I suppose the human will step in there.

 

As for completely driverless vehicles, look how many motorcycles are stolen.... won't these driverless cars fall victim to witness-less thefts and be scrapped for cash ?

 

There are so many scenarios we cannot think of yet, until they're here, and that's when the real development starts !