Triumph Torque Logo
 
Saturday, 23 March 2019 Register | Login

Message Board

NextRSSPreviousSocial Forums > The Hinckley Arms > Can self driving cars see bikes?
Info
.
Message
Bez
Buckinghamshire,
United Kingdom

Posts: 422
Sprint 900


icon11 Feb 2019 13:53
https://www.youtube.com/watch?v=V4U672QM...

With the Government giving the go ahead for tests of autonomous cars being tested on UK roads and Addison Lee to launch driverless taxi service in the capital by 2021 as part of the government's pilot scheme this clip is interesting.

It is testing Version 9 of Tesla's autopilot software and its bike detection seems to be a bit hit and miss on motorcycles filtering trough traffic. It seems to confuse bikes with cars at times and can miss the bike completely if traveling at speed.
-
Advertisement
Remove adverts by upgrading to a premier account
-
John Nelson
Coventry,
United Kingdom

Posts: 3,277
Speedmaster


icon11 Feb 2019 16:41
BMF and, I think, MAG are involved with the work that is being done setting the standards that self driving cars will have to work to. It is accepted that there are still issues with the sensors and software used and particularly that motorcycles are something that they don't work well with yet.

So long as they keep developing and don't turn the systems loose on the roads around here before it does work well with bikes I don't mind too much.

We do need to make sure that they don't even think about solving the problem by banning bikes off those classes of highways where the self driving car would be easiest to implement.
-
Bez
Buckinghamshire,
United Kingdom

Posts: 422
Sprint 900


icon11 Feb 2019 17:49
Good point well made :thumbup:

I can not imagine how an insurance claim would pan out in the event of an autonomous car changing course or lanes and taking out a motorcyclist because it failed to detect it.

Like you say hopefully the technology will improve but if it doesn't is important to make sure the easy option to get round it is not to restrict motorcycles through legislation or make filtering illegal.
-
-Rob NMTB -
Worcestershire,
United Kingdom

Posts: 245
Enthusiast


icon11 Feb 2019 18:20
49

I think liability issues need to be sorted before any self-driving vehicle is let anywhere near a public road.

Not that it's much consolation, if you've just been smeared across the road
-
Bouncy Pete
Bedfordshire,
United Kingdom

Posts: 691
Speed Triple 955i (02-04)


icon11 Feb 2019 21:38
Bez:
and can miss the bike completely if traveling at speed.


Sorry to sound like I'm supporting the car but If you mean the bike is travelling at speed then surely that's the fault of the bike, not the car.

If you mean the car is travelling at speed then the bike should no longer be filtering, again surely that's the fault of the bike, not the car.

https://www.youtube.com/watch?v=RjGe0Gii...

In that video it seems to detect a potential accident better than a human. Watch at 0:33 you can hear it's warning tone detecting the slowing car in front of the car it's actually travelling behind. It's also worth watching at 2:53

Personally I think this technology will have the same reception as flying gets. Millions of safe flights but one accident and it makes front pager news. Despite millions and millions of miles travelled safely there will always be someone who tells you very loudly that they are unsafe.

In 2017 1,793 people died on the Uk's roads and 170,993 were injured. As far as I'm concerned the sooner we take away human error the sooner those figures will fall.
-
Danny
SE Wales,
United Kingdom

Posts: 5,300
America (02->)


icon11 Feb 2019 21:52
I like driving and would never want a car/vehicle to drive for me and when I get my licence back I will start driving again.
-
Bouncy Pete
Bedfordshire,
United Kingdom

Posts: 691
Speed Triple 955i (02-04)


icon12 Feb 2019 06:03
Bouncy Pete:
Bez:
and can miss the bike completely if traveling at speed.


Sorry to sound like I'm supporting the car but If you mean the bike is travelling at speed then surely that's the fault of the bike, not the car.

If you mean the car is travelling at speed then the bike should no longer be filtering, again surely that's the fault of the bike, not the car.

https://www.youtube.com/watch?v=RjGe0Gii...

In that video it seems to detect a potential accident better than a human. Watch at 0:33 you can hear it's warning tone detecting the slowing car in front of the car it's actually travelling behind. It's also worth watching at 2:53

Personally I think this technology will have the same reception as flying gets. Millions of safe flights but one accident and it makes front pager news. Despite millions and millions of miles travelled safely there will always be someone who tells you very loudly that they are unsafe.

In 2017 1,793 people died on the Uk's roads and 170,993 were injured. As far as I'm concerned the sooner we take away human error the sooner those figures will fall.


What I should have pointed out was that in the original YouTube video we see a Tesla travelling at almost 60 MPH being overtaken by bikes which must have been travelling at 70 MPH even though the traffic in the next lane was in some cases virtually stationary and yet this thread implies the car travelling on auto pilot is the biggest threat. It's not, it's all the other cars.

Look at the way humans will swap lanes just to gain a few feet in queuing traffic, autonomous cars do not do this.

Judging by the YouTube video I posted I'd also say that the Tesla may well have detected the bikes but hadn't been programmed to display that sort of threat.
-
lord homertrix stand up
derbyshire,
United Kingdom

Posts: 24,598
Daytona 955i (99-01)


icon12 Feb 2019 07:01
If someone in power get their palm greased it will happen even if it's a risk to bikes, probly just say banned from autopilot areas unless bike fitted with a transducer. The fun part is all the God dammed cyclist that get mowed down.:grin:
-
Bez
Buckinghamshire,
United Kingdom

Posts: 422
Sprint 900


icon12 Feb 2019 10:03
Bouncy Pete:
Bez:
and can miss the bike completely if traveling at speed.


Sorry to sound like I'm supporting the car but If you mean the bike is travelling at speed then surely that's the fault of the bike, not the car.

If you mean the car is travelling at speed then the bike should no longer be filtering, again surely that's the fault of the bike, not the car.

https://www.youtube.com/watch?v=RjGe0Gii...

In that video it seems to detect a potential accident better than a human. Watch at 0:33 you can hear it's warning tone detecting the slowing car in front of the car it's actually travelling behind. It's also worth watching at 2:53

Personally I think this technology will have the same reception as flying gets. Millions of safe flights but one accident and it makes front pager news. Despite millions and millions of miles travelled safely there will always be someone who tells you very loudly that they are unsafe.

In 2017 1,793 people died on the Uk's roads and 170,993 were injured. As far as I'm concerned the sooner we take away human error the sooner those figures will fall.


I'm certainly no Luddite and agree that most advances in technology lead to great improvements. The overall system seems good but we are looking at specifically at its ability to see and detect motorcycles.

There is no point in replacing the flawed human response of seeing a bike by a computer that also makes the same mistakes and that is why it is important that motorcyclist don't get overlooked in this before it is rolled out.

You point out that the bikes in the clip should not have been traveling at speed or filtering in the conditions but these autopilot systems are going to have to work in real world situations. If everyone rode and drove to the letter of the Highway Code and adhered to every speed limit then those high accident and fatality figures would be dramatically reduced, but that is not how real life works.

As has been pointed out what we dont want is a 'good enough is good enough' system being rolled out and if there is a problem with bikes getting hit then the knee jerk reaction is to clamp down on motorcycles freedom to use all of the road network.
-
lord homertrix stand up
derbyshire,
United Kingdom

Posts: 24,598
Daytona 955i (99-01)


icon12 Feb 2019 10:19
What if steptoes horse got splattered under a autopilot bus
-
Bez
Buckinghamshire,
United Kingdom

Posts: 422
Sprint 900


icon12 Feb 2019 10:47
lord homertrix stand up:
The fun part is all the God dammed cyclist that get mowed down.:grin:


:grin::speechless:
-
Bouncy Pete
Bedfordshire,
United Kingdom

Posts: 691
Speed Triple 955i (02-04)


icon12 Feb 2019 18:37
I totally agree that it's important that technology doesn't cause accidents to any road user, especially the vulnerable road users.

The trouble is, I can't see the problem. In the original video the Tesla drives straight and true and at no point does it impede the motorcycle, nor is it likely to, detected or otherwise. I've not seen anything that has suggested the Tesla is a threat to the bikes. The only danger to the bikes is the near stationary cars they are passing at 70 MPH.

As I've already said, autonomous cars don't keep swapping lanes just because one lane is moving faster than the other. Humans do and they are the danger to motorcycles.

Personally I think the car actually does detect the bike but doesn't display everything it sees. Otherwise it would be an unnerving distraction thus it generally only displays potential threats and is re-assuring the driver it's seen them.

A while ago there was a BBC news article that had a video made at the Thatcham research center. The video showed a Tesla crashing into a tent that was supposed to represent a broken down van. the Tesla was following a car which swerved around the tent but the Tesla drove into the tent. I have two issues with that video.

1 - Most humans would have had the same crash.
2 - A tent doesn't have the same radar reflection a a solid object so obviously it's not going to reproduce the correct response. Seriously, would you want a full blown emergency stop if a Tesco's carrier bag blows in front of your car?
-
Bez
Buckinghamshire,
United Kingdom

Posts: 422
Sprint 900


icon12 Feb 2019 20:39
The 'Tent test' does sound like a load of nonsense :lol: Probably someone setting the car up to fail. Presumably the car has to be programed to recognize objects that it is likely to encounter on the roads and 'tent' probably wasn't high on the list.

I take your point about autonomous cars not randomly lane swapping.

At the end of the clip the guy suggests that it might be down the current 360 degree camera system not having a high enough frame per second capture for the reason it sometimes misses the fast moving bikes. The cars individual cameras capture 25 frames per second. A iphone can shoot 240fps so as he points out an upgrade on this might help.

However slightly worrying is that in the clip below the car does not see the bike directly in front of it and the bike doesn't seem to be traveling at speed (its a Harley Davidson)

https://www.youtube.com/watch?v=RXkoag9V...
-
Grumpy
dunstable,
United Kingdom

Posts: 137
Daytona T595


icon12 Feb 2019 21:24
Bouncy Pete:
I totally agree that it's important that technology doesn't cause accidents to any road user, especially the vulnerable road users.

The trouble is, I can't see the problem. In the original video the Tesla drives straight and true and at no point does it impede the motorcycle, nor is it likely to, detected or otherwise. I've not seen anything that has suggested the Tesla is a threat to the bikes. The only danger to the bikes is the near stationary cars they are passing at 70 MPH.

As I've already said, autonomous cars don't keep swapping lanes just because one lane is moving faster than the other. Humans do and they are the danger to motorcycles.

Personally I think the car actually does detect the bike but doesn't display everything it sees. Otherwise it would be an unnerving distraction thus it generally only displays potential threats and is re-assuring the driver it's seen them.

A while ago there was a BBC news article that had a video made at the Thatcham research center. The video showed a Tesla crashing into a tent that was supposed to represent a broken down van. the Tesla was following a car which swerved around the tent but the Tesla drove into the tent. I have two issues with that video.

1 - Most humans would have had the same crash.
2 - A tent doesn't have the same radar reflection a a solid object so obviously it's not going to reproduce the correct response. Seriously, would you want a full blown emergency stop if a Tesco's carrier bag blows in front of your car?

1 Why would " most humans " have had the same crash ? If the car your following swerves round a obstruction why wouldn't you ?
2 a tent might not have the same radar reflection as a car but it might be quite similar to that of a pedestrian or cyclist or biker or anything else soft nd squidgy
-
Bouncy Pete
Bedfordshire,
United Kingdom

Posts: 691
Speed Triple 955i (02-04)


icon12 Feb 2019 21:54
Grumpy:
Why would " most humans " have had the same crash ? If the car your following swerves round a obstruction why wouldn't you ?


Watch this video from 0:33 https://www.youtube.com/watch?v=RjGe0Gii...

Grumpy:
a tent might not have the same radar reflection as a car but it might be quite similar to that of a pedestrian or cyclist or biker or anything else soft nd squidgy


At 2:53 it detects a soft squishy animal is about to move into the cars path and it takes avoiding action.
-
Bez
Buckinghamshire,
United Kingdom

Posts: 422
Sprint 900


icon25 Feb 2019 18:13
https://www.morebikes.co.uk/56239/hondas...

Looks like Honda's Goldwing and others are now also adopting this technology.
-
NextRSSPreviousSocial Forums > The Hinckley Arms > Can self driving cars see bikes?
-
Forum Jump:
-
Moderators: ~Bluelabel~ Boycie Homertrix
Who's Online?
Guests:76
Members:0
Premiers:0

top
Triumph Torque
- About Us
- Community Guidelines
- Contact Us
- Statistics
Site Information
- Site Help
- Site Status
- Site Updates
Member Information
- Login / Register
- Merchandise
- Obituaries