Advocacy & Safety - Cyclist/Vehicle relations in the age of Robotic Driving

Last night I was cycling home from work. At one point I needed to make a left-hand turn across a 4-lane road with no turn lane, so in preparation for doing so I took the right lane when there was no traffic in it, about 1/2 mile before the turn. My normal plan is to continue in the right lane until there is light traffic in the left lane, at which point I signal, switch into the left lane, and slow or stop until it's safe to cross the two oncoming traffic lanes.

This time it was different though. As I got closer to the turn, I noticed some headlights coming up behind me rapidly. The road is 30mph but this person was coming up much faster than that. I would say it was going at least 50, maybe even 60mph, judging by how fast it approached me. I got scared and started thinking about veering into the gutter, not sure if this person even saw me (which they should have, I had just changed my rear light batteries the day before, and at the current level of charge I should be visible for almost a whole mile still). I guess they eventually did manage to see me, because they finally slowed down and then sat on my tail until they could merge into the left lane. When they did that they layed on the horn as they sped by me. I turned to look and saw it was some blonde driving a gigantic SUV, frantically doing something on a cell phone with her right hand and holding the horn with her left.

I was annoyed. I rarely give drivers a one-fingered salute, but last night I was feeling generous and gave her two. I just couldn't believe how selfish someone has to be to become that annoyed that my presence wouldn't allow her to go way past the speed limit, passing people on the right, while texting or whatever the hell she was doing with the phone.

I have no doubt that we all have stories like this. Something like this happens to me several times a week, where a driver expresses anger that I am cycling on the road. Now, since I am not a mind-reader, I cannot claim to know why they are annoyed at us with certainty. But judging from comments I have heard from colleagues and opinions voiced on many internet forums, universally they seem to be annoyed because we subtracted several seconds from their day by having to force them to pay more attention to the road and slow down to go around us.

As a software developer, over my career I have witnessed a trend towards automating virtually anything that humans can do that doesn't require creativity. Back when I was in college I wrote a paper on how eventually cars will become automated and controlled with software so that people no longer have to drive them. It's not a far-fetched idea; after all, almost all airplanes are mostly automated now. The Space Shuttle launch and landing are also both completely automated, with the exception of a single man-controlled operation during landing (a human must lower the landing gear; since the landing gear on the shuttle cannot be retracted, if they were lowered too early the shuttle would burn up, and thus they decided it was too difficult a task to trust to 1970's automation). Many science-fiction novels and movies often display this concept as well.

But it always seemed like a far-off idea that may not happen in our lifetimes... or is it?

http://spectrum.ieee.org/automaton/robotics/artificial-intelligence/how-google-self-driving-car-works



Once a secret project, Google's autonomous vehicles are now out in the open, quite literally, with the company test-driving them on public roads and, on one occasion, even inviting people to ride inside one of the robot cars as it raced around a closed course.

Google's fleet of robotic Toyota Priuses has now logged more than 190,000 miles (about 300,000 kilometers), driving in city traffic, busy highways, and mountainous roads with only occasional human intervention. The project is still far from becoming commercially viable, but Google has set up a demonstration system on its campus, using driverless golf carts, which points to how the technology could change transportation even in the near future.


This is earth-shattering in my opinion. Not only for drivers, but for cyclists as well. Since most drivers are annoyed that we slowed them down slightly, and in my experience it's usually their own fault for not paying enough attention in the first place, it seems to me that if there's an automated car that handles the driving for them, then the majority of drivers complaints about cyclists can become completely nullified.

Think about it; a car that doesn't get confused and knows exactly how to react when there's a cyclist on the road. The best part is that the people in the car probably won't care in the least; they'll be busy texting or watching TV or whatever the hell it is that people want to do in cars if they don't have to actually pay attention to the road anymore.

Does anyone else think that as we transition towards automated cars, a new era of driver/cyclist cooperation will be unleashed? How far off is it? 5 years? 10?

Is there an active part we can play in the development of these systems? Should we attempt to contact Google and form some sort of advocacy group to let them know that they should take road cyclists dead serious in their software? They showed the software recognizing a cyclist in the video in that link, and they said that they've driven these cars thousands of miles through San Francisco, where there's lots of cyclist activity already, so maybe they already know exactly how to work with road cyclists?

I'm excited. I look forward to the day when "I didn't see him" and "I was texting" are phrases that no longer terrify the cycling community.

Does anyone else think that as we transition towards automated cars, a new era of driver/cyclist cooperation will be unleashed? How far off is it? 5 years? 10?

Yes. I think this will finally make vehicular cycling fully practical for all of us, and it will reduce the number of collisions on the road overall. Robot cars will probably also change how we deal with cars and may even change our whole notion of ownership of cars.

How soon? That is hard to say... Google has done and is continuing to do extensive testing on various robot cars, and has persuaded Nevada to change their state laws so robot cars can be legally used on the roads there, AND humans inside such vehicles can use electronic devices.

The hardware is almost there, the testing is being done, it is only a matter of time.

The downside is that many motorists believe that "driving is fun," and are not likely to want to purchase a robot vehicle. (There is tremendous association in this country between one's status, ego and the vehicle one owns, that the situation to overcome this social enigma alone will be difficult)

I tend to think that over the next 20 or so years, fuel for current vehicles will become increasingly expensive and at some point the only vehicles that will be available to run on "new fuel" will be robot cars, and then a transition to robot cars will take place.

We'll probably still have road raging motorists, but they'll be passengers yelling out the window instead of drivers who could control their car as a weapon.


The downside is that many motorists believe that "driving is fun," and are not likely to want to purchase a robot vehicle. (There is tremendous association in this country between one's status, ego and the vehicle one owns, that the situation to overcome this social enigma alone will be difficult)
Unfortunately, this type of person is also the type most likely to hate cyclists.

We'll probably still have road raging motorists, but they'll be passengers yelling out the window instead of drivers who could control their car as a weapon.

Let 'em. Words have ceased being able to hurt me a long time ago.

Reality is, by the time they can get this by the lawers, we'll be out of oil.

A robotic car will free up the driver to use both hands to throw stuff at us.
Also, drivers will demand an override if the cars refuse to pass a bike within 6 inches.

Now, throw in to the equation the hacker who hates cyclists.

We'll probably still have road raging motorists, but they'll be passengers yelling out the window instead of drivers who could control their car as a weapon.
.

I don't think so - I can't ever recall being yelled at by a passenger in a cab, bus or a limo, which would be more or less the same thing. I think that the road range that we see comes from the fact that driving a motor vehicle is a stressful activity, and when someone under stress encounters an obstacle, they tend to react badly. As a passenger in an autonomously piloted vehicle, I believe that the stress levels will be much lower, and the likelihood of road rage will be much lower.

I don't think so - I can't ever recall being yelled at by a passenger in a cab, bus or a limo, which would be more or less the same thing. I think that the road range that we see comes from the fact that driving a motor vehicle is a stressful activity, and when someone under stress encounters an obstacle, they tend to react badly. As a passenger in an autonomously piloted vehicle, I believe that the stress levels will be much lower, and the likelihood of road rage will be much lower.

While I agree with you regarding the potential for lower stress in a driven vehicle vice being the driver... I have to chuckle at you never being harassed by passengers of a vehicle.

I once had several cups of beer thrown at me by passengers of a party bus that was leaving a local football game. I was only hit by some splashes of the beer, not directly... but certainly I was the "target" of several "passengers."

On the plus side, I have also been "catcalled" by female passengers in passing cars... of course this was many many years ago when I was young and "studly." (us "older gents" don't get "catcalled" much anymore... guess I should leave the beer belly off the handlebars... )

(us "older gents" don't get "catcalled" much anymore... guess I should leave the beer belly off the handlebars... )

C'mon Gene. You need to get back to Europe more. Even a guy like me can get some attention from the French...

I would still be worried because the technology will never be there. Roads are dynamic and constantly changing each day. The two companies who are responsible for making the maps for GPS use human drivers and they are doing a terrible job! All the mistakes I discovered on the road that would have driven me over a cliff and in one case off a bridge are just part of human error involved in map making. Yes it is humans and not robots making the maps on our GPS's and all units have advisories warning users not to depend on them 100% of the time and use caution. They are not responsible if the GPS tells you to drive down a one way street in the opposite direction. Google GPS related accidents and you'll see page afer page.

I use a GPS all the time and have lost signals in New York City skyscrapers, rain storms and forest.

The space shuttle is not a good example. It lands in an area void of trees, tall buidlings with good weather. Plus it's well maintained with experts checking all the systems. Do you think a 15 year old car badly abused is going to have all thoese sensors working properly?

I would still be worried because the technology will never be there.
The technology already is there (http://techcrunch.com/2010/10/09/google-automated-cars/).



Do you think a 15 year old car badly abused is going to have all thoese sensors working properly?
I think it's entirely possible to make the sensors work better, on average, than the current human sensors are working.

I never thought about it in terms of for the cyclists! Welp here's to hoping we run out of oil so the car companies let it happen...

I would still be worried because the technology will never be there. Roads are dynamic and constantly changing each day. The two companies who are responsible for making the maps for GPS use human drivers and they are doing a terrible job! All the mistakes I discovered on the road that would have driven me over a cliff and in one case off a bridge are just part of human error involved in map making. Yes it is humans and not robots making the maps on our GPS's and all units have advisories warning users not to depend on them 100% of the time and use caution. They are not responsible if the GPS tells you to drive down a one way street in the opposite direction. Google GPS related accidents and you'll see page afer page.

I use a GPS all the time and have lost signals in New York City skyscrapers, rain storms and forest.

The space shuttle is not a good example. It lands in an area void of trees, tall buidlings with good weather. Plus it's well maintained with experts checking all the systems. Do you think a 15 year old car badly abused is going to have all thoese sensors working properly?

The systems do not rely fully on GPS, but only use GPS as a general guide as to how to get somewhere, otherwise they "see" the road and hazards and drive according to what they see, with sensors that don't blink, are not distracted, don't doze off, and don't have an agenda.

I get the mail I deliver put into order my a machine, controlled by a computer. I get a few mistakes a day in the letters, and few mistakes in the magazines. Every time I find one I think about cars being controlled by someone other than a human driver. Remember, the cars we're talking about already being tested are being set up by the experts that designed them. When they are all over the place by the millions, they will be set up by the average technician.

I get the mail I deliver put into order my a machine, controlled by a computer. I get a few mistakes a day in the letters, and few mistakes in the magazines. Every time I find one I think about cars being controlled by someone other than a human driver. Remember, the cars we're talking about already being tested are being set up by the experts that designed them. When they are all over the place by the millions, they will be set up by the average technician.

As a software engineer it is usually my experience that machines screw up because someone cheaped out on something. Some idiot manager somewhere decided it would take too long to test or fix a known problem, so they let it pass, because there's no real penalty for when it screws up. The customer might get mad but we already have their money and they usually just wait for a bug fix.

This will not happen with cars. Someone gets killed with one of these things, the lawyers will circle like sharks and destroy whomever is at fault.

This is one of those rare cases where it truly doesn't pay to sweep bugs under the carpet.

I just see it becoming more problematic. The only way I see for an automated car to become aware of a cyclist 'taking the lane' or any other lane position, is when cyclists have to start wearing some mechanical device, that would warn an automated car. Human drivers are like automatons anyway.

The only way I see for an automated car to become aware of a cyclist 'taking the lane' or any other lane position, is when cyclists have to start wearing some mechanical device, that would warn an automated car.

Not true. You give the computer 'eyes' of cameras and it can see a cyclist just as easily as a person. AI keeps advancing.

The Space Shuttle launch and landing are also both completely automated, with the exception of a single man-controlled operation during landing

There have never been even two shuttles in the air at the same time, and they've always had priority in the airspace when launching and landing. That means lots of physical room (pretty much the entire sky) for error on everything except final approach. If there had been 254 million more shuttles all on independent schedules, I doubt they would have trusted even current automation capability.

The systems do not rely fully on GPS, but only use GPS as a general guide as to how to get somewhere, otherwise they "see" the road and hazards and drive according to what they see, with sensors that don't blink, are not distracted, don't doze off, and don't have an agenda.

And can't make a value judgment in a crisis situation. What will it pick when all the choices result in collision, the softest target, the lowest value target, etc.? How quickly can it differentiate between a 100lb deer and a 100lb child that darts out in the road too close to stop and decide whether to hit the obstacle or risk crashing into something else? I guarantee if it wrecks avoiding the deer or flattens the child, the developers and the car manufacturer will be looking at a really expensive lawsuit.

Now, throw in to the equation the hacker who hates cyclists.

Or just a kid who figures out how to vandalize the sensors in a way that doesn't immediately cause a shutdown-worthy error condition.

Or a kid who figures out it's fun to swerve on his motorcycle at the last second and watch the robot cars go into the ditch avoiding him.

Or a group of kids in a pickup with a bunch of cardboard boxes; get in front of a robot car and drop a 3'x3' obstacle right in its path.

And can't make a value judgment in a crisis situation.
Huh? Computers can and do make value judgments all the time. Chess and poker playing computers do so, for example, and they have both beat human world champions.

http://www.popsci.com/technology/article/2011-10/googles-explains-tech-behind-their-autonomous-cars

The car is remarkable in its restraint, yielding to pedestrians during a left turn at an intersection. But what's even more remarkable, is its ability to be more "aggressive" when it needs to be. For instance, it's able to compensate when other motorists don't exactly adhere to right of way rules. In response, the car will inch forward into the intersection, signaling the other drivers its intention.




I guarantee if it wrecks avoiding the deer or flattens the child, the developers and the car manufacturer will be looking at a really expensive lawsuit.
Of course.

Human mental abilities are not on some un-reachable pedestal over computers. Computers can fly and land airplanes in crowded airspaces and have done so for years, for example. Heck, the Reaper flies combat missions on its own.

I figure even freak scenarios where the system fails or where people try to circumvent the system would still end up with a net safety increase over the current situation of people not paying attention, driving drunk, simply inept, etc. Therefore it has my full support.

http://www.popsci.com/technology/article/2011-10/googles-explains-tech-behind-their-autonomous-cars
Looks like it buzzed a cyclist at 8m15s. It seems to have adjusted its course according to a stationary obstacle model, meaning it cut back in early.

And can't make a value judgment in a crisis situation. What will it pick when all the choices result in collision, the softest target, the lowest value target, etc.? How quickly can it differentiate between a 100lb deer and a 100lb child that darts out in the road too close to stop and decide whether to hit the obstacle or risk crashing into something else? I guarantee if it wrecks avoiding the deer or flattens the child, the developers and the car manufacturer will be looking at a really expensive lawsuit.

What and humans CAN do this?

We kill 40,000 fellow humans a year based on this "keen judgement" you attribute to humans... humans that also make the decision to distract themselves with texts, drugs and rock and roll. And you are now concerned that we should not let a tireless dedicated set of computers do this task? :rolleyes:

Look I'll be the first to say that robot cars will not be perfect, but certainly what we have and do now is far far from perfect...

If you want to hold up the judgement of "perfect" against robot cars, hold that same judgement to human motorists too.

What and humans CAN do this?

We kill 40,000 fellow humans a year based on this "keen judgement" you attribute to humans... humans that also make the decision to distract themselves with texts, drugs and rock and roll. And you are now concerned that we should not let a tireless dedicated set of computers do this task? :rolleyes:

Look I'll be the first to say that robot cars will not be perfect, but certainly what we have and do now is far far from perfect...

If you want to hold up the judgement of "perfect" against robot cars, hold that same judgement to human motorists too.

Machines should be able to evolve in such a way to ensure that humans are no longer a danger to themselves and others...I wonder how these artificial decision makers will deal with that threat...hmm!?!?

THANK YOU for bringing this issue up. Yes, its coming, and probably not that far into the future.. (within a decade, I'd say)


BUT, I think, NO, robotic cars cannot NOW even under a best case scenario be trusted to SEE bicyclists and motorcyclists unless the bicycling community gets organized to be involved in the evolving discussion from the beginning.

A committee of bicyclists should be formed to create a set of goals and as quickly as possible, draft a robust RF communications protocol for bicycle/car communication/recognition - similar to the IETF RFC process.. see http://ietf.org In fact, the IETF could help get it going.

The IETF is one organization that should be a starting point because almost all telecommunications is now TCP-IP based.

For example, see this:
http://www.ietf.org/rfc/rfc4838.txt

They are way ahead of the curve..

There are existing IPv6 networking protocols that are probably already close to what is needed.

Some entity should be formed to act as the point of contact for determining and then communicating the bicycling community's side of the various issues.

http://spectrum.ieee.org/static/singularity

This will not happen with cars. Someone gets killed with one of these things, the lawyers will circle like sharks and destroy whomever is at fault.

This is one of those rare cases where it truly doesn't pay to sweep bugs under the carpet.

All crashes between autonomous cars and others will be blamed on the car company and their software. And they will get a LOT of slack, also they will fight cases with all their might. They wont give an inch. It could potentially be very bad for bicyclists if bicyclists don't get tools to warn cars internal computers of their presence in verifiable ways.

I think that car companies will get a free pass because of the "traffic crisis on American roads" They will get a very low cap stuck on medical payments (Google "usual and customary") and pain and suffering awards and lowball compensation for lost wages (factor in the statistical likelihoods of layoffs, etc.) and then injured or killed people wont be able to get lawyers, like what happens in California malpractice suits now.

Suppose the cap is $250k The lawyer gets a third to a half, and in many current cases the person's employer's health insurance company gets first dibs on the rest.. (Google "subrogation") and since these cases can take years to settle, lawyers don't want them. Basically class actions or nothing.

Life is cheap in America now. And selling new cars to people who would have kept the old ones forever if not forced to buy new sounds to me as if it would be an easy sell in Washington, I bet. Nothing is more of an emotional issue for a lot of people. Everybody with money being forced to buy a new car is not an unlikely outcome. Look at the Lightspeed controversy with GPS. Well connected companies with hidden agendas are the rule, not the exception.

"Progress is our most important product"

I always figured that our cars will end up still having manual mode, but a maximum speed of 30 mph while you're in manual mode. If you want to go faster, you have to get on a designated high speed road, where the computers and robotics take over.

THANK YOU for bringing this issue up. Yes, its coming, and probably not that far into the future.. (within a decade, I'd say)


BUT, I think, NO, robotic cars cannot NOW even under a best case scenario be trusted to SEE bicyclists and motorcyclists unless the bicycling community gets organized to be involved in the evolving discussion from the beginning.

A committee of bicyclists should be formed to create a set of goals and as quickly as possible


This is pretty much why I brought it up. While the video showed the car navigating when a single cyclist rode by on the side of the road, it led me to wonder if the software is at all programed to interact appropriately with a cyclist who has "taken the lane". Does the current programming just treat cyclists as pedestrians and not vehicles? In which case, will the car simply slow to a crawl and follow the cyclist even though it should be looking for a way to (safely) pass? If they handle this incorrectly, it could lead to greater danger for cyclists, and annoyance for the passengers.

This is something that we, as a community, must address. The sooner the better, imo. Are there any bicycle advocacy groups that are willing to take on this role?

Do you have the name(s) of the author(s) of the video you saw? People at Google or other organizations that have a research function are easy to contact. Usually there is a coressponding author and email address.

It might be good to create a forum here, or if thats not appropriate, a mailing list of interested folks, that would be a good way to talk the issue out step by step.

We need to get moving with the bike community and get a feel for what people want. If we can get a consensus that yes, its coming and we want to be included in the debate, then we can say that we speak for bicyclists and can represent the community. Thats super important, but time is just as important.

The DARPA Grand Challenge folks might know something about this.

Old automobile drivers (most cars are not compatible with drive by wire retrofits) might get frozen out of new highways and they are not going to be happy. A badly planned effort on our part that ends up having us be identified with the people who are saying they can't commute to work unless they buy new cars could be very bad..

Given the history - the streetcars, National City Lines thing, etc, I think that could happen.. Its just what we need.. NOT

If we could create some low-tech-high-tech method to warn automobile of bikes around corners... that would be very good.

I always figured that our cars will end up still having manual mode, but a maximum speed of 30 mph while you're in manual mode. If you want to go faster, you have to get on a designated high speed road, where the computers and robotics take over.

Thats sounds about right.. the thing that will piss people off is that that dedicated high speed highway is todays regular highway. They won't be able to use their older car the way they use it now.

It could end up being like the digital TV fiasco or the mess thats currently happening with Lightspeed and GPS.

There is this:http://www.youtube.com/watch?v=SqBw7XapJKk&feature=player_embedded

Do you have the name(s) of the author(s) of the video you saw? People at Google or other organizations that have a research function are easy to contact. Usually there is a coressponding author and email address.

It might be good to create a forum here, or if thats not appropriate, a mailing list of interested folks, that would be a good way to talk the issue out step by step.

We need to get moving with the bike community and get a feel for what people want. If we can get a consensus that yes, its coming and we want to be included in the debate, then we can say that we speak for bicyclists and can represent the community. Thats super important, but time is just as important.

The DARPA Grand Challenge folks might know something about this.

I find it pretty much impossible to believe that these "robot cars" would ever be widely deployed in the real world without the ability to detect cyclists operating in a wide variety of different ways. The reason is that they will always need to be able to detect other dynamic hazards in realtime anyway, such as pedestrians, animals in the road, and so on. Even if some sort of legislation were passed to keep cyclists off the roads where the robot cars operate (and that already exists, as it will likely be deployed on limited access freeways first), the cars will still need the ability to detect and avoid unknown hazards. They will have a large number of sophisticated sensor systems in order to be able to do this, probably eventually in a superhuman fashion. For example, a robot car could use infrared imaging to detect a cyclist's body heat. Last I checked, humans can't see in the infrared.

Whenever they ever make it into more complex urban environments, they will also likely be augmented with other systems such as sophisticated communication/collaboration between different vehicles present in the road at the same time. I don't see why it would be hard at all for a bike to be outfitted with similar computerized devices to alert nearby robot cars to its exact position and presence...after all, many of us ride with small computers on our bikes all the time already. In any case, robot cars will not rely on just one source of information (such as one source of sensor data, or one source of location fixes, or one collaborative communication device). In a deployed system, there will be many layers of redundant failsafe mechanisms, and I expect this to eventually make driving safer than having cars under the direct control of human beings (just like aircraft have been made safer by placing various aspects of their control in the hands of computers).

Thats sounds about right.. the thing that will piss people off is that that dedicated high speed highway is todays regular highway. They won't be able to use their older car the way they use it now.

It could end up being like the digital TV fiasco or the mess thats currently happening with Lightspeed and GPS.

What "mess that's currently happening with lightspeed and GPS"? Are you referring to the current scientific debate over the OPERA experiment at CERN? That's not really a major factor affecting the accuracy of GPS as a practical system, so much as it's a scientific curiosity with implications for our major physical theories (as the effect observed is very small). GPS obviously already accounts for lightspeed, relativity, and so on, or it wouldn't be anywhere near as accurate as it is. There were some rather ill-informed articles out about GPS supposedly not accounting for the relativity effects, but those articles were incorrect.

As for the digital TV transition, I never really saw it as a "fiasco", except that the government kept extending the deadline and dragged it out longer than they should have. Digital TV is a clearly superior technology to old analog TV, and sometimes you just have to move on and change things to make things better. Refusing to change simply because it's a small expense or inconvenience for a short time is really stupid, and I'm sick of Americans whining about that kind of thing. Another example is the failed metric system conversion: it failed purely due to this short-sighted resistance to change for its own sake. The government should have just manned up and forced the change, just like most other countries did, and we'd be better off today. This is the way with any major technological or standards shift, and it's not a problem except in the very short term when people just need to get used to the new way of doing things. So it will be for robot cars, eventually.

I find it pretty much impossible to believe that these "robot cars" would ever be widely deployed in the real world without the ability to detect cyclists operating in a wide variety of different ways. The reason is that they will always need to be able to detect other dynamic hazards in realtime anyway, such as pedestrians, animals in the road, and so on. Even if some sort of legislation were passed to keep cyclists off the roads where the robot cars operate (and that already exists, as it will likely be deployed on limited access freeways first), the cars will still need the ability to detect and avoid unknown hazards. They will have a large number of sophisticated sensor systems in order to be able to do this, probably eventually in a superhuman fashion. For example, a robot car could use infrared imaging to detect a cyclist's body heat. Last I checked, humans can't see in the infrared.


The open question in my mind is "how will the cars interact with cyclists?". Let's be honest, the vast majority of people in this country see cyclists as pedestrians who should be forced to ride on the sidewalk. It's rather ridiculous how many people I talk to who are completely unaware of the fact that cyclists are legally required to act as traffic, not pedestrians. I would not put it past the designers of these systems to not realise this fact either.

If they program the system to act like we're pedestrians, we have a problem, because that will annoy everyone and cause more friction, and may lead to new laws forcing cyclists off roads in order to facilitate robotic driving. I don't have the answer to this question; I have no idea if they are programming the cars to treat us like we are slow traffic. If they are not doing this, it is in our best interests to get this remedied as soon as possible, before it's too late.

I find it pretty much impossible to believe that these "robot cars" would ever be widely deployed in the real world without the ability to detect cyclists operating in a wide variety of different ways. The reason is that they will always need to be able to detect other dynamic hazards in realtime anyway, such as pedestrians, animals in the road, and so on. Even if some sort of legislation were passed to keep cyclists off the roads where the robot cars operate (and that already exists, as it will likely be deployed on limited access freeways first), the cars will still need the ability to detect and avoid unknown hazards. They will have a large number of sophisticated sensor systems in order to be able to do this, probably eventually in a superhuman fashion. For example, a robot car could use infrared imaging to detect a cyclist's body heat. Last I checked, humans can't see in the infrared.

Whenever they ever make it into more complex urban environments, they will also likely be augmented with other systems such as sophisticated communication/collaboration between different vehicles present in the road at the same time. I don't see why it would be hard at all for a bike to be outfitted with similar computerized devices to alert nearby robot cars to its exact position and presence...after all, many of us ride with small computers on our bikes all the time already. In any case, robot cars will not rely on just one source of information (such as one source of sensor data, or one source of location fixes, or one collaborative communication device). In a deployed system, there will be many layers of redundant failsafe mechanisms, and I expect this to eventually make driving safer than having cars under the direct control of human beings (just like aircraft have been made safer by placing various aspects of their control in the hands of computers).

Oddly enough, cyclists currently often depend on a "single sensor" right now, that doesn't see 360 degrees around the car, is not deployed at all times, and does not specifically see body heat as an indicator of an obstacle... and is often slow to react, and tends to cloud judgement decisions with a rather skewed value system... and by the comments expressed here, some folks seem to want to keep it that way.

What "mess that's currently happening with lightspeed and GPS"? Are you referring to the current scientific debate over the OPERA experiment at CERN? That's not really a major factor affecting the accuracy of GPS as a practical system, so much as it's a scientific curiosity with implications for our major physical theories (as the effect observed is very small). GPS obviously already accounts for lightspeed, relativity, and so on, or it wouldn't be anywhere near as accurate as it is. There were some rather ill-informed articles out about GPS supposedly not accounting for the relativity effects, but those articles were incorrect.

No he was referring to the LightSquared rollout of LightSpeed 4G LTE interfering with GPS signals. I am not familiar with the details of the controversy however.

No he was referring to the LightSquared rollout of LightSpeed 4G LTE interfering with GPS signals. I am not familiar with the details of the controversy however.

Gotcha, I wasn't aware of that one. It looks like some of the GPS companies have alleged interference with GPS signals from their new network. Personally, I find it hard to believe that the FCC would tolerate that if it's actually true, given that GPS has national security implications.

The open question in my mind is "how will the cars interact with cyclists?". Let's be honest, the vast majority of people in this country see cyclists as pedestrians who should be forced to ride on the sidewalk. It's rather ridiculous how many people I talk to who are completely unaware of the fact that cyclists are legally required to act as traffic, not pedestrians. I would not put it past the designers of these systems to not realise this fact either.

If they program the system to act like we're pedestrians, we have a problem, because that will annoy everyone and cause more friction, and may lead to new laws forcing cyclists off roads in order to facilitate robotic driving. I don't have the answer to this question; I have no idea if they are programming the cars to treat us like we are slow traffic. If they are not doing this, it is in our best interests to get this remedied as soon as possible, before it's too late.

Given the complexity of the various tasks associated with driving, I don't think that software designed to automatically navigate and drive a motor vehicle would be able to make many assumptions about how other entities "should" behave. There is enough unpredictable behavior out there, from cyclists, pedestrians, and motorists, that I don't think you could make many assumptions about other road users actually obeying most of the laws, for example. So I think such cars will probably sense their environment and avoid obstacles in realtime. Of course we don't want their introduction to become a political opportunity to water down cyclists' rights, so I agree with you there. But I don't believe that it's a particularly realistic scenario that cyclist rights might be watered down in order to reduce the technical challenges associated with robot cars, because similar technical challenges would need to be overcome anyway in order for robot cars to be reasonably safe for their occupants and other motor vehicles on the road. I could see it being used as a political bludgeon against cyclists, but it wouldn't be because restricting cyclists' rights would actually make sense. If anything, I would imagine an advanced robot car would be BETTER at avoiding cyclists than many human drivers are, due to advanced sensors, lack of fallible emotion, lack of aggressive impulses, lack of drunk or distracted driving, etc. A computer won't have a few too many beers and then get on the road anyway. So, basically, I don't think we have much to worry about unless the introduction of robot cars becomes an opportunity for preexisting cycling-haters to carry out an anti-cyclist agenda unrelated to the actual technical issues.

The first child that is struck by one of these will shut down the program permanantly. This is really a non-issue.

(just like aircraft have been made safer by placing various aspects of their control in the hands of computers)

Uh, yeah. Tell that to a lot of Airbus pilots. Particularly the ones that have crashed planes because the computer decided to override simple emergency maneuvers.

There are two GPS bands, the consumer band is the one that would be effected. The GPS systems that would see the most problems are the extremely sensitive DGPS systems that are currently being used in agriculture (autonomus agriculture) and construction (guiding construction equipment) Also consumer GPS euipment will probably be effected, but like Type 15 networking eqipment (802.11 b, g, n etc) THAT IS A LOW PRIORITY.

In other words, if everyone in the country has to spend $300 on a new GPS unit, their attitude is "GREAT!"


Gotcha, I wasn't aware of that one. It looks like some of the GPS companies have alleged interference with GPS signals from their new network. Personally, I find it hard to believe that the FCC would tolerate that if it's actually true, given that GPS has national security implications.

Uh, yeah. Tell that to a lot of Airbus pilots. Particularly the ones that have crashed planes because the computer decided to override simple emergency maneuvers.

On the flipside, how about the pilot who crashed Colgan Air Flight 3407 in Buffalo a few years back, because he was fatigued and overrode the computer, which if left alone would have prevented the flight from crashing?

Bugs can be fixed in faulty software. Idiots, however, will always attempt to drive fatigued, drunk, etc.

Just this morning I almost got rammed by 4 idiots who apparently have no idea that when their windows are frosted over, they should scrape them. God I hate people sometimes.

The first child that is struck by one of these will shut down the program permanantly. This is really a non-issue.

Nonsense. Children get hit by cars all the time. You don't see us banning cars now do you?

Hell, an elderly person rammed a building here in Buffalo last month, killing two people inside it. He didn't get any punishment whatsoever, and is allowed to keep driving.

Robotic driving will be such a beneficial product to so many people that there will simply be no way to prevent it from proliferating once people realise those benefits. I imagine the trucking industry will be the most vocal about it at first. Can you imagine the dollar signs in their eyes? They'll get to fire every single trucker and have goods moving on the roads 24/7, without having to deal with vacations, days off, or fatigued drivers. Then the elderly and the incapacitated (blind, deaf, quad/paraplegics) will be next. No longer will they have to rely on others to take them places, or worry if they're going to hurt someone. The drinkers are going to love it too; no more need to get a taxi or a designated driver. Get smashed, have your car drive you home. The biggest market will be the lazy, ie everyone. Now you can twitter and facebook and read the news and put on your makeup while going to work, and who cares? Your car is driving, not you.

There's no way this isn't going to happen.

I find it pretty much impossible to believe that these "robot cars" would ever be widely deployed in the real world without the ability to detect cyclists operating in a wide variety of different ways.

First and most important thing.. You're right, if "in the real world" means on our current highway system, with bicyclists, pedestrians, automobiles and trucks. But, unfortunately, as we are seeing, to many people, the bicyclists and pedestrians and even the non autonomous vehicles might be excluded - in their vision.

Also, I'm no aware of any computer or sensor that is smart enough to reliably detect pedestrians and bicycles yet with the 100% accuracy that would be needed unless the pedestrians and bicyclists were to do something, like carry some kind of signalling device, that was absolutely foolproof preferably verifiably so.

On the low end that could be something as simple as a Blueooth or IR (or both) based device that simply said "someone's here" Or it could be a device that engaged the other devices around it in a handshake in which each device would acknowledge the others presence and authenticate the others identity for the purpose of logging it.. which would mean exchange cryptographically signed credentials.. a digital signature.. Then each vehicle's driver's computer would contain a log that could be proof in court that the other vehicle "had seen it".

Thats a mechanism that I think - once implemented, could be very useful to bicyclists even in a non-autonomous situation.

The situationists talk about the dishonesty of many politicians in pretending that they are naive about political and social realities in order to argue pursuasively to people who in fact are in reality unaware of them, in order to prevent their becoming aware. I forget what they call it. (its an interesting area of thought..)

Well, the autonomous vehicle science is moving along much more rapidly than most of us realize (I know that I am constantly amazed by the changes in a year or two) The science may resolve most of the issues much more quickly than society becomes ready to deal with the potential social issues of requiring a society that pretends it can afford something like new cars universally but really can't. Its similar to the DTV "debate" and especially the health insurance "debate". he disturbing pattern seems to be to hold a lot of discussions behind closed doors between industry "stakeholders" and then come out with a fait accompli, but not admit that the entire thing has already been nailed down shut without any public input at all, then hold some public meetings, pretend to accept public comments, etc, and then say "Done" with a terrible outcome for the public.

If groups of bicyclists are on top of this situation, especially bicyclists who are also active in the computing community, which is necessary player here because politicians and lobbyists are not so technically literate..participate now, that is by far their best strategy, their best and perhaps only shot at real participation, because it is FAR more meritocratic than Washington ever could be. Computing is a language that politicians rarely speak.

To draw the same analogy with healthcare. The healthcare bill was written by Liz Fowler, a lobbyist for Anthem healthcare who had previously been a staffer for Jim Baucus, a Democratic Senator. She was allowed to do that because she was literate in the field, more literate than any of the Senate staffers.

As I understand it, actuaries have a LOT of concerns with the bill. Now, imagine iif a significant number of the public who needed to use the parts of the bill that are concerning people were acuaries..

The outcome would have been very different.. saving the country years of grief. We would have been presented with a far better deal. But we werent, because the fox was guarding the henhouse.

Lets not have any illusions.. they want to ban bicylists from highways.. Look at the history.. for example, Google "National City Lines".

I think that its quite probable that the first deployments of autonomous driven vehicles will be on sections of highway where bicyclists, pedestrians and non-autonomous vehicles are absolutely PROHIBITED.

As they sometimes say in the software industry, "Thats not a bug, its a feature". The automobile industry would like nothing better than to force the most affluent class of consumers to buy and expensive cars.

Even if some current cars (the ones that already use electronic control systems) can be retrofitted - they will find a way to make that next to impossible.

The little guy just doesn't matter any more.

The only way the bicyclists are going to be heard in this debate are by trying to get in there early, before the debate gets framed around a situation that "oh, we are so sorry" coincidentally eliminates bicycles from the picture. is to get in there and start everyone involved in the process to be thinking about bicyclists as part of their designs.. The best way to do that is to join the debate, and speak the lingo, which I am guessing in technical terms is the emerging, evolving lingo of "the Internet of things" (Google that phrase) .. or "ubiqutous computing", (yet another amazing technically innovative concept that came out of pioneering Xerox PARC research in the 70s) and the IETF "RFC" (request for comments/Internet draft) process, and the DARPA Grand Challenge type projects.. and probably others..

Whatever system evolves will undoubtably use IPv6, its what its made for.. It could be a very positive thing, by creating a way so that bicyclists can be formally and reliably recognized from a distance - (visualize heads up displays on airplanes) not just function- in a mixed car/bicycle environment, I think the best way to do this is by bicyclists using some kind of transponder.

I think the only highways that will initially go to fully automated are that small subset of highways that are currently extremely congested entering and leaving cities, like Manhattan in New York. I live in New Jersey and the highways going from NJ to NY are so very congested that its not unusual to see an hour delays when there is an accident. Its really a crisis situation. In this environment, the autonomous car will be promoted as a savior that will allow who they see, but dont say as "the people who really need to commute" (based on their incomes) to continue to live in the exurbs, and commute into the city (while reading the newspaper or eating breakfast, or even catching a nap, which will hold a lot of appeal.)

Right now, when I (rarely) find myself in the morning rush hour traffic (I'm always on a bus) I notice that most of the cars out there are brand new, expensive cars. Thats what they see, and they may argue that its not a price sensitive market. Bicyclists are no even on their radar screen.

Lets make it so, (literally) we ARE.

...

OTOH, .. Caveat..

I was thinking yesterday, like everybody else, we might be making some faulty assumptions about the economy.

If it crashes, for example, that kind of gridlock traffic could cease to be a problem for a while.

Nobody knows.

The open question in my mind is "how will the cars interact with cyclists?". Let's be honest, the vast majority of people in this country see cyclists as pedestrians who should be forced to ride on the sidewalk. It's rather ridiculous how many people I talk to who are completely unaware of the fact that cyclists are legally required to act as traffic, not pedestrians. I would not put it past the designers of these systems to not realise this fact either.

If they program the system to act like we're pedestrians, we have a problem, because that will annoy everyone and cause more friction, and may lead to new laws forcing cyclists off roads in order to facilitate robotic driving. I don't have the answer to this question; I have no idea if they are programming the cars to treat us like we are slow traffic. If they are not doing this, it is in our best interests to get this remedied as soon as possible, before it's too late.

Yes yes yes...

Exactly..