Ninety5rpm

Comparing the Segway to AVs is like comparing Pet Rocks to PCs.

tyrion

I and most of the people I know in tech didn't buy into the Segway hype. It was the classic solution in search of a problem. Self driving cars are not a solution in search of a problem.

I-Like-To-Bike

Correct, they are being hyped as an alleged low cost solution to the problem of taxi rides that cost too much because of the presence of a driver; or as the alleged hyped up solution to the problem of injury, death and damage that are the result of automobile collisions.

JayNYC

This is a BIKE forum, right? This thread doesn't seem to be much about bikes anymore…

I could really care less when self-driving cars come out – they will eventually. In the meantime cars are being equipped with more and more technology for collision avoidance, and I'd assume that tech uses some of the same sensors that self-driving cars use. So the question for us is how can we be more visible to those sensors? It may help now, and will help in the future as self-driving and collision avoidance tech gets deployed more and more… A lot of collision avoidance systems are based on radar/lidar, so my guess is that our goal should be to appear as clearly as possible on radar.

I was cycling through Hermosa Beach last week. I did it one day on a steel bike and a couple days later on a carbon bike. They have radar detectors set up to tell people they're speeding. It had no problem figuring out I was going over the 8mph limit when I was riding the steel bike, but had problems seeing me and reporting my speed when I was on the carbon bike. Same thing happens to fiberglass boats – they're not seen on radar very easily so boat owners have to buy "radar reflectors" and mount them on their masts and then they can be seen on radar miles away. Yeah, the radar reflectors for boats are huge compared to bikes, but we don't have to be seen miles away – just a few hundred feet away, so ours can be smaller. Ideally, the radar reflective bits would be placed inside the frame, but there could be small radar reflector modules for older bikes.

So maybe what we need are radar reflectors for our bikes that will allow radar-based systems to see us. I'm guessing if you had one and you rolled up next to a Mercedes at a stop light their sensors would go off. And what's on the high-end cars now will be on the average car in 5-10 years.

But down the road I can see where cyclists might need to have some type of active transponder on their bikes that communicates with the self-driving cars around us. Those transponders could just come standard in bike lights. Once there's some type of standard, I think we'll be safer than we are now. We just need to make sure the self-driving car can "see" us. Then if we act in a predictable manner, given that it's a computer that isn't easily distracted, it should act in a manner that's far more predictable than the average human driver, and our level of safety goes up.

Ninety5rpm

Do you still doubt that AV technology is much cheaper than the cost of a human driver?

Do you still doubt that AVs are much less likely to be involved in collisions (including with bikes) than are human driven cars?

Ninety5rpm

This issue comes up every few pages. No, cyclists don't need transponders. If cyclists needed transponders to be "seen" by AVs, then pedestrians would too. Not gonna happen. No need for it to happen anyway.

JayNYC

But wouldn't they benefit from radar reflectors (and eventually transponders)? I mean you could make the same argument about lights. You don't need them, but they can help an awful lot.

And unlike pedestrians, bikes are in the traffic lanes with cars, not just at the side of the road and at intersections. A time will come when the self-driving cars will talk to each other to coordinate movements. At that point, a transponder will be our way of joining that conversation. A transponder could tell them my planned route, my typical speed, etc. It could also tell me that a car I'm approaching does indeed "see" me, which can be really useful. Or imagine if, with a transponder, cars would lock their doors until a bike passes to avoid doorings. Knowing that the doors are locked and can't be opened would be wonderful as I cycle past stopped and parked cars.

I-Like-To-Bike

Yes.

Provide the source of your evidence that establishes the cheapness of operating vehicles with AV technology;
provide the source of your evidence that establishes that AVs are much less likely to be involved in collisions (including with bikes) than are human driven cars.
Do not reference your personal calculations and/or guesswork.

Ninety5rpm

They don't have to be cheap. They just have to be cheaper than the cost of a human driver, which is about $20/hour.

The dearth of crashes the Waymo cars have been in after millions of miles of driving is ample evidence of their safety.

Ninety5rpm

Notwithstanding Musk's dubious claim that he might be able to pull it off without Lidar, everyone actually doing Level 4/5 is using Lidar, which is light. No need for radar reflection to be seen. The transponder idea is a solution looking for a problem.

As to the v2v communications, I think that can bring only marginal benefit not worth the complexity and cost. Because it's a chicken egg problem, they have to figure out how to make them work without v2v communications anyway. And once they do that, v2v is of marginal use.

Maelochs

I have raised a couple of these issues in other threads here ... and was roundly condemned. Around here, you are only allowed to think certain thoughts freely.

That auto-door-lock idea is interesting but it will need some tweaking. Imagine being trapped in a burning car after hitting a bicycle ... the rider runs away but leaves the bike next to your car and you can't get out.

I suggested that it only made sense that every vehicle on the road would communicate ... I was told that it was impossible ... by people using several dozen makes of computer manufactured over a decade, in many countries, using a wide variety of software, and all connecting to the Internet via different means ... and some of them on their phones, no doubt.

No, that degree if interconnection was just not possible, i was told ... by people on the World Wide Web.

I was talking about manufacturers simply sticking an RFID chip in their frames, something people in other posts on other pages had actually asked for, so that stolen bikes could be tracked. Considering how many people bring their cell phones when they ride and how many use a GPS program when they ride ... but I was told that that would be surrendering too much privacy.

I still think it will be offered as a sales feature and a lot of riders would like it---track a stolen bike, and interact more positively with the collision sensors in cars.

But ... one does not know. Everything in the world could change suddenly and everything we think now could prove to be misguided. Oh, well.

Enjoy the ride, I say.

tyrion

It's not that complicated. It just has to detect something moving above a certain speed into the path of the opened door. Static things - a tight parking spot - shouldn't activate the lock. This feature is useful for more than just saving bicyclists, it can save the door opening person from getting smacked by a passing car.

New Audi A8 automatically locks doors if there?s a chance a cyclist might be doored | road.cc

Ninety5rpm

You're free to think whatever you want. While you're at it, please answer this question. For a system that can detect a rider's leg moving up and down, and every other little thing about the cyclist, what additional useful information from a transponder could possibly be gained?

noisebeam

The AV systems use light waves not radio waves. Anyway in dozens of cases I;ve never had radar overlook me when on CF bike. The frame is a tiny bit relative to the human on it.

I-Like-To-Bike

IOW zero credible evidence or facts that can lead to any credible predictions about the actual cost of owning/operating a rental driver less vehicle business and no solid evidence about how driver less cars will perform while operating as taxi service rather than being restricted to closely controlled test routes, closely controlled release of data about said tests.

And of course no evidence about how well they will operate on city streets throughout a metropolitan area without any real time human control or oversight while in operation.

SHBR

I see we have stepped to new lows.

Rather than discuss issues related to the original topic, we should resort to personal attacks.

Stay classy.

Transponders are here, most of us carry one, its called a mobile phone.

The I-phone will be obsolete, next it will be the I-chip.

Eventually this won't be optional, its a precursor of things to come.

Rollfast

I'll disable it then. The last thing I need is a bunch of calls when I'm on my bike enjoying life.

Maelochs

So you are saying, since I can hear cars coming I shouldn't look for cars?

A chiop in a bike could feed a lot of information a lot more quickly using less processing power, allowing the car's computer to operate more efficentl;y.

I am no tin favor or opposed. Myself as SHBR mentioned, I use a phone to track my rides ( I don;'t get calls and would ignore them if I did.) But the idea of being connected is not odd ...as I mentioned.But whatever ... after a while these threads just become a place where people can take out their frustrations on each other, and build up new ones.

If I say AVs are good people who like AVs say how smart I am. i say chips might be inserted in bikes, people who don't like chips say I am an idiot. if the same person likes AVs but hates chips ... what,. am i an idiot/savant?

It gets ridiculous.

If we sought greater understanding instead of personal validation we might actually discuss instead of arguing.

JayNYC

If we chip our dogs, chipping our bikes isn't a bad idea. Though, unlike dogs, bikes do have serial numbers engraved on them.

Not sure RFID is the right tech for getting cars to sense us. Most RFID only has a range of 1 meter, though newer generation RFID has a bigger range. But RFID is about IDing stuff. I don't need them to know my name. I just need to see me as a bike or as some sort of thing to avoid.

And the day when vehicles will talk to each other is coming. It's completely possible. And just like there are different versions of HTTP and SSL/TLS operating simultaneously, there will be different versions of the car talk protocol operating simultaneously.

And as far as being trapped – only lock one side of the car, and only for a short period of time. After which there's a loud beeping and flashing light during your over-ride of the warning.

Interesting. This is what I think we should be focusing on. If a bike collision system is on the A8 today, it'll be available on Golfs & Jetta's in ±5 years, and standard issue in ±10 years. And other manufacturers will be following VW's lead.

So the question is how can we up our chances of being seen by such systems? AFAIK these systems are mostly based off radar (LIDAR is more for self-driving cars). So how can we increase our visibility to radar? Are carbon bikes seen as easily as steel bikes, etc?

[It mentions that it works for the first three minutes the car is off. I assume it also works when the engine is running.]

When you go beyond things like radar reflectors, to active transponders there's a bunch of stuff you can do… The bike can tell the car the intended or likely route the person will take. Will they be turning at the next intersection, etc. That would require routing software to be running, or it could learn your habits. If you do the same commute from work at the same time of day each day, it could give a probability that you'll be turning.

It could also tell the car things like our capacity to accelerate (if knows the grade and something about how much power you're personally capable of outputting). When the car knows the parameters of our behavior it can better react to our being there.

But it's more about the car telling us stuff. Does the car see us? Are the car doors locked so we don't have to worry about being doored? Is the car planning on turning? Does the car sense a 3rd party danger we should be aware of (like another car coming at us from an angle the car can see, but we can't see)?

Ninety5rpm

Of course not. But your hearing system is not nearly as reliable as AV vision systems absolutely need to be.

Surely you're aware of the bizarre proclivity the press and society at large has with noting whether a fatally run-over cyclist was wearing a helmet. I can see it now.

CYCLIST KILLED. NO CHIP; NO FOUL.



I'm not following. Look, either

1. AVs are able to reliably detect, recognize and avoid hitting cyclists and pedestrians without transponders and, so, transponders are pointless, OR
2. AVs are unable to reliably detect, recognize and avoid hitting cyclists and pedestrians without transponders and, so, transponders are mandatory.

Further, adding transponder detection to the system (what frequency? what direction? How to interpret? Who decides?) adds complexity to the system (more complexity = more opportunity for failure) for dubious benefit.

Sorry, but it's a ridiculous idea.

mr_bill

V2X is what is called an orthogonal issue.

Quite useful frankly.

Especially since there are opaque objects in the world. (And clearly obtuse objects in the world too.)

-mr. bill

Ninety5rpm

Unless the "information" is definitive, it's basically useless for collision avoidance. If you usually go left, but sometimes go straight, that's not information the AV can use to avoid a collision. Similarly, if you usually follow the directions from your routing software, but not always, that too is not information that can be relied upon. The AV still needs to be prepared for you to do either, which it can do just as well without having a clue as to what you usually do or what your routing software indicates you will probably do.

As a programmer, I just can't see what information a transponder on a bike can provide that an AV can utilize to improve safety.

If you want to make an anti-theft device, sure, but that's an entirely different subject for a different forum, let alone thread.

Ninety5rpm

V2X is V2V (vehicle-to-vehicle) plus V2I (Vehicle-to-infrastructure). These are optimizations that may be added to the system later, but they are not safety or anti-collision methods. It's to allow vehicles to safely tailgate each, more efficiently maneuver around each other, to know when an intersection is clear and will be clear despite lack of sight line to cross traffic, etc. Since the AV is in control, it can transmit its intentions and plans reliably. A bike transponder cannot do that for a cyclist who is control.

Again, bike transponders have no viable function here. Another example: Say an AV is coming to a smart intersection that says it's clear because there is no cross traffic. How does the intersection know there is no cross traffic? With it's own systems. Do those systems rely on bike transponders to detects bicyclists? They could, of course, but they better not rely on them, or they couldn't detect bicyclists without transponders. And, again, if it can detect bicyclists without transponders, what's the point of the bike transponder?

JayNYC

Also a programmer and have taught stats. And actually, probabilities are something the AV can absolutely use. They're making assumptions about other people's probable actions several times a second. Even probable confirmations of those assumptions can help them make better assumptions. And those assumptions can be fed back to the bike. "Car A assumes you're continuing straight". My assumption is when we get to the point where this is happening (15? years) we'll have signaling capabilities on our bikes, and a simple interface that shows us the environment around us and our course through it. Yes, you can change your mind (you're human), but you'll have feedback about your expected course from the cars around you and can use those assumptions to make your course change in a way that keeps everyone safe.

Ninety5rpm

Again, seems like a lot of complexity for marginal benefit.