mixteup

Not advocating, not commenting, but I got this e-mail from Thomas Net - Looks like it's coming :

https://news.thomasnet.com/featured/...inid=221591629

Road Fan

I'm glad you pointed out that lateral awareness of peds and pedalcyclists and forward awareness are both critical. Cars can be deadly to vulnerable road users ahead and to the side. Simplifying the problem only to forward object detection is inadequate. I see alerting the driver as the minimum countermeasure, but only one step of collision prevention, and one that could be counterproductive if traffic is very complex (a "target-rich environment").

The question in the technology and what can actually be done, is time. There aren't a lot of seconds available to prevent a collision. Time for the car to compute that there is a transponder or beacon in the path, time to confirm whether the signal correlates to an object or not, time to determine if the best safety measure is to brake or to steer, time to prepare the vehicle for heavy braking or steering, and time for the brakes to scrub off speed or for the steering to deflect the car. Might want to add in time to assess road traction to adjust the estimated time when braking must begin. In case of a warning intended to motivate the driver to prevent a collision,

These timing requirements are usually long enough that the car's safety system effectiveness has to be limited in speed. The timing requirements translates to a distance requirement based at least on road speed, and there is an upper limit for each type of sensor. 5 GHz V2V transponders have pretty significant ranges, sometimes more than the 150 to 200 m of a 77 GHz radar. So the car company might have to state in the owner's manual that the car safety system cannot be effective above some speed. For peds and pedalcyclists that's probably ok in cities, but for suburban to rural areas with 50+ mph 2 lane roads (with expected additional speeding), the crazy pickup could probably still git us.

This is all set against processing speed, data bus throughput, and sensor update timing. With the coming of monstrously powerful multi-processors from the likes of Nvidia, the computing aspect might be nailed, or we can see that it will be. If it takes 10 samplings of a sensor hit on a pedestrian, the time for the sensor (a little intelligent system all by itself) to generate each new sampling certainly figures in here as well. This is where a beacon (or a transponder that also acts like a beacon) could help a lot. It should be able to determine its own position rather reliably (GPS and differential GPS) and simplify the task left to the in-car system.

Somehow cybersecurity must also be considered, at least man-in-the-middle attacks. If it is a braking or steering system, functional safety must also be followed to manage the potential for induced steering or braking errors.

So what's the point? WPHamilton mentioned "program architecture" in a previous post. I want to show some of the system architecture issues gleaned from my current experience in the collision mitigation/avoidance field and some more general topics in safety of electronic systems. I think the auto companies are coming around to understanding what this really takes, and that not to engineer the systems really carefully can actually have terrible consequences. Glossing over the downsides should now be a dead issue, at least with awareness of the Uber/Tesla problems, and with Volvo's proof that a solution exists.

Clyde1820

If it's a little something-or-other, it ought to work similarly to donning a safety light. Suitable for pedestrians, as well.

Hopefully, this sort of device-X recognition will be in many (or most) vehicles in the future. So long as the device isn't ever tied to someone's identity (and I can't see why it would ever need to be), this sort of thing makes sense. Provide simple notifications and warnings, in the cockpit, and on "auto-driving" units provide avoidance/braking mechanisms to deal with it.

Of course, there are minor downsides with portable/wearable units like that, if/when vehicles become "automatic" ... Imagine someone lining up a dozen of these things across the roadway, as a prelude to a carjacking. Wouldn't be hard to do. Lay out a dozen of them, wait for the "automatic" car to come along, and zip zip it's done.

Road Fan

The automatic car should be equipped with a means to confirm if the line of beacons is a line of individuals. Of course in the design, the relative risks of error have to be evaluated.

Hypno Toad

Took me a minute ... but I just returned from ProMat in Chicago and now I can share details for the industrial RFID array reader. Zebra is one company in this space, this is the press release from the show:
Repeating myself from earlier: this product is for industrial/warehouse applications and would not be ready for deployment on vehicles as is. However, the tech is here today and should be modifiable to use a passive (no battery and cheap) RFID tag on bikes and autonomous vehicles can 'see' them.

dynodonn 

Gawd, I'd love to use a detection device on my bike, especially if it caused a motorist's vehicle to avoid a close pass or a hit from behind. Of course, my son in law hates driving my daughter's new car, with it's autocorrect steering.

Maelochs

I t woudl be interesting to see what a car did if another car, coming head-on, crossed the center line while the first car was passing a bike on the first car's right.

gecho

I see a lot of positives for this V2V technology.

Traffic lights would be aware that a cyclist is approaching so no more dealing with faulty inductive loops or poorly aligned camera detectors. In my city after this winter 90% of the intersections I approach no longer detect me as the camera is pointed in the wrong place, no detection, no green light ... ever.

To address the near universal habit of drivers blowing past the stop line at stop signs even visibility is poor, automatic braking systems could prevent you from flying over someone's hood or getting t-boned. Ditto for vehicles backing out of driveways and people opening doors into traffic.

Daniel4

Detector beacons should be optional and free for bicycles and pedestrians but mandatory for motor vehicles.

alvaroe16

I was surprised to see so many people arguing against having a device that communicates to drivers there is a cyclist ahead. I saw some good points with regards to batteries and privacy but even then, I would personally love to have a way to let my fellow drivers that I am in the road with them. Any warning is better than nothing. This is the same reason I use a flashing light.

One idea I have been thinking for a while is using my GPS data (either Garmin or an app on my phone (Strava Beacon, etc)) to upload my location to the cloud. Currently, I use this to let my family members know where I am. I wouldn't mind if this same data is uploaded (anonymously) to a database of current cyclist on the road. It wouldn't record that Joe Doe is on the road, It would only record that a cyclist is on the road, at a certain speed.

From there, car makers can access this data and populate their maps. The reality is that there are a lot of drivers currently using Google Maps, Apple Maps, Android Car or AppleCarplay while they drive. The number of drivers and cars with this service will only increase in the future. These apps can communicate to the database and populate hazards on the road. An app called Waze (now owned by Google) uses driver source data to notify drivers that a police or hazard is X miles ahead - so the technology to update hazards on maps exist.

SUMMARY:
- Your GPS/cellphone communicates your location and speed - anonymously
- Map services/Cars access database and warn drivers know that there is a cyclist on the road: "Watch out, cyclist in the road in 1 mile"

Would you be against this kind of technology? what are your concerns?

Daniel4

I heard about this a while back and the first thing I thought of was that all the cars would avoid the road where the police is located. Not generally a bad thing but you know what bad drivers do in the presence of a cop, they momentarily turn into good drivers. And then all those alternate roads will become filled with bad drivers too.

So I would have preferred that the app would not only identify where the cops and the bicycles are but also randomly and falsely input many more locations where they are too. The good driver would take caution - as he should - to drive carefully everywhere but the bad driver - not trusting the app - may eventually get nabbed, as he should.

This is because drivers should drive carefully not because they don't want to get caught, but because driving carefully is the right thing to do.

velojym

Another crutch. Car drivers will assume all bikes have working transponders, and won't even bother looking for us anymore... at least some of those who may have otherwise been looking.

Hypno Toad

Frankly, the beacon thing is a losing tech for collision avoidance. A better option is on the road today:

(All cars sold by Volvo Car with a model year of 2014 or later is equipped with City Safety Generation II)
Volvo Cars and POC to demonstrate life-saving wearable cycling tech concept at International CES 2015

Footnote: in case you didn't see the full story from the Uber autonomous vehicle fatal crash - Uber disabled emergency braking in self-driving car: U.S. agency

This post is from this thread: https://www.bikeforums.net/advocacy-...ety-flaws.html

Chris0516

ditto

Feldman

You should not have to "hardware up" to ride a ******g bicycle on the roads! I believe wholeheartedly that if driverless cars can't be programmed to not kill walkers or cyclists, that it is legitimate self defense for those of us who walk or pedal to do whatever is necessary to disable those cars whether they belong to us or not. No difference between that and kicking the crap out of the pitbull that goes after my little beagle.

Hypno Toad

Based on this post, should we go wreck every car on our block? These are the cars are kill 35,000 people in the US each year. It'd be kinda hard for autonomous vehicles to be worse than people driving cars.

Rollfast

Wow, that went down with the Ty-D-Bol Man's dinghy in record time.

Rollfast

They are assuming that Rollfast and Shelby will rise like zombies. I'm pushing for new Colsons myself.

Feldman

Well, all of those human-driven cars are driven by different people. One programmer or team of programmers would be responsible for software design for all of those autonomous cars--fewer minds directing a larger number of cars. To me it's easy to make the distinction when viewed that way.

Rollfast

Edited after comment in question was moderated.

Hypno Toad

This is just so stupid:

Also a little victim blaming from Uber?:
https://www.washingtonpost.com/trans...op-jaywalkers/

Additionally, this NOVA episode was great reporting on this topic: https://www.pbs.org/wgbh/nova/video/look-whos-driving/
EDIT - after posting ... BF nanny changes the spokeswoman's last name like it's a *naughty word*?!? must be naughty in another language.

rydabent

But some B'crats would love to, to keep track of you.

Hypno Toad

I hope your tinfoil hat isn't getting too tight.

rydabent

Personal attacks are stupid.

Hypno Toad

My apologizes, I figured your post was satirical and I responded in kind. I didn't expect anybody was buying into conspiracy theories like government tracking programs.