Will Tesla autopilot two hardware support V2V and V2I?

pbender

New Member
Joined
May 18, 2017
Messages
1
Location
Pahrump Nv.
Country
Country
#1
Will Tesla autopilot two hardware support V2V and V2I?
I understand that there will be hardware to support WiFi but the question is is will it support the 802.11p protocol. This protocol will allow
Dedicated Short Range Communications ( DCRC )There has been much research on this technology over the past several years and the government has played a very big role in this research. I am not an autonomous vehicle engineer but I cannot see how full autonomy (level 5) can be achieved without this type of communication.
 

orcinus

Well-Known Member
Joined
Jul 10, 2017
Messages
243
Location
Grad Zagreb
Country
Country
#2
That’s what i’d like to know too.
Back during my stint in journalism 6-7+ years ago, V2V was the cornerstone of all plans for future automotive transport.

Then it suddenly dropped from everyone’s radar.
 

Rich M

Top-Contributor
TOO Supporting Member
Joined
Jul 28, 2017
Messages
741
Location
Berks, PA
Tesla Owner
Model 3
Country
Country
#3
I would really hope that all 2-way radio hardware is modular and can be swapped out. 4G and 5G broadband are a staple for autonomous vehicles more so than V2V. 4G is also upgrading on a monthly basis. This means even if the 3's radios support every LTE frequency being used as well as every transmission mode (such as 3-way carrier aggregation) it will likely be behind the curve in a couple years, just as your two year old smartphone is already.
 

orcinus

Well-Known Member
Joined
Jul 10, 2017
Messages
243
Location
Grad Zagreb
Country
Country
#4
4G (LTE) is unreliable, especially at highway speeds and with frequent BTS handovers.
Not to mention the latencies are humongous.

I wouldn’t rely on it for any real-time messaging between vehicles, or anything that’s super-time sensitive.

It’s fine for fleet management, updates, cloud-based navigation and self driving training / ML / hints.
But for vehicles signaling intent to each other, and resolving conflicts, it’s nearly useless.

5G won’t make this any different. The higher the bandwidth, and the higher the frequency, the worse it gets.
 

JWardell

Legendary Member
Joined
May 9, 2016
Messages
3,584
Location
Boston
Tesla Owner
Model 3
Country
Country
#5
After hearing about V2V for what must be over a decade now I'm surprised there is still nothing out there. Of course standards take forever and technology changes too fast. I can see Tesla spearheading this, making their own and open sourcing it, to kick the industry in gear.
 

Rich M

Top-Contributor
TOO Supporting Member
Joined
Jul 28, 2017
Messages
741
Location
Berks, PA
Tesla Owner
Model 3
Country
Country
#6
While mobile broadband isn't instant communication, I wouldn't call it unreliable. V2V is only good for a few hundred feet due to power limitations (and don't forget Elon's anti-antenna stance on aesthetics further limiting RF performance). This distance shrinks in the city due to the noisy RF environment.
This is why I suspect manufacturers spent their money developing local sensors like cameras, radar and sonar to make split second decisions. For anything further than a car's length or two away, mobile broadband is more than responsive enough to provide general traffic flow information.
 

orcinus

Well-Known Member
Joined
Jul 10, 2017
Messages
243
Location
Grad Zagreb
Country
Country
#7
Major idea of V2V was to make it a mesh network, as opposed to simple peer-to-peer.
Few hundred feet between two cars is plenty, to get a message across from car to car, from, say, one end of a traffic jam, to another, or at crossroads.

There's a lot of issues (mesh networking in general never quite took off because routing such a network dynamically is a *very* hard problem - first home routers that use mesh networking successfully are only showing up now, thanks to cloud based machine learning). And relying on LTE is way way easier, for sure. I'm just saying that it isn't quite up to the original task V2V was meant for.

I'm sure you're right, though, and it will serve as a stop-gap measure, until someone starts tackling local/mesh V2V properly (perhaps Tesla, perhaps Volvo, perhaps someone else - both Volvo, Mercedes and a few asian car makers have been working on an implementation for a while now).

By the way, check this out for some fun:

 

Rich M

Top-Contributor
TOO Supporting Member
Joined
Jul 28, 2017
Messages
741
Location
Berks, PA
Tesla Owner
Model 3
Country
Country
#9
One more feather in the hat of mobile broadband for V2V is this: wireless carriers are already getting cozy with automakers for connected car exclusivity. They want all vehicle comms to go through your wireless provider while on the road. Direct V2V comms doesn't require a 3rd party, which means your wireless provider doesn't get paid, and we can't have that! ;)
 
4

4701

Guest
#11
I believe there is no big reason to have V2V direct connectivity.
Why not keep it all in a central server? Like Tesla does today!
Fleet can still learn about potholes, roadkills, construction, black ice
or whatever... from server on demand 24/7. Unlimited distance ahead.
Why should be there low latency communication between vehicles
close to each other? If there is no vehicle in near proximity? What then?

If vehicle in front of another vehicle has an AEB event, LED brake lights react
faster than signal can travel from one car to another. Slight extra coding of
light module to blink brake lights rapidly (less than 100ms cycles) for 3-4 times
is all that is necessary for vehicle in the back to react appropriately in case of
not satisfactory following distance.

There are many solutions for traffic safety/optimization. No comparative analysis AFAIK.
 

JWardell

Legendary Member
Joined
May 9, 2016
Messages
3,584
Location
Boston
Tesla Owner
Model 3
Country
Country
#12
Of the many features of V2V, emergency braking ahead is an important one. And that requires direct, low-latency communication. It can't easily be encoded in brake lights when they are blocked by a vehicle or two in between.

I like the brake light encoding idea as a fallback, but it would require adding extra sensors. The cameras' frame rate is much to slow. Think of the pulsed IR system commonly used by fire trucks to trigger traffic lights instead.

There's no reason why there can't be both a direct link and a cloud link. In fact there should be.
[We already have a universal V2V cloud link now: it's called Waze! Unfortunately your brain needs to make the final hop]
 
4

4701

Guest
#13
Multi-vehicle-pileup is usually the result of bad drivers. Actually every single driver that had the accident.
There are two main variables that make up the main reason for rear ending
and many-vehicle-rear-ending:

1) tailgating - aka incorrect following distance according to speed, vehicle capabilities and traction
2) reaction time - anything besides not looking front vehicle taillights is extremely bad, compared to computer.

I often ignore one of those two variables: sometimes I follow very closely and sometimes I use my phone.
But I never do those two simultaneously. This resulted in no rear-ending nor even close-calls during 8 years of my driving.
Around a quarter million of kilometers.

If all vehicles have AEB and vehicle forces minimum appropriate following distance (BMW keeps red car icon on the dash
if following distance is not acceptable) V2V has little effect on end result. First vehicle has AEB event and following vehicle
reacts to that AEB within a fraction of a second with no V2V at all.
In case of appropriate following distance, no car will crash as "appropriate following distance" already takes into account
human high latency. In case of computer, there will be lots of spare space.
 

orcinus

Well-Known Member
Joined
Jul 10, 2017
Messages
243
Location
Grad Zagreb
Country
Country
#15
Multi-vehicle-pileup is usually the result of bad drivers. Actually every single driver that had the accident.
There are two main variables that make up the main reason for rear ending
and many-vehicle-rear-ending:

1) tailgating - aka incorrect following distance according to speed, vehicle capabilities and traction
2) reaction time - anything besides not looking front vehicle taillights is extremely bad, compared to computer.

I often ignore one of those two variables: sometimes I follow very closely and sometimes I use my phone.
But I never do those two simultaneously. This resulted in no rear-ending nor even close-calls during 8 years of my driving.
Around a quarter million of kilometers.

If all vehicles have AEB and vehicle forces minimum appropriate following distance (BMW keeps red car icon on the dash
if following distance is not acceptable) V2V has little effect on end result. First vehicle has AEB event and following vehicle
reacts to that AEB within a fraction of a second with no V2V at all.
In case of appropriate following distance, no car will crash as "appropriate following distance" already takes into account
human high latency. In case of computer, there will be lots of spare space.
Still not enough.
With V2V, the vehicles at the back can start braking *immediately* when crash happens at the front of the convoy.
Without V2V, there is a necessary latency while the message ("encoded" in the braking of vehicles) is spreading like a wave through a convoy.

Also, with V2V, safe tailgating (platoons) is possible, reducing energy consumption of the platoon as a whole (increased range, better energy savings, even lesser environmental impact).
 

JWardell

Legendary Member
Joined
May 9, 2016
Messages
3,584
Location
Boston
Tesla Owner
Model 3
Country
Country
#16
Arnis, your logic is based only on all cars having AEB. That will not be true for another 40+ years. Most cars on the road are anything but brand new. Logic and engineering must be developed to coexist with old technology. V2V will start improvements now, and in 40 years when every car has AEB then only then can you argue it is not needed. By that time we probably won't have cars anywhere. Assuming the teleporters can communicate with each other of course.
 

Model34mePlease

Well-Known Member
Joined
Jun 3, 2017
Messages
274
Location
Monterey, CA
Country
Country
#17
Still not enough.
With V2V, the vehicles at the back can start braking *immediately* when crash happens at the front of the convoy.
Without V2V, there is a necessary latency while the message ("encoded" in the braking of vehicles) is spreading like a wave through a convoy.

Also, with V2V, safe tailgating (platoons) is possible, reducing energy consumption of the platoon as a whole (increased range, better energy savings, even lesser environmental impact).
Moreover, the breaking rate of sequential vehicles has to be coordinated or more collisions will be triggered.

The issue of safe following distance is important, but the total carrying capacity of a road can be increased with closer packing of vehicles. If that can be done safely at speed it is a benefit for everyone.
 

Model34mePlease

Well-Known Member
Joined
Jun 3, 2017
Messages
274
Location
Monterey, CA
Country
Country
#18
Arnis, your logic is based only on all cars having AEB. That will not be true for another 40+ years. Most cars on the road are anything but brand new. Logic and engineering must be developed to coexist with old technology. V2V will start improvements now, and in 40 years when every car has AEB then only then can you argue it is not needed. By that time we probably won't have cars anywhere. Assuming the teleporters can communicate with each other of course.
Something akin to 'platooning' might be the 'killer app' that causes a fast turnover of vehicles to new technology. Granted that the pricing is very different, look how fast smartphones took over the universe. This kind of transition over about a decade has happened for several technologies over the last 50 years: PCs, local networks, internet, etc.
 
4

4701

Guest
#19
With V2V, the vehicles at the back can start braking *immediately* when crash happens at the front of the convoy.
There will be no rear-end crash due to AEB.

Also, with V2V, safe tailgating (platoons) is possible
No, as vehicle without V2V can not follow nor tailgate. 50% V2V vehicles is not enough.

Arnis, your logic is based only on all cars having AEB. That will not be true for another 40+ years.
Less. Average lifespan of vehicle is around 20 years. AEB will be mandatory in US from 2022. AFAIK in EU it is even more popular than in US. Likely due to Euro NCAP requirements today.

Moreover, the breaking rate of sequential vehicles has to be coordinated or more collisions will be triggered.
It kinda is coordinated anyway. According to how fast vehicle is approach the one in front. But it is simple to encode that into brake light blinking as well. Though there will be no cascade effect as EVERY vehicle has it's own safe following distance buffer. The longer the chain, the softer the braking. Not harsher. It gets harsher and harsher if following distance is not appropriate.
EU vehicles today already have Brake Force Display. First vehicles came out in 2004. Though no definite standard set.
Usually one level is rear fog lights added. Next one is blinking brake lights (or only middle brake light).
And in case of big speed reduction, hazards. Sorry US, blinkers are not the same as brake lights. Everywhere else....
 

orcinus

Well-Known Member
Joined
Jul 10, 2017
Messages
243
Location
Grad Zagreb
Country
Country
#20
There will be no rear-end crash due to AEB.
You're absolutely sure that there will never ever be any rear-end crashes after every car has AEB?
Are you willing to bet your money and life on that?

Also, what happens when not every car has AEB?
What if the AEB simply cannot brake fast enough to avoid a collision?
What if the crash isn't caused by a car?
What if a semi hits the first car in the lane sideways?

No, as vehicle without V2V can not follow nor tailgate. 50% V2V vehicles is not enough.
Not relevant, as platooning vehicles all have V2V by its very definition.

It kinda is coordinated anyway. According to how fast vehicle is approach the one in front. But it is simple to encode that into brake light blinking as well. Though there will be no cascade effect as EVERY vehicle has it's own safe following distance buffer. The longer the chain, the softer the braking. Not harsher. It gets harsher and harsher if following distance is not appropriate.
Braking all vehicles precisely at the moment of the crash is still safer and leaves more margin for error than relying on propagation through the whole chain and everyone using the same distance buffer.