Tesla Autopilot Vs Volvo Autopilot (POOL)

[vortexz]

Here is a demonstration of Tesla Vs Volvo Autopilot :

Volvo autonomous car engineer calls Tesla’s Autopilot a ‘wannabe'

Source: Volvo autonomous car engineer calls Tesla’s Autopilot a ‘wannabe'

Which one do you belive won the battle ?

Start your votes

 

[TesAus]

It's always easier to have the superior system when you are "coming soon".

Every other manufacturer seems to keep benchmarking or comparing themselves against what Tesla has had in customers hands for months or years.

 

[ediot]

to be fair, the video can be very biased by cherry picking single incidents. I am a Tesla owner and i LOVE the car, and i both laughed and cringed at the volvo guys royally screwing up something so simple.

I wouldn't want to underestimate the progress that other manufacturers have made just because i "root for" Tesla. Mercedes could have come up with the near perfect algorithm for autopilot and unveil it tomorrow for all i know, and it could blow Tesla's out of the water. What i can observe tho, is how probable this is, and how quick is each manufacturer's autopilot program learning.

Tesla does have a massive, MASSIVE fleet of human driven vehicles on the road that autopilot learns things from, and they don't scale it up by hiring more people to drive the cars around like some sort of UAT, but real world drivers behaving in real world situations, which has massive advantages.

e.g. if you're a programmer, you either decide or get told by a team whether to program a car to do evasive maneuvers for a certain sudden situation. But Tesla can learn from massive real world data, where some would rather brake as hard as possible to take a certain but minimised impact (which, is the ONLY legal way to drive in many countries) whereas others would check for empty space and swerve to avoid the accident. Certain people drive more cautiously and may already have taken their foot off the throttle a bit before something becomes an issue.

What Tesla has, is the massive feed available to them for accurate, real world empirical data instead of some "expert" panel groups who can't possibly consider the whole range of situations that come up. I'm rooting for Tesla, and I think they have an autopilot program that has the best learning potential, regardless of how good they are compared to other car maker's "claims" at the moment.

TL;DR All the non-tesla car makes get the benefit of doubt for their claims of superior autopilot technology, but Tesla has objectively the best autopilot learning platform by far because we're all part of their learning already.

 

[Johan]

Poor guy. Gloating is bad form. I think Volvo will have an excellent system soon. Volvo didn't use the term "wannabe", that was the headline writer.

 

[3s-a-charm]

Poor guy. Gloating is bad form. I think Volvo will have an excellent system soon. Volvo didn't use the term "wannabe", that was the headline writer.

Agreed... no need to label Tesla fans as being cocky autopilot snobs. The video is childish in my opinion.

 

[hockeythug]

Volvo believes that anything under Level 4 is dangerous and a waste of time. Apples and oranges comparison though for what they want to launch next year though. We will see what version 2 of the autopilot hardware is capable of.

Also, that Volvo "fail" in the video was just some idiot behind the wheel not knowing that car didn't have the extra pedestrian detection package.

 

[vortexz]

part 2

 

[calisnow]

Volvo is vapor ware. When they release a product in the market let's do a comparison. For now there is no comparison because there is no Volvo product. A beta test of 100 cars in London doesn't count.

 

[Mike K]

To be fair, Volvo seems to hate on all level 3 autonomous cars, not just Tesla.

To an extent I see their point. Auto-Pilot as it sits is basically intelligent cruise control with lane keeping. It's nice and it's convenient but view a couple videos on Youtube and you'll quickly find that many Tesla owners are treating it a if they can just read a book and let the car drive itself on the expressway. One video shows an owner rifling through a ticket he just received, showing it to the camera while the car drives in the background. He wasn't in any way ready to take over control of the vehicle if necessary.

So in that respect, I think Volvo's point is that a level 3 system is so close to being good that it gives a false sense of security which could potentially end up being less safe as the driver is coddled into that false sense of security and not ready for a situation in which they'd need to resume control of the vehicle.

To that, I agree. Auto-pilot is great but in it's current iteration I think Tesla could do a better job of advertising what it is and what it isn't.

 

[CHG-ON]

I don't appreciate companies criticising a launched product when they have not launched their own into the real world. For Volvo, of whom I am a fan, their comments are all theory for now.

 

[green1]

No different from the dozens of "tesla killer" cars that will all launch "soon" and take all of tesla's market share... the common denominator is always that the "better" product never actuality exists yet, or if it does you can't actually buy it.

Tesla has had this system on the road, in the hands of normal consumers, for a year and a half. If and when anyone else launches a system to the public, that people can actually buy and use, then we can compare the two. But not until then.

 

[Super_Popular]

Volvo's system will be good when it comes out but they really should watch what they say.

WARNING. The first video is somewhat graphic.

 

[vortexz]

here:

 

[Spidy]

And now try it with a 2013 Model S...

 

[ABCCBA]

Does Level 4 have the capability to read traffic lights?

 

[Saghost]

to be fair, the video can be very biased by cherry picking single incidents. I am a Tesla owner and i LOVE the car, and i both laughed and cringed at the volvo guys royally screwing up something so simple.

I wouldn't want to underestimate the progress that other manufacturers have made just because i "root for" Tesla. Mercedes could have come up with the near perfect algorithm for autopilot and unveil it tomorrow for all i know, and it could blow Tesla's out of the water. What i can observe tho, is how probable this is, and how quick is each manufacturer's autopilot program learning.

Tesla does have a massive, MASSIVE fleet of human driven vehicles on the road that autopilot learns things from, and they don't scale it up by hiring more people to drive the cars around like some sort of UAT, but real world drivers behaving in real world situations, which has massive advantages.

e.g. if you're a programmer, you either decide or get told by a team whether to program a car to do evasive maneuvers for a certain sudden situation. But Tesla can learn from massive real world data, where some would rather brake as hard as possible to take a certain but minimised impact (which, is the ONLY legal way to drive in many countries) whereas others would check for empty space and swerve to avoid the accident. Certain people drive more cautiously and may already have taken their foot off the throttle a bit before something becomes an issue.

What Tesla has, is the massive feed available to them for accurate, real world empirical data instead of some "expert" panel groups who can't possibly consider the whole range of situations that come up. I'm rooting for Tesla, and I think they have an autopilot program that has the best learning potential, regardless of how good they are compared to other car maker's "claims" at the moment.

TL;DR All the non-tesla car makes get the benefit of doubt for their claims of superior autopilot technology, but Tesla has objectively the best autopilot learning platform by far because we're all part of their learning already.

This. Beyond that, I'm sure all the Tesla drivers are contributing to AP learning every day. I think it's a safe bet that all the AP cars make a record every time the driver takes control away from AP and keep all of the available data for that moment in a log file to upload to Tesla when convenient - and that the AP software team go through that data to find what the algorithm isn't catching that humans are and try to roll it into future updates.

I wouldn't be too surprised if they were also running the software in the background when humans are driving, comparing the system planned response with the human response and again logging cases where there's a big difference for upload, but that's more speculative.

Having hundreds of thousands of volunteers flogging the corners of the algorithm for you for years is a huge advantage - receiving in near real time all of the exceptions is another one.

Until someone else adopts Tesla's OTA software approach, it's going to be very hard to take the lead in self driving away from them.

 

[P85DBeast]

Just like ive been saying for the past year since ive owned an S "No other car and take the Model S position this is the only car that does what is does the way it does" not any car does what the S better!