Interesting stuff. I wonder to what extent Google and Tesla could benefit from sharing each others datasets, Tesla has far more real world data than Google at this point in time but Google has the better virtual environment to test in.
Inside Waymo's Secret World for Training Self-Driving Cars
11–20 of 146 posts
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#12Interesting stuff. I wonder to what extent Google and Tesla could benefit from sharing each others datasets, Tesla has far more real world data than Google at this point in time but Google has the better virtual environment to test in.
Not all data is equal. Waymo test vehicled likely record a bunch of data from all the sensors and can download them all at the end of the data.
Tesla just enabled in the last few months the option for people to upload data to them. Determining which data to upload is its own problem as well.
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#13Clearly the progress being made in the area of self-driving cars is undeniable. Seeing the articles and discussions crop up on a daily basis are making me wonder if we are headed for an AI Winter scenario in this area within the next decade or if this is the real deal and we will see self-driving cars at dealerships within 20 years. This is so far from being my area of expertise. Just one observer's thoughts/question…
I don't disagree that AI is over-hyped. However, I think the research environment is not as likely to dry up this time. That is because much of the machine learning research is coming from industry, particularly the mega-corps. And for extremely important disruptive events such as AI, they are willing to drop $$ into collecting a stable of research scientists to not fall behind in this space. So while the AI winter w…
Agreed. As a Google engineer, I think a lot of our AI efforts are "safe" (as in we'll keep investing in them for a long time) because they're already providing substantial business value. For example,
- TTS and speech synthesis have lots of uses in Android (phones, Wear, Auto, etc.)
- Object recognition is very useful for photo search
- Face recognition is also useful for photo search (if you tag people in Google Photos)
- Neural machine translation provides better translation accuracy (for languages which we have enough training data for)
It's possible Google will cut back on some more speculative ML research efforts, but certainly not those four, I would think.
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#14Interesting stuff. I wonder to what extent Google and Tesla could benefit from sharing each others datasets, Tesla has far more real world data than Google at this point in time but Google has the better virtual environment to test in.
Not all data is equal. Waymo test vehicled likely record a bunch of data from all the sensors and can download them all at the end of the data. Tesla just enabled in the last few months the option for people to upload data to them. Determining which data to upload is its own problem as well.
> Tesla just enabled in the last few months the option for people to upload data to them.
Did you mean Tesla never captured data before that? If yes then I disagree. Sorry.
More here - https://electrek.co/2017/07/12/tesla-global-fleet-electric-m...
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#15Earlier quoted context omitted.
I don't disagree that AI is over-hyped. However, I think the research environment is not as likely to dry up this time. That is because much of the machine learning research is coming from industry, particularly the mega-corps. And for extremely important disruptive events such as AI, they are willing to drop $$ into collecting a stable of research scientists to not fall behind in this space. So while the AI winter w…
Agreed. As a Google engineer, I think a lot of our AI efforts are "safe" (as in we'll keep investing in them for a long time) because they're already providing substantial business value. For example, - TTS and speech synthesis have lots of uses in Android (phones, Wear, Auto, etc.) - Object recognition is very useful for photo search - Face recognition is also useful for photo search (if you tag people in Google Pho…
As a fellow google engineer, I'd be curious to know which (public) research you think isn't marketable/valuable.
As you've mentioned, object detection (which is a considerable portion of CV) is marketable. Meta learning (Vizier, "Learning to learn with gradient descent with gradient descent") are incredibly valuable from a business perspective. Model compression is important for neural networks on mobile/embedded devices, etc.
One of the very interesting things about ML being such a corporation driven field is that it's incredible how quickly relatively recent research makes its way into products.
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#16> for an average of once every 890 miles, or 0.80 disengagements per 1,000 miles.
Isn’t that 1.12~ per 1000mi?
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#17Interesting stuff. I wonder to what extent Google and Tesla could benefit from sharing each others datasets, Tesla has far more real world data than Google at this point in time but Google has the better virtual environment to test in.
Not all data is equal. Waymo test vehicled likely record a bunch of data from all the sensors and can download them all at the end of the data. Tesla just enabled in the last few months the option for people to upload data to them. Determining which data to upload is its own problem as well.
https://qz.com/694520/tesla-has-780-million-miles-of-driving...
Tesla has an enormous amount of data at their disposal. The recent change concerned video footage from the onboard cameras.
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#18Earlier quoted context omitted.
Not all data is equal. Waymo test vehicled likely record a bunch of data from all the sensors and can download them all at the end of the data. Tesla just enabled in the last few months the option for people to upload data to them. Determining which data to upload is its own problem as well.
https://qz.com/694520/tesla-has-780-million-miles-of-driving... Tesla has an enormous amount of data at their disposal. The recent change concerned video footage from the onboard cameras.
And not much lidar data, right? I don't work on this, but I imagine the quality of the data set makes a big difference.
From what I've learned, lidar and radar are typically used for object detection (i.e., avoiding other cars; perhaps augmented with camera data), while cameras are used for things like lane detection and traffic sign detection. If Tesla is trying to solve the same hard problems as everyone else, using weaker equipment, I wouldn't bet on them having a lot of success.
Also, kind of the point of this article is that Waymo has a test area where they can get real data along with the ground truth. That seems much more valuable than unlabeled/random data.
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#19The fact that Waymo revealed their "secret" tools for advancing this crucial technology implies that either:
1) They believe no one can quite catch up before they can launch the technology. Since they know several competitors have huge resources and brilliant people, it means they are quite close to launch.
2) These tools have become open secret within the industry, so no harm is done to their competitive position by revealing it to the public, only good PR to be gained perhaps to attract more bright engineers.
I suspect 2) is more likely since several key players have moved around so much in the past couple of years. Relatively high-level knowledge about how autonomous vehicles are being developed at Waymo might have become well-known within the industry by now.
Re: Inside Waymo's Secret World for Training Self-Driving Cars
#20That's pretty fucking cool.