Pirelli P Zero PZ5 Was Born from Artificial Intelligence

[borski]

https://www.caranddriver.com/news/a64796129/pirelli-p-zero-pz5-tire-details/

 

[Rogue]

Something legitimate for once of the use of "AI". This appears to be stimulation studies where the computer has been given a set of parameters to tune that normally a person would tune. Great things. Legitimate use cases.

Hope to see the new tire on the next set, and may it be sturdy enough to handle the potholes without bubbles.

 

[hmp10]

https://www.caranddriver.com/news/a64796129/pirelli-p-zero-pz5-tire-details/

Maybe this is one of the reasons that Pirelli was able to homologate the PZ5 so quickly after introduction for several cars, including the Gravity?

Jonathan Benson of "Tyre Reviews" has already done specific measurements of the tire's performance and pronounced it the new state of the art for UHP road tires.

 

[idiot900]

Hard to tell from the article how this was "AI" as opposed to just manually varying parameters in their FEA simulation. However recently in the sciences there have been more applications of machine learning to make better guesses about choosing model parameters from very large parameter spaces for simulating physical processes, so maybe this is something like that.

 

[Rogue]

Hard to tell from the article how this was "AI" as opposed to just manually varying parameters in their FEA simulation. However recently in the sciences there have been more applications of machine learning to make better guesses about choosing model parameters from very large parameter spaces for simulating physical processes, so maybe this is something like that.

Hence the quotes in my post around "AI". Essentially, anything that a computer does in a semi-automatic or automatic manner nowadays is labeled as AI.

I work in the physical simulation spaces. There is not an AI (eg. Artificial Intelligence) aka "intelligence" of the machine in this space or any other.

As noted above, Pirelli is almost certainly just automating the process of adjusting parameter inputs related to the physical model (material, physical design, etc.) within a known pre-defined space (ie. not producing novelty in the sense of processes, defining of parameters, etc). To be fair, most companies struggle to even do this basic step, and the resulting outputs in productivity are astounding when done correctly.

 

[illopp00]

Hence the quotes in my post around "AI". Essentially, anything that a computer does in a semi-automatic or automatic manner nowadays is labeled as AI.

I work in the physical simulation spaces. There is not an AI (eg. Artificial Intelligence) aka "intelligence" of the machine in this space or any other.

As noted above, Pirelli is almost certainly just automating the process of adjusting parameter inputs related to the physical model (material, physical design, etc.) within a known pre-defined space (ie. not producing novelty in the sense of processes, defining of parameters, etc). To be fair, most companies struggle to even do this basic step, and the resulting outputs in productivity are astounding when done correctly.

Yes; although that article only used AI once, which is very restrained compared to many AI buzzword soups. One of the areas I have a more promised and keen interest in is like NVDIA's Omniverse, where they can use multiparameters and other forms of machine learning to do more simultaneous and complex simulations and inferences faster over a wider range of conditions. A lot of physics-based modeling is typically statistically based from curve fitting over test data with some 'fudge' factor coefficients. I went deep into a rabbit hole in my earlier years of trying to reverse engineer tire physics models and they really cannot capture all the range of rubber flexing motions and temperature and compound dependencies.