technomation
Member
Interesting comparison you give, a laptop. People used to regularly upgrade their PCs and laptops just for a little more performance and to gain some new functionality and graphics capability. Now processor performance is so high with no exciting gains in functionality that few are compelled to upgrade, regardless of the insignificant cost. Going from AP 1.0 to 2.0 is a different story because much new functionality is to be gained, e.g., recognizing stop signs and traffic lights, planning miles-ahead curve slowdowns and speedups, properly reacting to more complex scenarios with motorcycles and pedestrians, and other safety features sure to come in the near future...but NOT if your new car is on the old hardware.
With more sensors and cameras on board, the amount of data to be analyzed will escalate, an actual big-data analytics problem, real-time in your car. A new, more powerful processor will also be required. Elon has said That Tesla is doing only “narrow AI,” but that doesn’t mean that over their software-development roadmap that they want to be confined to simple logistical-regression AI algorithms. To advance to L3/L4 and beyond, Tesla's dev team may want to add some bells and whistles to their current AI algorithms, e.g., stochastic gradient descent or maybe some new Deep Learning algorithms. For this more advanced AI and all the big data from more cameras and sensors that has to be processed in real-time, they’re going to need to switch from the simpler EyeQ3 to NVIDIA Tegra or some other more capable processor. There’s just a whole lot more going on in the autonomic driver-assist world than with boring laptops. That's why I'm concerned about missing the AP 2.0 boat with my Model S P100D order that I placed just two weeks ago (the new hardware probably costs less than 1 percent of my total vehicle cost). Tesla has been lauded for its software upgradeability. They ought to consider upping their game with hardware retrofitability.
With more sensors and cameras on board, the amount of data to be analyzed will escalate, an actual big-data analytics problem, real-time in your car. A new, more powerful processor will also be required. Elon has said That Tesla is doing only “narrow AI,” but that doesn’t mean that over their software-development roadmap that they want to be confined to simple logistical-regression AI algorithms. To advance to L3/L4 and beyond, Tesla's dev team may want to add some bells and whistles to their current AI algorithms, e.g., stochastic gradient descent or maybe some new Deep Learning algorithms. For this more advanced AI and all the big data from more cameras and sensors that has to be processed in real-time, they’re going to need to switch from the simpler EyeQ3 to NVIDIA Tegra or some other more capable processor. There’s just a whole lot more going on in the autonomic driver-assist world than with boring laptops. That's why I'm concerned about missing the AP 2.0 boat with my Model S P100D order that I placed just two weeks ago (the new hardware probably costs less than 1 percent of my total vehicle cost). Tesla has been lauded for its software upgradeability. They ought to consider upping their game with hardware retrofitability.