Tesla has its own supercomputer, and with its 1.8 eflops and ten petabytes of NVME memory, the new computer is expected to completely revolutionize the company’s autopilot.
Tesla’s AI boss even goes so far as to rank the new computer fifth in the top 500 supercomputers and proudly proclaims that it far outstrips Nvidia and others. But perhaps some readers are wondering – why does Tesla need a supercomputer? The company wants to realize autonomous driving according to the Vision Only approach in the near future and needs a huge amount of computing power for this.
With the new computer, Tesla can replace the previous radar and lidar sensors, which were necessary for the car’s navigation, with modern cameras to react even faster. The model here is humans, who can also react quickly and usually correctly on sight. The total of eight cameras that will be used in new Teslas will collect 36 images per second and process this data at incredible speed to ensure that not a single detail is overlooked.
The advantage of this vision-only approach is that it enables high-resolution images that were inconceivable with previous methods, but it also requires a supercomputer to make the vision a reality. What do you think of this approach to autonomous driving? Feel free to post it in the comments.