Tesla's Supercomputer & Self-Driving Cars
Project Dojo
World's 5th Most Powerful Supercomputer
Many automotive experts consider Tesla to be the global leader in autonomous driving. Tesla is now using the world's 5th most powerful supercomputer to help train its vehicles to drive autonomously. The AI supercomputer system is called Dojo. Right now, humans and computers at Tesla are training its self driving neural network and Dojo is crunching the driving data that Tesla is getting from its network of customers on the road. Dojo is automating the process. Tesla plans to replace the radar, LIDAR sensors and camera combination it has used to navigate its self-driving vehicles with a vision-only system using cameras and Dojo hosting a neural network. CEO Elon Musk says vision is more precise and far faster than radar and LIDAR.
Tech Specs
Dojo is impressive. It has 5,760 GPUs delivering 1.8 EFLOPS, 10 Petabytes of "hot tier" NVME storage and 640 terabits per second of switching capability. Each Tesla car will be equipped with 8 cameras collecting footage that is fed to the supercomputer, where it gets processed real-time. Tesla says the system is much faster and more reactive than a driver, provides 360 degree awareness and cannot be distracted. Some examples of functionality offered by Tesla: the system provides emergency braking to prevent a pedestrian from being hit and warns the driver about traffic lights.
Do Drivers Trust Autonomous Vehicles (AVs)?
According to an April 2021 survey of 1,247 consumers, 51% responded that they are not comfortable using an AV because of safety concerns. A majority also have little faith in the ability of automakers to develop safe, AV technology. However, in the survey, Tesla ranked first as the automaker they would have the most trust in to develop AVs. Dojo, the world's 5th fastest and most powerful supercomputer may help to build that trust factor. For journalist Edward Kane's latest book on future travel vehicles, go to https://read.amazon.com/kp/embed?asin=B095XDNZVD&preview=newtab&linkCode=kpe&ref_=cm_sw_r_kb_dp_VAMZVDMC1XDWG679G2NC
Comments
Post a Comment