Cloudification of Transportation
There are too many repetitive tasks happening across all companies, so it's necessary to share and divide the work among them
Tesla uses expensive sensors during training time and distilling that into vision only model. Scaling hardware problems is much harder than software problems. Large amount of human data and expensive sensory data is done for each pre-training and fine-tuning.
For fully autonomous driving to be commercially viable, it must demonstrate significantly higher safety levels than human driving. To achieve this, LiDAR sensors are essential. The argument that cameras alone are sufficient because humans drive with vision only may be flawed without demonstrating overwhelmingly superior safety performance.
Autonomous Driving Intensions
Autonomous Driving Extensions

Seonglae Cho
