These days I read a lot about Artificial Intelligence. I think this is the most crucial topic for the next coming years. In the past few days, I read about Tesla's new supercomputer, which they use for AI purposes. I was curious about Andrej Karpathy's presentation, how it was structured!
If you find interesting the presentation, you can find it on the following link:
Surprisingly the supercomputer part was not interesting! They built a conservative supercomputer, eliminating the next-gen architectures which rose in the past years.
Understanding how self-driving cars learn about the environment gave me another idea that could enhance these systems' safety. If we notice, objects like pedestrians or bicyclists near the road could be uploaded to a predictive system. The predictive system can predict the movements of these objects. Even if the object is not visible to the camera, it could prepare for possible situations. I assume later these systems will arise! Active communication could solve a lot of issues which not possible to simulate. I think those counties that implement these features for smart roads will have the safeties places!
I would be curious why these models are not used currently and why manufacturers like Tesla wanted to simulate situations without collective intelligence.
These collaborations between cars and even with the environment around the road will help a lot in the future.
Workshop on Autonomous Driving at CVPR'21 — https://www.linkedin.com/embeds/publishingEmbed.html?articleId=8150726550112760791]