I wonder how much of this is actual progress and how much is marketing fluff.
Reinforcement learning in perfect simulations isn’t the same as messy real-world scenarios.
The data rate of 5 GB/s is nuts. That’s a lot of sensory input to process in real time.
Curious how Tesla handles memory retention without catastrophic forgetting.
If they get policy-sharing between robots right, training speed will be unprecedented.
I hope there’s transparency on how Tesla ensures ethical training of these bots.
Love the Apple comparison—Tesla really is becoming the “Apple of robotics.”
Still waiting for FSD to be “solved” first before trusting a humanoid robot from Tesla.
Who’s auditing the ethics of how these robots are being trained?
If it’s so advanced, why haven’t we seen more practical demonstrations yet?