){kind=logo}
){kind=logo}
Overcoming Data Challenges for Physical AI
This panel will address the most limiting factor facing humanoid deployment today – we do not have enough, good data. The discussion will explore what it means for humanoid training data to be “good”, where this data is collected from both in simulation and from reality, and the tradeoffs when deciding on a data collection method. We will also explore the potential for leveraging current, suboptimal data for continuous humanoid learning, and the need to utilize training data from multiple embodiments as robots move towards their optimal form factor.
Speakers
