Kevin Black
Kevin Black is a researcher affiliated with Physical Intelligence, a San Francisco-based company developing foundation models and learning algorithms for robotic control. He is known for his work on training-time real-time control (RTC) methods, which simulate latency during post-training to teach models to generate future actions conditioned on already committed actions. His research contributes to the development of vision-language-action systems and robotic manipulation algorithms.
“We use GPT-6-Astra as System 2 and π0.5 [12] fine-tuned on RoboDojo as System 1 (VLA)”
Source→“Kevin Black... Co-author of the π0.5 paper [Ref 12]”
Source→“Kevin Black - Developer of π0 and π0.5, the primary low-level policies evaluated”
Source→“training-time RTC simulates this latency during post-training and teaches the model to generate future actions conditioned on already committed actions (Section 3.3, citing Black et al. [5])”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.