
Ep#13 Instant Policy: In-Context Imitation Learning via Graph Diffusion
RoboPapers
00:00
Exploring Graph Structures in Robotic State Representation
This chapter explores the use of graph structures to represent states and interactions within robotic systems. The speakers highlight the benefits of this approach, including flexibility in state sequences and the capacity for graph expansion during inference, drawing from past project experiences.
Transcript
Play full episode