A graph-based diffusion model trained only on procedurally generated pseudo-demonstrations lets a robot perform new manipulation tasks from one or two test-time demonstrations.
To add noise to the action expressed as (TEA ∈ SE(3), ag ∈ R, we first project TEA to se(3) using a Logmap, normalise the resulting vectors, add the noise as described by Ho et al
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Instant Policy: In-Context Imitation Learning via Graph Diffusion
A graph-based diffusion model trained only on procedurally generated pseudo-demonstrations lets a robot perform new manipulation tasks from one or two test-time demonstrations.