A heterogeneous SE(3)-equivariant graph policy (HEPi) outperforms Transformer and homogeneous equivariant baselines on a new seven-task manipulation reinforcement learning benchmark.
This shows that any object node can exchange information with the actuator nodes after a single layer of the object-actuator update
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Geometry-aware RL for Manipulation of Varying Shapes and Deformable Objects
A heterogeneous SE(3)-equivariant graph policy (HEPi) outperforms Transformer and homogeneous equivariant baselines on a new seven-task manipulation reinforcement learning benchmark.