- Rating
- 2
- Confidence
- 2
- Informativeness
- 1
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper addresses cross-embodiment imitation learning. The authors argue that treating a demonstration as a sequence of actions to reproduce is the wrong formulation due to the difference in source and target robots. They propose Demo-JEPA, which translates source demonstrations into trajectories in a shared JEPA representation space, and then optimizes actions toward them in target robot's action space. Experiments on RLBench and real-world manipulation tasks are reported across three suites of increasing distribution shift.
Critical strengths
-
It is a good reframing of the problem. A demonstration is not an optimal trajectory to reproduce considering the difference in source robot and target robot. It is a sound choice to operate in a JEPA latent space rather than pixel space.
-
The authors recognize that source-target temporal misalignment is the main vulnerability, and propose engineering solutions to mitigate this issue.
-
Experiments cover both simulation and the real world
Critical weaknesses
-
It is unclear what the major contributions are. The world model is adopted from V-JEPA 2.1. The only newly learned component is the dreamer predictor. Once the latent goal is given, the rest is standard goal-conditioned latent MPC on the target.
-
Scene-configuration mismatch is motivated but never evaluated. Although target-side initial conditions are varied, but source and target are never systematically placed in different configurations.
-
Transfer is evaluated toward one target (Franka arm), and all three arms are similar 6 or 7 DoF manipulators, so the morphological gap tested is modest.
-
Baseline coverage is thin: no VLA baseline and no target-only behavior-cloning baseline. So I cannot see what the source demonstration contributes over training on target data alone.
Questions for authors
-
What does conditioning on the source demonstration contribute relative to target data alone?
-
Since the simulation source demonstrations are generated by retargeted replay of the target's trajectories, does trajectory retargeting dominate all three simulation suites by construction rather than by privileged information?