You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
However, in the original paper, different image encoders are used for different cameras.
@alexander-soare Is this design choice? (e.g. single encoder is better than multiple encoders?)
Chi, Cheng, et al. "Diffusion policy: Visuomotor policy learning via action diffusion." The International Journal of Robotics Research (2023): 02783649241273668.
The text was updated successfully, but these errors were encountered:
@HiroIshida thanks for raising this. It was a choice for simple over complicated, mostly for speed of development. We're not against using multiple image encoders going forward.
In the diffusion policy implementation, it seems that RGB encoder is shared by all camera images.
lerobot/lerobot/common/policies/diffusion/modeling_diffusion.py
Line 185 in c351e1f
However, in the original paper, different image encoders are used for different cameras.
@alexander-soare Is this design choice? (e.g. single encoder is better than multiple encoders?)
Chi, Cheng, et al. "Diffusion policy: Visuomotor policy learning via action diffusion." The International Journal of Robotics Research (2023): 02783649241273668.
The text was updated successfully, but these errors were encountered: