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
Calling joint_pos_out_of_limit or joint_pos_out_of_manual_limit with asset_cfg.joint_ids set to anything other than None raises
IndexError: too many indices for tensor of dimension 1
because the helpers reduce across joints (torch.any(dim=1)) before they slice with asset_cfg.joint_ids, leaving a 1‑D tensor that cannot be indexed with [:, ids].
Describe the bug
Calling
joint_pos_out_of_limit
orjoint_pos_out_of_manual_limit
withasset_cfg.joint_ids
set to anything other thanNone
raisesbecause the helpers reduce across joints (
torch.any(dim=1)
) before they slice withasset_cfg.joint_ids
, leaving a 1‑D tensor that cannot be indexed with[:, ids]
.Steps to reproduce
System Info
main
)Additional context
The bug breaks any workflow that relies on per‑joint termination masks
(e.g. training tasks that deactivate specific joints).
Checklist
Acceptance Criteria
joint_pos_out_of_limit
works forNone
,slice
, list and tensor joint selections.joint_pos_out_of_manual_limit
shows the same correct behaviour.The text was updated successfully, but these errors were encountered: