datagen-stack-unsafe25-joint-5cam

stack_retrieve demonstrations (unstack three objects from a target object onto a pile beside it, then retrieve the target into a goal region) with per-episode safety labels.

Purpose: built for a standalone, temporary experiment; not part of the main data release. For general SFT use datagen-stack-v1-joint-5cam.

Contents

Episodes	1,120 (28 base tasks × 40)
Unsafe episodes	280
Frames	2,656,420
Unique language prompts	8
Robot / control	FrankaPanda / absolute joint
FPS / resolution / format	30 / 256×256 / LeRobot v2.1

Composition

1120 episodes, 40 per base task × 28 tasks, of which 280 (25.00%) are unsafe successes. The unsafe count per task ranges 0–26: the family total of 25% is split across tasks in proportion to each task's natural unsafe rate. The remaining episodes are safe successes (840 newly collected). Violated clauses among the unsafe episodes: target_upright 242, stack_upright 222, no_stack_dropped 4 (an episode can violate several). Prompts are the plain task instruction; safety labels appear only in episode_labels.jsonl.

Sibling repos built from the same pool: datagen-stack-allsafe40-joint-5cam · datagen-stack-unsafe25-joint-5cam · datagen-stack-unsafe25-safetytag-joint-5cam · raw pool datagen-stack-safetylabeled-raw.

Collection

Demonstrations come from the same cuRobo planner pipeline, bench tasks and cameras as datagen-stack-v1-joint-5cam, with one difference: the LTL safety monitor ran on every step but did not reject unsafe attempts. Every successful attempt was kept and labeled safe or unsafe by the monitor (fixed draw stream starting at draw index 1000). An episode is unsafe if it violates at least one of the family's safety clauses:

Clause	LTLf	Meaning
no_stack_dropped	G (!any_stack_dropped)	no stacked object falls to the floor
stack_upright	G (all_stack_upright)	all stacked objects stay upright
target_not_dropped	G (!target_dropped)	the target never falls to the floor
target_upright	G (target_upright)	the target stays upright

Schema

Identical to the v1 dataset (LeRobot v2.1, 30 fps, 256×256 h264). state(8)=[arm_q(7),gripper(1)]; actions(8)=[arm_q[t+1](7),gripper_cmd(1)] (default SFT target); actions_commanded(8)=cuRobo command. Cameras: image_opposite, image_left, image_right, image_left_shoulder, wrist_image.

episode_labels.jsonl (one row per episode): episode_index, task (base task), source (new, or the v1 repo the episode was copied from), source_ref (task_XXXX/traj_YYY in the raw pool repo, or the v1 episode_XXXXXX), draw_index, safe, violated_constraints, violation_frame (first violating frame in this episode, null if safe), length.

License

Released under CC BY-NC 4.0 (see LICENSE). The generated artifacts — joint state/action trajectories, rendered demonstration videos, and language prompts — are © 2026 IDEAS Lab, Northwestern University.

The underlying scene and object assets shown in the rendered videos come from BEHAVIOR-1K and remain subject to its license; this dataset does not redistribute them as assets.