Retrieve Roll¶
Agent runs
Each mode of this task runs once per run. Est. cost is tokens at list price (data/prices.yml), never a bill; see the RoboWits runs for every task of a run.
Codex + GPT-6 Luna, reasoning xhigh (ChatGPT login)default run
| Trial | Started | Agent time | Model requests | Tokens in / out | Est. cost | Grade |
|---|---|---|---|---|---|---|
| U✗ 1h 30m | 09-26 07:36 | 1h 30m | 131 | 10.7M / 107k | $0.193 | grader_error 0, missing_trajectory 1 |
| L✗ 1h 30m | 09-29 18:51 | 1h 30m | 127 | 12.3M / 114k | $0.202 | deterministic 1, grader_error 0, live_success 0, n_actions 1240, replay_success 0, video_rendered 1 |
Task instruction (upstream)
Use the long rod to retrieve the out-of-reach roll and move it to the green target area.
Playback speed
What RoboWits states about this task
| Success criteria | 1. Roll overlaps target area by at least 50% 2. Roll remains within table bounds |
| Objects | green target area hollow roll long rod |
| Physics | rigid |
| Episode budget | 200 control steps |
| Default control mode | EE_ABS |
| Evaluation scenes | 50 |
| Published mutations | 7 |
| BlenderKit assets | 1 (not redistributable — mounted from the host) |
| Registry id | robowits/27-retrieve-roll-v0 |
| Environment class | RetrieveRollEnv |
| Defined in | gs_gym/envs/robowits/27_retrieve_roll.py |
From https://github.com/UMass-Embodied-AGI/RoboWits @ 9cc30ae.
Tags¶
Task DomainManipulation
Why this task is interesting¶
Not yet written.
Capability notes¶
Not yet written.
Oracle demo review¶
Not yet reviewed.