Stand Bulb¶
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 00m | 09-26 05:29 | 1h 00m | 141 | 16.1M / 121k | $0.264 | deterministic 1, grader_error 0, instance_exact 1, instance_ok 1, n_actions 1225, video_rendered 1 |
| L✗ 1h 30m | 09-29 20:24 | 1h 30m | 328 | 36.3M / 218k | $0.541 | deterministic 1, grader_error 0, live_success 0, n_actions 2372, replay_success 0, video_rendered 1 |
Task instruction (upstream)
Stand the tall object with curved bottom stably within the green target area.
Playback speed
What RoboWits states about this task
| Success criteria | 1. Tall object overlaps target patch by at least 80% 2. Tall object is upright (Z extent dominates XY extents) 3. Tall object is stable (low velocity) 4. Objects remain within table bounds 5. Tall object is not being held |
| Objects | stabilizing ring tall object with curved bottom target patch |
| Physics | rigid |
| Episode budget | 200 control steps |
| Default control mode | EE_ABS |
| Evaluation scenes | 50 |
| Published mutations | 7 |
| BlenderKit assets | 2 (not redistributable — mounted from the host) |
| Registry id | robowits/16-stand-bulb-v0 |
| Environment class | StandBulbEnv |
| Defined in | gs_gym/envs/robowits/16_stand_bulb.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.