Cover With Lid¶
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 05m | 09-26 05:00 | 1h 05m | 155 | 19.0M / 115k | $0.276 | deterministic 1, grader_error 0, instance_exact 1, instance_ok 1, n_actions 647, video_rendered 1 |
| L✗ 1h 30m | 09-29 22:42 | 1h 30m | 218 | 24.0M / 201k | $0.403 | deterministic 1, grader_error 0, live_success 0, n_actions 2459, replay_success 0, video_rendered 1 |
Task instruction (upstream)
Cover the pot precisely using one lid.
Playback speed
What RoboWits states about this task
| Success criteria | 1. Exactly one lid is above the pot 2. Lid overlaps pot by at least 80% 3. Lid size is appropriate (85%-140% of pot area) |
| Objects | lid large lid medium pot without lid small lid |
| Physics | rigid |
| Episode budget | 200 control steps |
| Default control mode | EE_ABS |
| Evaluation scenes | 50 |
| Published mutations | 5 |
| BlenderKit assets | 2 (not redistributable — mounted from the host) |
| Registry id | robowits/13-cover-with-lid-v0 |
| Environment class | CoverWithLidEnv |
| Defined in | gs_gym/envs/robowits/13_cover_with_lid.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.