Balance Board¶
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 20% | 09-26 08:03 | 1h 30m | 104 | 13.7M / 153k | $0.240 | deterministic 1, grader_error 0, instance_exact 1, instance_ok 1, n_actions 1145, video_rendered 1 |
| L✗ 1h 30m | 09-29 16:08 | 1h 30m | 118 | 13.6M / 105k | $0.212 | deterministic 1, grader_error 0, live_success 0, n_actions 834, replay_success 0, video_rendered 1 |
Task instruction (upstream)
Place the board onto the wall and keep it balanced.
Playback speed
What RoboWits states about this task
| Success criteria | 1. Board is resting on top of the wall near its center 2. Board is balanced (nearly horizontal, not tipping/moving) 3. Objects remain within table bounds |
| Objects | board small wall support cube |
| Physics | rigid |
| Episode budget | 200 control steps |
| Default control mode | EE_ABS |
| Evaluation scenes | 50 |
| Published mutations | 6 |
| BlenderKit assets | 1 (not redistributable — mounted from the host) |
| Registry id | robowits/29-balance-board-v0 |
| Environment class | BalanceBoardEnv |
| Defined in | gs_gym/envs/robowits/29_balance_board.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.