Single-stage · PnPSinkToCounter¶
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 RoboCasa runs for every task of a run.
Codex + GPT-6 Luna, reasoning xhigh (ChatGPT login)default runclosed
Not in this run: not in the sample (robot_coding_bench PR
Task instruction (upstream)
Pick the tomato from the sink and place it on the plate located on the counter.
Playback speed
Recorded by us: our reference solution, replayed by the verifier (the scene camera).
What RoboCasa states about this task
| Success criteria | 1. The object touches the plate and its centre is within 0.7 × the plate's horizontal radius of the plate's centre (horizontal distance). 2. The plate touches the counter. 3. The gripper's grip site (between the fingertips) is more than 25 cm from the object's centre: let go and move the hand away. A "centre" is a body's origin; "touches" means a MuJoCo contact. 4. unlimited (privileged): both fresh-process replays of the handed-in trajectory end in the same state, and the check holds on it 5. limited (standard): the one episode (no reset) is recorded by the service and replays to the same state; the check holds on it live and in the replay |
| Family | robocasa/PnPSinkToCounter |
| Robot | Franka Panda on an Omron mobile base with a torso lift (PandaOmron) |
| Category | Single-stage · kitchen pnp |
| Instance | the initial state of official demonstration episode 0 (demo_1, v0.1/single_stage/kitchen_pnp/PnPSinkToCounter/2024-04-26_2); MuJoCo state, model arrays, task state and RNG frozen in instance.npz (SHA-256 337584e93136…, checked on load) |
| Deliverable | (T, 12) native controller commands, 1 ≤ T ≤ 3000, 20 Hz |
| Reference Solution | The demonstration's own actions (original: 273 actions), executed from the frozen scene by the robot's own controller: 273 actions in solution/oracle.npz. The demonstration is a reference solution: it is not in the image and the agent cannot reach it. |
| Limited Mode | Standard mode (robot as a service, eai-standard/2.1) of robocasa-pnp-sink-to-counter-i00-privileged: the same frozen instance and success check, served by the sim sidecar (images/robocasa/standard/server_casa.py, tool module casa_tool), which records the episode, replays it in a fresh simulator and judges it; the verifier grades the sidecar's record in a container of its own. The agent sees the robot's cameras (RGB-D, calibrated), its own joints and end effector, and the instruction. |
| Oracle | full |
| Base Image | ghcr.io/mll-lab-nu/eai-robocasa:0.1.0 |
| Agent Budget | 3600 s of wall clock per mode |
| Environment Source | https://github.com/robocasa/robocasa/blob/756598a5be52e052339bb2d957426e39015c2afb/robocasa/environments/kitchen/single_stage/kitchen_pnp.py#L369 |
| Task Dirs | robocasa-pnp-sink-to-counter-i00-privileged, robocasa-pnp-sink-to-counter-i00-standard |
From https://github.com/robocasa/robocasa @ 756598a (v0.2), as defined in our task definitions @ 8c5594a43.
Tags¶
CapabilityControl & Coordination
Task DomainManipulation
Why this task is interesting¶
Not yet written.
Capability notes¶
Not yet written.
Oracle demo review¶
Not yet reviewed.