Wipe a line of dirt off the table with a sponge-like tool
robosuite-wipe-pandarobosuiteFranka PandaManipulationhard
Instruction
A Franka Emika Panda (7-DoF) with its parallel-jaw gripper works at a table; its base is at the -x end, facing +x. The scene is robosuite's Wipe environment (MuJoCo).
Goal: Wipe all the dirt (100 small markers along a line on the table) off the table with the wiping tool mounted on the arm. A marker counts as wiped when the tool's bottom face passes over it while pressed flat on the table. robosuite's Wipe check: every marker is wiped. The episode ends by itself when the last marker goes.
The episode ends as solved the moment the check first passes (you do not need to call robo done then).
This robot
- The action has 6 numbers:
robo act DX DY DZ DROLL DPITCH DYAW, each in [-1, 1]: robosuite's OSC_POSE end-effector command, 20 steps per second, in the world frame. DX/DY/DZ = 1.0 asks for a 5 cm move (held at 1.0 the hand travels about 1.1 cm per step); DROLL/DPITCH/DYAW rotate the hand about the world x/y/z axes (1.0 asks for 0.5 rad). robo move-to X Y Zis available (it keeps the tool's orientation);robo gripis not.- World frame, metres: +x points away from the robot's base, +y to the robot's left, +z up.
robo observefields:hand_pos(the tool's reference point, a few millimetres above the middle of its face, metres),hand_quat(x, y, z, w);markers_left,dirt_markers_xy(x, y of every marker still on the table),proportion_wiped,wipe_centroidandwipe_radius(centre and spread of the remaining dirt),hand_force_N(force on the tool, newtons; it rises when the tool presses on the table);table_height.robo observe --imagesaves theagentviewcamera; add--camera robot0_eye_in_handfor the wrist camera. Images are 320x320.- The step budget is 600 steps.
- The action has 6 numbers (there is no gripper). Keep the tool flat (pointing straight down) and pressed lightly onto the table while sweeping; wiping does not count while any other part of the arm touches the table or a joint is at its limit.
move_tois available;gripis not.
How the robot is controlled and scored
You are controlling a simulated robot. Read the task below, then solve it by running the robo command in your shell (start with robo info and robo observe). Keep going until the task is done, then call robo done once. Do not stop to ask questions; there is no human to answer.
How to control the robot
You are the robot's policy. You act only through the robo command in your shell. There is no other way to move the robot, and you cannot read or change the simulator, the scoring, or other files to succeed; the episode server judges the final physical state itself.
robo info # action space, available skills, step budget
robo observe # robot and object state as numbers
robo observe --image # also saves a camera image and prints its path (open it to look)
robo act DX DY DZ GRIP [--repeat N] # low-level action, applied N times (N <= 50)
robo move-to X Y Z [--grip G] # skill: move the gripper toward a point (if enabled for this task)
robo grip G [--steps N] # skill: hold position and set the gripper (+1 close, -1 open)
robo done "short summary" # end the episode and ask for scoring
robo give-up "reason" # end the episode without claiming success- Positions are in metres in the world frame (x, y on the table plane, z up).
- The episode has a fixed step budget (see
robo info); every simulated step counts, including skills. - Unless the task says otherwise, success is judged about 10 steps after you call
robo done, with the robot holding still, so the goal must still be true when the robot stops. - Work in small steps and re-observe after each motion. Call
robo doneexactly once when finished.
Run this task
ROBOUSE_ORACLE_TOKEN=$(openssl rand -hex 16) \
bench eval run \
-d arise-initiative/robosuite@0.2 \
--registry https://robouse.ai/hub/registry.json \
--agent oracle \
--include robosuite-wipe-pandaPinned to robohub commit e472b1a1e041. The verifier and the reference solution are not published.
| Trial | |||||
|---|---|---|---|---|---|
| GPT-6 Astra · Codex | Solved | 140 / 600 | 78 s | Trial Compare | |
| Kimi K3 · Claude Code | Solved | 242 / 600 | 9 min | Trial Compare | |
| Kimi K3 · mini-swe-agent | Solved | 319 / 600 | 16 min | Trial Compare | |
| GLM-5.3 · Claude Code | Solved | 261 / 600 | 9 min | Trial Compare | |
| GLM-5.3 · mini-swe-agent | Not solved | 600 / 600 | 21 min | Trial Compare | |
| GLM-5.3 · Codex | Infra failure | 54 / 600 | 51 s | Trial Compare |
No model matches.Clear the filter or change the models in the comparison set.
These trials ran on version 0.1, now deprecated (superseded by 0.2).