Pendulum swing-up
dmcontrol-pendulum-swingupdmcontrolDeepMind Control Suite bodiesManipulationmedium
Instruction
A torque-limited pendulum from the DeepMind Control Suite (MuJoCo): a 1 kg ball on a 0.5 m massless rod, swinging in the x-z plane (z up) about a hinge with light damping. pole_angle is in radians: 0 = pointing straight up, +-pi = hanging straight down; a positive angle tilts the ball towards +x. The motor is too weak to lift the pendulum directly (gravity's torque reaches 4.9 N m, the motor gives at most 1 N m), so it has to be swung up. Simulated time advances only when you act; the scene is paused while you think. A feedback controller can run as a script: Python 3 with numpy is available, and robo observe --json / robo act ... --json print machine-readable output.
Task
Swing the pendulum up and keep it upright. It starts at rest near the bottom (pole_angle = -3.08 rad).
Success: the pendulum within 30 degrees of upright (cosine of pole_angle >= cos 30 deg), the task's own reward condition, held for 100 consecutive steps (2 s of simulated time), judged by the episode server from the simulated state after every step. The episode ends as solved the moment that happens; robo done before that scores 0.
Controls. robo act TAU [--repeat N] applies a torque of TAU N m at the hinge, TAU in [-1, 1] (positive increases pole_angle), for N steps of 20 ms. There are no skills.
Observation. robo observe reports dm_control's own observation, orientation = [cos, sin] of the pole angle and velocity = [angular velocity (rad/s)], plus pole_angle and pole_angvel by name. Every task also reports reward (dm_control's own reward for the last step, for information), in_target (whether the success condition holds right now), hold_steps (for how many consecutive steps it has held) and hold_required, and time_s (simulated time). Cameras: fixed, lookat; robo observe --image [--camera C] saves a picture.
The step budget is 1000 steps (20 s of simulated time).
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 # the robot, its sensors, action groups, skills and step budget
robo observe # robot and scene state as numbers
robo observe --image [--camera C] # also saves a camera image and prints its path (open it to look)
robo act V1 V2 ... [--repeat N] # one low-level action (the action groups under Controls), applied N times (N <= 50)
robo skill NAME ARG ... # run a skill listed by `robo info`; it runs until it finishes and reports the result
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 (+z up); angles are in degrees unless a field says otherwise.
- The episode has a fixed step budget (see
robo info); every simulated control step counts, including the steps a skill runs. - Skills are ordinary controllers: they can fail, stop early or be blocked by the scene. Read what they report and re-observe.
- The episode ends as solved the moment the task's success rule holds (see Success);
robo donebefore that scores 0. - Call
robo doneexactly once when finished.
Run this task
bench eval run \
-d google-deepmind/dm-control@0.1 \
--registry https://robouse.ai/hub/registry.json \
--agent oracle \
--include dmcontrol-pendulum-swingupPinned to robohub commit bcd0d002b9df. The verifier and the reference solution are not published.