Isaac Sim robot workbench

Undo an experiment without restarting Kit: back to the spawn state in a few frames. Removes spawned props (unless keep_objects), restores the gains and coupling ratios the daemon spawned with, writes every joint and the root back to their spawn values at zero velocity, then settles `settle` steps. ok=False with a hint means use sim_reload: a non-finite state, or a joint that would not return. Spawn properties (USD, solver iterations, self-collision) are not touched; those always need sim_reload.

Capture the viewport as an image. Requires the daemon started with gui=True.

Put a plain panel (RGB 0..1) behind the robot, facing the capture camera; no color removes it. For clean footage, also start the daemon with ground=False.

Point the viewport camera, e.g. eye=[1.2, 1.2, 1.0] target=[0, 0, 0.3].

Place the capture camera by hand (sim_frame_robot does it automatically).

Change software-coupling ratios live, keyed by follower joint. 0 releases a follower. Couplings are declared in the robot config; a follower is driven to ratio x its leader's target.

Change the driven joints' PD gains and effort limit on the live articulation.

Command a named pose. amount blends toward it: 1 = as authored, 0.5 = halfway. Halfway from the lower limits by default, or from the current targets with from_current=True. Follow with sim_step, then sim_screenshot to verify.

Command the driven joints, normalized 0..1 (lower..upper limit). Pass `unit` as a full vector in driven order (see sim_status), or `joints` to set individual joints by name. Targets persist; advance with sim_step.

Add a prop to the live scene: a primitive or a USD file, with collision. shape: cuboid (size=[x,y,z]), sphere (radius), cylinder / capsule / cone (radius, height), or usd (usd=<path>, optional scale). Units are meters and kg. static=True: a fixed collider (a table, a wall). kinematic=True: a rigid body that contact cannot move. rot is a (w, x, y, z) quaternion. color is RGB 0..1. Spawns at /World/Objects/<name>; sim_step to let it fall and settle.

Report whether the sim daemon is up, its robot, driven joints, poses and parameters.

Advance the simulation n physics steps (dt from sim_status, default 1/120 s).

Compare several values of one live parameter inside the single session. param: stiffness, damping, effort, or "coupling:<follower>". For each value it applies the parameter and runs `test` ("wave" or "range") for `steps`. The original value is restored afterwards, even on error. A diverging solver is recorded as a result, not raised.

Start the daemon container and wait until it answers. Idempotent. robot: a robot config JSON, path relative to the repo root (e.g. examples/robots/franka.json); defaults to $ISAAC_MCP_ROBOT. usd: a USD path instead of / overriding the config's (every joint driven if no config). gui=True is needed for sim_screenshot; the host must have run `xhost +local:docker` once. ground=False spawns without a floor (use for range tests). extra_args go to scripts/simd.py verbatim (e.g. ["--pos-iters", "64"]). The first start downloads Nucleus assets and builds shader caches: minutes.

Sweep every driven joint through its range for n steps; return travel stats.

List a run's saved checkpoints with step and size.

List training runs: live or recent containers, and log directories holding checkpoints.

Tail a running training container's stdout/stderr.

TensorBoard scalars for a run: no tag lists tag names; a tag returns its series, downsampled.

Launch a headless Isaac Lab training run in its own container; returns at once. task is a registered gym id (e.g. Isaac-Cartpole-v0). run_name defaults to '<task>_<timestamp>'; keep it to find the run later. extra_args go to the training script verbatim (e.g. Hydra overrides). device pins one GPU index. Independent of the daemon and of this MCP session.

Container state, checkpoints, and latest TensorBoard scalar values for one run.

Stop a training run's container.