> ## Documentation Index > Fetch the complete documentation index at: https://dimensional-adventure-x.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # The Learning Loop > Collect demonstrations, build a dataset, train a policy with LeRobot, and run it back on the robot. This is the full loop, on one page: ``` demonstrate ──▶ session.db ──▶ dataset ──▶ train ──▶ run on the robot (VR / teach) (recorded) (LeRobot (ACT) (policy module) or HDF5) ``` Each arrow is one command. By the end you will have taught a robot a task by showing it, not programming it. ## 1. Demonstrate Two ways to produce demonstrations. Both record the same thing — camera frames, joint states, and episode markers — into a timestamped session database. ### With a VR headset Best when you have a Quest 3. You see what you are doing, and the arm tracks your hand. ```bash theme={null} dimos --simulation run learning-collect-quest-xarm7 # MuJoCo, no hardware dimos run learning-collect-quest-piper # real Piper + RealSense ``` Open `https://:8443/teleop` in the Quest browser, accept the certificate, tap Connect. Then: | Button | Action | | ------------ | ------------------------------------------------------- | | **A** (hold) | Engage — the arm tracks your controller only while held | | **B** | Start recording / save the episode | | **Y** | Discard the episode in progress | A take is: hold **A**, move into place, press **B**, do the task, press **B** again. The terminal confirms every save: ``` [collect] ▶ RECORDING episode (state=recording saved=0 discarded=0) [collect] ✓ SAVED episode (state=idle saved=1 discarded=0) ``` Save each good take with **B** before quitting — an episode still recording at shutdown is dropped. ### By hand-teaching Best for arms you can move by hand. The Galaxea A1Z runs gravity compensation while you guide it through the task: ```bash theme={null} uv run dimos a1z teach --camera-index 0 --task "pick up the object" ``` `SPACE` starts an episode and saves it, `g` toggles the gripper, `d` discards — and pressing `d` between episodes undoes the last save. Real hardware only for now; there is no A1Z simulator yet. Replay any episode to check what you captured: ```bash theme={null} uv run dimos a1z replay ~/.local/state/dimos/recordings/a1z_teach_.db ``` ### What you end up with Either way, the recorder prints the path of a session database: ``` ~/.local/state/dimos/recordings/session_xarm7_20260622_120000.db ``` One file per run, never overwritten. It holds three streams: `color_image`, `coordinator_joint_state`, and the episode start/save/discard markers. Record enough successful episodes to cover the variation the policy will encounter, including camera pose, lighting, object pose, and motion timing. ## 2. Build a dataset DataPrep reads the session database, aligns camera and joint streams onto one clock, splits at the episode markers, and writes a training dataset: ```bash theme={null} dimos dataprep build \ --source ~/.local/state/dimos/recordings/session_xarm7_20260622_120000.db \ --config dimos/learning/dataprep/example_config.json ``` The config maps recorded streams to dataset features — which stream is the observation image, which is the state, what rate to resample at. Copy `example_config.json` and adjust; the A1Z ships its own at `dataprep/galaxea_a1z_state_config.json`. By default the **action** for each frame is the *next* frame's measured joint state, which is what next-state behavioral cloning expects. Output is a LeRobot v3 dataset by default; pass `-f hdf5` for HDF5. Check what you built: ```bash theme={null} dimos dataprep inspect data/datasets/session ``` You get features, shapes, dtypes, and episode counts — worth a look before spending GPU hours. Each dataset also carries a `dimos_meta.json` recording exactly how it was built. ## 3. Train Training happens in [LeRobot](https://github.com/huggingface/lerobot), pointed straight at your DataPrep output. No upload step, no format shuffling: ```bash theme={null} uv sync --extra lerobot # once uv run lerobot-train \ --dataset.repo_id=my_task \ --dataset.root=./data/datasets/session \ --policy.type=act \ --policy.device=cuda \ --policy.push_to_hub=false \ --output_dir=outputs/my_task_act \ --job_name=my_task_act \ --wandb.enable=false ``` Use `--policy.device=mps` on Apple silicon or `cpu` for a slow smoke test. Policy quality depends on demonstration count, consistency, and coverage; a completed training run does not by itself mean the policy will generalize. The checkpoint you deploy lands at: ``` outputs/my_task_act/checkpoints/last/pretrained_model ``` ## 4. Run it on the robot `LeRobotPolicyModule` wraps the checkpoint as a DimOS module: camera frames and joint states in, joint commands out, at the policy's control rate. On the A1Z it is one command: ```bash theme={null} uv run dimos a1z run-policy \ outputs/my_task_act/checkpoints/last/pretrained_model \ --task "pick up the object" \ --duration 20 ``` It will not surprise you: loading and hardware initialization ask for confirmation, and inference starts only once live camera and joint observations are flowing. For other arms, compose `LeRobotPolicyModule` into a blueprint next to a camera and the coordinator. The module never moves hardware at startup — its `execute_learned_policy` skill starts inference explicitly, and `stop_learned_policy` halts it with the robot holding position. For the complete A1Z dependency setup, exact dataset conversion command, camera selection, safety notes, and multi-policy agent blueprint, use the [Galaxea A1Z hardware and learning guide](/dimos/robot/manipulators/galaxea_a1z/README.md). ## When something looks wrong Work backwards through the loop. A policy behaving strangely is usually a dataset problem; a dataset problem is usually a recording problem. * `dimos dataprep inspect` — do shapes, rates, and episode counts match what you expect? * `dimos a1z replay` — does the recorded motion look like what you demonstrated? * Recordings from before the `coordinator_joint_state` stream rename need a config pointing at the old name — or just re-record.