aryankakad/cube_picking_bspline_diffusion

1---2datasets: aryankakad/cube_picking_bspline3library_name: lerobot4license: apache-2.05model_name: bspline_diffusion6pipeline_tag: robotics7tags:8- lerobot9- robotics10- bspline_diffusion11---12 13# Model Card for bspline_diffusion14 15<!-- Provide a quick summary of what the model is/does. -->16 17 18This is a **bspline_diffusion** policy trained with [LeRobot](https://github.com/huggingface/lerobot).19 20 21 22 23<!-- A short demo is worth more than any description! Record a GIF/video of the policy24running on your robot, upload it to this repo, and embed it here:25<p align="center">26  <img src="https://huggingface.co/<hf_user>/<policy_repo_id>/resolve/main/demo.gif" width="60%"/>27</p>28-->29 30This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).31 32See the [full LeRobot documentation](https://huggingface.co/docs/lerobot/index).33 34 35---36 37## Model Details38 39- **License:** apache-2.040- **Robot type:** `so_follower`41- **Cameras:** `front`, `gripper`42 43 44## Inputs & Outputs45 46The policy consumes these observation features and produces these action features.47 48**Inputs**49 50| Feature | Type | Shape |51| --- | --- | --- |52| `observation.state` | STATE | `(6,)` |53| `observation.images.front` | VISUAL | `(3, 480, 640)` |54| `observation.images.gripper` | VISUAL | `(3, 480, 640)` |55 56**Outputs**57 58| Feature | Type | Shape |59| --- | --- | --- |60| `action` | ACTION | `(112,)` |61 62 63## Training Dataset64 65- **Repository:** [aryankakad/cube_picking_bspline](https://huggingface.co/datasets/aryankakad/cube_picking_bspline)66- **Episodes:** 5067- **Frames:** 2618068- **Frame rate:** 30 FPS69- **Task(s):** "Pick and place cube"70 71<a class="flex" href="https://huggingface.co/spaces/lerobot/visualize_dataset?path=aryankakad/cube_picking_bspline">72<img class="block dark:hidden" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl.svg"/>73<img class="hidden dark:block" src="https://huggingface.co/datasets/huggingface/badges/resolve/main/visualize-this-dataset-xl-dark.svg"/>74</a>75 76 77## Training Configuration78 79| Setting | Value |80| --- | --- |81| Training steps | 100000 |82| Batch size | 32 |83| Optimizer | adam |84| Learning rate | 0.0001 |85| Seed | 1000 |86| LeRobot version | 0.6.1 |87 88---89 90## How to Get Started with the Model91 92New to LeRobot? These guides cover the full workflow:93 94- **[Install LeRobot](https://huggingface.co/docs/lerobot/main/en/installation)** — set up the `lerobot` package.95- **[Hardware setup](https://huggingface.co/docs/lerobot/main/en/hardware_guide)** — assemble, wire, and calibrate your robot and cameras.96- **[Record data & train a policy](https://huggingface.co/docs/lerobot/en/il_robots)** — the end-to-end imitation-learning walkthrough.97- **[CLI cheat-sheet](https://huggingface.co/docs/lerobot/main/en/cheat-sheet)** — quick reference for the `lerobot-*` commands.98 99The short version to run and train this policy:100 101### Run the policy on your robot102 103```bash104lerobot-rollout \105  --strategy.type=base \106  --robot.type=so_follower \107  --robot.port=<your_robot_port> \108  --robot.cameras="{ <camera_1>: {type: opencv, index_or_path: <index_or_path>, width: 640, height: 480, fps: 30}, <camera_2>: {type: opencv, index_or_path: <index_or_path>, width: 640, height: 480, fps: 30}}" \109  --policy.path=aryankakad/cube_picking_bspline_diffusion \110  --task="Pick and place cube" \111  --duration=60112```113 114Replace the remaining `<...>` placeholders with your own values: `--robot.port` and the camera names/indices are specific to your machine, and the camera names must match the observation keys this policy was trained on.115 116When `--strategy.type=base` is used the script doesn't record the episodes. Skipping duration will make the policy run indefinitely. For more information look at [rollout documentation](https://huggingface.co/docs/lerobot/main/en/inference).117 118### Train your own policy119 120```bash121lerobot-train \122  --dataset.repo_id=${HF_USER}/<dataset> \123  --policy.type=bspline_diffusion \124  --output_dir=outputs/train/<policy_repo_id> \125  --job_name=lerobot_training \126  --policy.device=cuda \127  --policy.repo_id=${HF_USER}/<policy_repo_id> \128  --wandb.enable=true129```130 131_Writes checkpoints to `outputs/train/<policy_repo_id>/checkpoints/`._132 133---134 135## Evaluation136 137<!-- Report real-robot results here: run the policy several times per task and count the138successes. Delete the "No evaluation results" line and fill in this table instead:139 140| Task | Trials | Successes | Success rate |141| ---- | ------ | --------- | ------------ |142| pick the lego brick | 10 | 8 | 80% |143 144Also worth noting: anything that affects difficulty (new object positions, lighting,145distractors, a different robot of the same type, ...).146-->147 148_No evaluation results have been provided for this policy yet._149 150---151 152## Citation153 154If you use this policy, please cite the method linked in the description above, along with LeRobot:155 156```bibtex157@misc{cadene2024lerobot,158    author = {Cadene, Remi and Alibert, Simon and Soare, Alexander and Gallouedec, Quentin and Zouitine, Adil and Palma, Steven and Kooijmans, Pepijn and Aractingi, Michel and Shukor, Mustafa and Aubakirova, Dana and Russi, Martino and Capuano, Francesco and Pascal, Caroline and Choghari, Jade and Moss, Jess and Wolf, Thomas},159    title = {LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch},160    howpublished = "\url{https://github.com/huggingface/lerobot}",161    year = {2024}162}163```