diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 6832e0f..84ebccb 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -name: Autodata CI +name: AutoData CI on: pull_request: diff --git a/.github/workflows/nightly.yml b/.github/workflows/nightly.yml index 09f8e34..c855a84 100644 --- a/.github/workflows/nightly.yml +++ b/.github/workflows/nightly.yml @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -name: Autodata Nightly +name: AutoData Nightly # 03:30 UTC is off-peak and offset from other repos' nightlies to reduce GPU-runner contention. on: diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 5a7cc69..f9216b7 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -1,6 +1,6 @@ -## Autodata Contribution Rules +## AutoData Contribution Rules -This document describes the rules for contributing to Autodata +This document describes the rules for contributing to AutoData #### Signing Your Work diff --git a/CONTRIBUTORS.md b/CONTRIBUTORS.md index 822cd85..2b20333 100644 --- a/CONTRIBUTORS.md +++ b/CONTRIBUTORS.md @@ -1,6 +1,6 @@ -# Autodata Developers and Contributors +# AutoData Developers and Contributors -This is the official list of Autodata project developers and contributors. +This is the official list of AutoData project developers and contributors. Guidelines for modifications: diff --git a/README.md b/README.md index d0fe36e..d22af22 100644 --- a/README.md +++ b/README.md @@ -1,6 +1,6 @@
-# Autodata +# AutoData ### Scalable Robot Demonstration Generation for Robot Learning @@ -20,27 +20,27 @@ ## Overview -**Autodata** is a trajectory data-generation framework built on +**AutoData** is a trajectory data-generation framework built on [NVIDIA Isaac Lab](https://github.com/isaac-sim/IsaacLab) and [Isaac Lab-Arena](https://github.com/isaac-sim/IsaacLab-Arena). Given a small set of annotated human demonstrations, it uses parallel simulation environments to generate diverse datasets of successful robot demonstrations for robot learning. -Autodata splits demonstrations into object-relative skill segments. During generation, it transforms those +AutoData splits demonstrations into object-relative skill segments. During generation, it transforms those segments to new scene configurations, connects them into complete trajectories, executes them in simulation, and records successful demonstrations in HDF5 datasets.

- Autodata generating robot demonstrations in parallel + AutoData generating robot demonstrations in parallel

-## Why Autodata? +## Why AutoData? Robot-learning policies require large and diverse collections of successful demonstrations. Gathering all of that data through human teleoperation is slow and expensive, even though a small set of demonstrations often already contains the task's essential skills. -Autodata scales those demonstrations across randomized object placements and scene configurations, reducing +AutoData scales those demonstrations across randomized object placements and scene configurations, reducing the amount of manual collection needed to produce datasets for policy training. ## Key Features @@ -162,25 +162,25 @@ For test-suite details and common commands, see ## License -Autodata is released under the [Apache License 2.0](LICENSE.md). +AutoData is released under the [Apache License 2.0](LICENSE.md). Third-party open-source software notices are provided in [THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md). Additional component-specific licenses are stored in [`docs/licenses/`](docs/licenses/). NVIDIA cuRobo is governed by the [NVIDIA Isaac Lab Additional Software and Materials License](docs/licenses/curobo-license.txt). -Autodata depends on Isaac Sim, which includes components distributed under proprietary licensing terms. See +AutoData depends on Isaac Sim, which includes components distributed under proprietary licensing terms. See the [Isaac Sim license](https://docs.isaacsim.omniverse.nvidia.com/latest/common/NVIDIA_Omniverse_License_Agreement.html) for details. ## Citation -If you use Autodata in your research, please cite: +If you use AutoData in your research, please cite: ```bibtex @misc{autodata2026, - title = {Autodata: Scalable Robot Demonstration Generation for Robot Learning}, - author = {{NVIDIA Autodata Contributors}}, + title = {AutoData: Scalable Robot Demonstration Generation for Robot Learning}, + author = {{NVIDIA AutoData Contributors}}, year = {2026}, url = {https://github.com/isaac-sim/AutoData} } @@ -193,7 +193,7 @@ Depending on the generation algorithm used, please also cite the original ## Acknowledgements -Autodata builds on NVIDIA Isaac Sim, Isaac Lab, and Isaac Lab-Arena. Its data-generation workflows incorporate +AutoData builds on NVIDIA Isaac Sim, Isaac Lab, and Isaac Lab-Arena. Its data-generation workflows incorporate ideas from MimicGen, DexMimicGen, and SkillMimicGen, with cuRobo providing GPU-accelerated motion planning for SkillGen workflows. @@ -204,6 +204,6 @@ foundational work.
-**Autodata** · [Documentation](https://isaac-sim.github.io/AutoData/) · [GitHub](https://github.com/isaac-sim/AutoData) +**AutoData** · [Documentation](https://isaac-sim.github.io/AutoData/) · [GitHub](https://github.com/isaac-sim/AutoData)
diff --git a/THIRD_PARTY_NOTICES.md b/THIRD_PARTY_NOTICES.md index 405ed2a..f5abdc4 100644 --- a/THIRD_PARTY_NOTICES.md +++ b/THIRD_PARTY_NOTICES.md @@ -5,7 +5,7 @@ SPDX-License-Identifier: Apache-2.0 # Third-Party Software Notices -Autodata includes or directly depends on the following open-source software. Each entry reproduces the +AutoData includes or directly depends on the following open-source software. Each entry reproduces the component's copyright and license notice and links to the canonical license text. ## IsaacLab-Arena diff --git a/autodata_core/__init__.py b/autodata_core/__init__.py index aafd21a..edf52bf 100644 --- a/autodata_core/__init__.py +++ b/autodata_core/__init__.py @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -"""Core data generation runtime for Autodata. +"""Core data generation runtime for AutoData. Public entrypoints: diff --git a/autodata_tests/utils/utils.py b/autodata_tests/utils/utils.py index 6041f1f..f9cbde4 100644 --- a/autodata_tests/utils/utils.py +++ b/autodata_tests/utils/utils.py @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -"""Shared assertion helpers for the Autodata test suite.""" +"""Shared assertion helpers for the AutoData test suite.""" import h5py import json diff --git a/conda_installer.sh b/conda_installer.sh index a3cc113..f48997a 100755 --- a/conda_installer.sh +++ b/conda_installer.sh @@ -17,7 +17,7 @@ usage() { Usage: $(basename "$0") [-c] [-i] [-h] -c Create conda env "${ENV_NAME}" with Python ${PYTHON_VERSION} - -i Install Isaac Sim, PyTorch, Isaac Lab, Arena, and Autodata + -i Install Isaac Sim, PyTorch, Isaac Lab, Arena, and AutoData into the "${ENV_NAME}" env (env must already exist; run with -c first) -h Show this help @@ -110,7 +110,7 @@ install_all() { echo ">>> Installing Isaac Lab - Arena" uv pip install --editable "${arena_root}" - echo ">>> Installing Autodata" + echo ">>> Installing AutoData" uv pip install --editable "${REPO_ROOT}" echo ">>> Installation complete." diff --git a/docker/Dockerfile.autodata b/docker/Dockerfile.autodata index 054a32a..021a309 100644 --- a/docker/Dockerfile.autodata +++ b/docker/Dockerfile.autodata @@ -40,7 +40,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \ ####################################### # Copy the Arena submodule once as it carries the nested IsaacLab checkout. # The .dockerignore drops Arena's other nested submodules (e.g. Isaac-GR00T) -# which Autodata does not use. +# which AutoData does not use. COPY submodules/IsaacLab-Arena ${WORKDIR}/submodules/IsaacLab-Arena # --- IsaacLab --- @@ -60,7 +60,7 @@ RUN /isaac-sim/python.sh -m pip install --force-reinstall daqp==0.8.5 RUN /isaac-sim/python.sh -m pip install -e ${WORKDIR}/submodules/IsaacLab-Arena ################## -# Autodata +# AutoData ################## # Copy the package sources + packaging metadata (exclude submodules installed above). COPY setup.py pyproject.toml README.md sitecustomize.py ${WORKDIR}/ @@ -96,7 +96,7 @@ RUN echo 'export PATH="/isaac-sim/kit/python/bin:${CUDA_HOME}/bin:${PATH}"' >> / && echo "alias pytest='/isaac-sim/python.sh -m pytest'" >> /etc/bash.bashrc \ && echo "alias pre-commit='/isaac-sim/python.sh -m pre_commit'" >> /etc/bash.bashrc \ && echo "alias ll='ls -alF --color=auto'" >> /etc/bash.bashrc \ - && echo "PS1='[Autodata] \[\e[0;32m\]\u \[\e[0;34m\]\w\[\e[0m\] \$ '" >> /etc/bash.bashrc \ + && echo "PS1='[AutoData] \[\e[0;32m\]\u \[\e[0;34m\]\w\[\e[0m\] \$ '" >> /etc/bash.bashrc \ && cp /etc/bash.bashrc /root/.bashrc # Set the entrypoint diff --git a/docker/run_docker.sh b/docker/run_docker.sh index 94a8022..051bdc0 100755 --- a/docker/run_docker.sh +++ b/docker/run_docker.sh @@ -56,7 +56,7 @@ while getopts ":d:crRvh" OPTION; do v) set -x ;; h) script_name=$(basename "$0") - echo "Build and run the Autodata dev container." + echo "Build and run the AutoData dev container." echo "" echo "Usage: $script_name [options] [command...]" echo "" @@ -149,7 +149,7 @@ DOCKER_RUN_ARGS=( "--env" "ACCEPT_EULA=Y" "--env" "PRIVACY_CONSENT=Y" "--env" "ISAACLAB_PATH=${ISAACLAB_PATH}" - # Make Autodata's sitecustomize compatibility hook visible to direct upstream script entrypoints. + # Make AutoData's sitecustomize compatibility hook visible to direct upstream script entrypoints. "--env" "PYTHONPATH=${WORKDIR}" # Used by the entrypoint to recreate the host user inside the container. "--env" "DOCKER_RUN_USER_ID=$(id -u)" diff --git a/docs/README.md b/docs/README.md index a0a5bdf..d2b9d00 100644 --- a/docs/README.md +++ b/docs/README.md @@ -1,4 +1,4 @@ -# Autodata Documentation — Developer Guide +# AutoData Documentation — Developer Guide The docs are built on the **host machine** (not inside Docker) using a dedicated Python 3.12 venv. diff --git a/docs/_ext/autodata_doc_tools.py b/docs/_ext/autodata_doc_tools.py index de99398..feb9f27 100644 --- a/docs/_ext/autodata_doc_tools.py +++ b/docs/_ext/autodata_doc_tools.py @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -"""Small source-read macros for the Autodata docs. +"""Small source-read macros for the AutoData docs. Mirrors the Arena docs tooling (``isaaclab_arena_doc_tools``): text substitutions applied to every RST page before parsing, so recurring snippets (docker commands, code links, clone URLs) are defined diff --git a/docs/_redirect/index.html b/docs/_redirect/index.html index 9b66fa5..05c298e 100644 --- a/docs/_redirect/index.html +++ b/docs/_redirect/index.html @@ -1,7 +1,7 @@ - Redirecting to the latest Autodata documentation + Redirecting to the latest AutoData documentation diff --git a/docs/conf.py b/docs/conf.py index 874c59b..2ea42c1 100644 --- a/docs/conf.py +++ b/docs/conf.py @@ -17,7 +17,7 @@ # -- Project information ----------------------------------------------------- -project = "Autodata" +project = "AutoData" copyright = "2026, NVIDIA" author = "NVIDIA" released = False # Indicates if this is a public or internal version of the repo. @@ -70,7 +70,7 @@ # The theme to use for HTML and HTML Help pages. See the documentation for # a list of builtin themes. html_theme = "nvidia_sphinx_theme" -html_title = f"Autodata {AUTODATA_VERSION_NUMBER}" +html_title = f"AutoData {AUTODATA_VERSION_NUMBER}" html_show_sphinx = False html_theme_options = { "copyright_override": {"start": 2026}, diff --git a/docs/images/System_Architecture.svg b/docs/images/System_Architecture.svg index 3713692..64b2f1e 100644 --- a/docs/images/System_Architecture.svg +++ b/docs/images/System_Architecture.svg @@ -1,5 +1,5 @@ - Autodata system architecture: contracts, typed data flow, modular generation and direct execution + AutoData system architecture: contracts, typed data flow, modular generation and direct execution @@ -23,7 +23,7 @@ - Autodata — System Architecture + AutoData — System Architecture Contracts, typed data flow, modular generation and direct execution diff --git a/docs/index.rst b/docs/index.rst index 51dd53b..781d636 100644 --- a/docs/index.rst +++ b/docs/index.rst @@ -1,7 +1,7 @@ -Welcome to Autodata! +Welcome to AutoData! ==================== -``Autodata`` is a trajectory data-generation framework built on top of +``AutoData`` is a trajectory data-generation framework built on top of `Isaac Lab `_ and `Isaac Lab-Arena `_. Given a handful of annotated human demonstrations, it uses parallel simulation environments to @@ -11,9 +11,9 @@ demonstration segments. .. figure:: images/autodata.gif :width: 100% :align: center - :alt: Autodata + :alt: AutoData - Autodata + AutoData The Problem @@ -24,14 +24,14 @@ demonstrations, and collecting those by human teleoperation is slow and expensiv what a policy needs to learn from a thousand demonstrations is already contained in ten: the same skill repeated under different object placements. -Autodata exploits that redundancy. A human demonstration is split into **subtasks** (each +AutoData exploits that redundancy. A human demonstration is split into **subtasks** (each a contiguous segment in which the robot's end-effector motion is driven by a single reference object). Because each segment is object-relative, it can be *transformed* to a new scene configuration and replayed. Stitching transformed segments together turns a handful of demonstrations into an arbitrarily large dataset. -Autodata +AutoData ======== Four pieces cooperate to generate data: @@ -53,9 +53,9 @@ Four pieces cooperate to generate data: :width: 100% :align: center :target: _images/System_Architecture.svg - :alt: Autodata system architecture — contracts, typed data flow, modular generation and execution + :alt: AutoData system architecture — contracts, typed data flow, modular generation and execution - The Autodata architecture: declarative contracts feed the Datastream read interface, + The AutoData architecture: declarative contracts feed the Datastream read interface, which the data generator and its algorithm plug-ins consume to produce waypoints, actions, and finally recorded HDF5 episodes. @@ -89,7 +89,7 @@ first generation with :doc:`pages/quickstart/first_data_generation`. License ======= -Autodata is licensed under the `Apache License 2.0 +AutoData is licensed under the `Apache License 2.0 `_. diff --git a/docs/pages/concepts/algorithms.rst b/docs/pages/concepts/algorithms.rst index d2bca1c..0502158 100644 --- a/docs/pages/concepts/algorithms.rst +++ b/docs/pages/concepts/algorithms.rst @@ -1,7 +1,7 @@ Generation Algorithms ===================== -Autodata ships three generation algorithms, selected with ``--alg`` on the CLI (the +AutoData ships three generation algorithms, selected with ``--alg`` on the CLI (the task descriptor's ``algo`` field independently selects which per-subtask ``algo_params`` schema the descriptor is parsed with — keep the two consistent). All three plug into the :doc:`data generator ` and share the same skeleton — split diff --git a/docs/pages/concepts/concept_overview.rst b/docs/pages/concepts/concept_overview.rst index fe4605a..4fe9c94 100644 --- a/docs/pages/concepts/concept_overview.rst +++ b/docs/pages/concepts/concept_overview.rst @@ -1,7 +1,7 @@ Architecture Overview ===================== -Autodata is organized around one idea: **the generation machinery should not know the +AutoData is organized around one idea: **the generation machinery should not know the simulator or the robot directly.** Everything the generator needs is reachable through small, explicit interfaces — a task-descriptor YAML for the task, an embodiment YAML for the robot, and a read interface (the Datastream) for the live environment. New tasks, robots, and diff --git a/docs/pages/concepts/task_descriptors.rst b/docs/pages/concepts/task_descriptors.rst index 997b800..dc1a771 100644 --- a/docs/pages/concepts/task_descriptors.rst +++ b/docs/pages/concepts/task_descriptors.rst @@ -1,7 +1,7 @@ Task Descriptors ================ -A **task descriptor** is a YAML file that declares everything Autodata needs to know +A **task descriptor** is a YAML file that declares everything AutoData needs to know about a task: its subtasks per end-effector, the signals that mark subtask boundaries, cross-arm constraints, and the data generation policy for the task. It is the single source of truth shared by the annotation tool and the data generator. Both load it, so annotations and generation can @@ -74,7 +74,7 @@ Subtasks A **subtask** is a contiguous segment of a demonstration in which an end-effector's motion is dictated by a single object, such as reaching for a cube or placing it on another cube. A subtask ends and the next one begins when the object dictating the motion changes. During -generation, Autodata selects a recorded subtask segment, transforms it into the current scene, connects the +generation, AutoData selects a recorded subtask segment, transforms it into the current scene, connects the robot's current trajectory to the segment, and replays it. The ``subtasks`` section of the YAML schema groups subtasks by ````. The eef name key must exactly match diff --git a/docs/pages/quickstart/installation.rst b/docs/pages/quickstart/installation.rst index a5df95a..7bbecfc 100644 --- a/docs/pages/quickstart/installation.rst +++ b/docs/pages/quickstart/installation.rst @@ -1,8 +1,8 @@ Installation ============ -Docker is the recommended way to install Autodata. The dev container setup includes Isaac Sim, -Isaac Lab, Isaac Lab-Arena, and Autodata, providing a reproducible environment +Docker is the recommended way to install AutoData. The dev container setup includes Isaac Sim, +Isaac Lab, Isaac Lab-Arena, and AutoData, providing a reproducible environment without modifying the host Python installation. The repository is bind-mounted into the container, so edits on the host are live inside it. @@ -126,7 +126,7 @@ From the repository root, create the ``autodata`` environment with Python 3.12: ./conda_installer.sh -c Activate the environment and install Isaac Sim, CUDA-enabled PyTorch, Isaac Lab, Isaac Lab-Arena, -and Autodata: +and AutoData: .. code-block:: bash @@ -159,7 +159,7 @@ SkillGen additionally requires cuRobo. Before installing it, review the NVIDIA c ``setup_conda_env.sh``. Those scripts set ``PYTHONHOME`` and ``PYTHONPATH`` to use Kit's bundled packages, which can cause conda to fail during the cuRobo installation. -Activate the Autodata environment, install the CUDA 12.8 toolkit, and configure the build for your +Activate the AutoData environment, install the CUDA 12.8 toolkit, and configure the build for your GPU's compute capability: .. code-block:: bash @@ -171,7 +171,7 @@ GPU's compute capability: export LD_LIBRARY_PATH="$CUDA_HOME/lib:$LD_LIBRARY_PATH" export TORCH_CUDA_ARCH_LIST="$(nvidia-smi --query-gpu=compute_cap --format=csv,noheader | head -n1)+PTX" -Install the cuRobo commit tested with Isaac Lab and used by the Autodata cuRobo container: +Install the cuRobo commit tested with Isaac Lab and used by the AutoData cuRobo container: .. code-block:: bash diff --git a/docs/pages/quickstart/support_matrix.rst b/docs/pages/quickstart/support_matrix.rst index 2a28893..5272669 100644 --- a/docs/pages/quickstart/support_matrix.rst +++ b/docs/pages/quickstart/support_matrix.rst @@ -5,12 +5,12 @@ Support Matrix ============== -This matrix describes the supported Autodata v0.1.0 stack. Upstream projects may support -additional platforms, but configurations outside this matrix are not validated by Autodata. Use -the recursively pinned submodules from the Autodata repository instead of independently selecting +This matrix describes the supported AutoData v0.1.0 stack. Upstream projects may support +additional platforms, but configurations outside this matrix are not validated by AutoData. Use +the recursively pinned submodules from the AutoData repository instead of independently selecting Isaac Lab or Isaac Lab-Arena revisions. -.. list-table:: Autodata platform and resource support +.. list-table:: AutoData platform and resource support :widths: 25 75 :header-rows: 1 @@ -33,7 +33,7 @@ Isaac Lab or Isaac Lab-Arena revisions. * - Disk - Minimum: 100 GB SSD. Recommended: 500 GB SSD * - Docker installation - - Docker Engine with the NVIDIA Container Toolkit; recommended Autodata installation + - Docker Engine with the NVIDIA Container Toolkit; recommended AutoData installation * - Conda installation - Optional Linux installation using conda, ``uv``, and Python 3.12 * - Isaac Sim diff --git a/docs/pages/references/release_notes.rst b/docs/pages/references/release_notes.rst index 939f679..e94ef4b 100644 --- a/docs/pages/references/release_notes.rst +++ b/docs/pages/references/release_notes.rst @@ -4,7 +4,7 @@ Release Notes v0.1.0 ------ -This initial release of Autodata provides a standalone framework for transforming a small +This initial release of AutoData provides a standalone framework for transforming a small set of annotated demonstrations into larger robot-learning datasets. It brings data-generation algorithm parity with Isaac Lab Mimic without depending on the Isaac Lab Mimic package and covers the workflow from demonstration annotation through parallel generation, validation, and replay. @@ -29,12 +29,12 @@ Key features of this release include: opt-in cuRobo image for SkillGen, and an optional conda installation with pinned Isaac Sim, Isaac Lab, and Isaac Lab-Arena revisions. - **Migration and developer documentation:** A migration guide for moving Isaac Lab Mimic tasks to - Autodata, a published platform support matrix, and unit, end-to-end, and data-generation + AutoData, a published platform support matrix, and unit, end-to-end, and data-generation performance tests. Known limitations: -- **Platform support:** Autodata currently supports Linux x86_64 systems with an NVIDIA RTX GPU. +- **Platform support:** AutoData currently supports Linux x86_64 systems with an NVIDIA RTX GPU. See the :doc:`support matrix <../quickstart/support_matrix>` for the complete hardware and software requirements. - **SkillGen:** SkillGen is currently single-arm only and requires the optional cuRobo diff --git a/docs/pages/references/troubleshooting.rst b/docs/pages/references/troubleshooting.rst index c31fee5..89cd483 100644 --- a/docs/pages/references/troubleshooting.rst +++ b/docs/pages/references/troubleshooting.rst @@ -251,7 +251,7 @@ network access before debugging cuRobo or planner settings. SkillGen Planning Failures -------------------------- -An occasional planning failure is expected: when ``guarantee_success: true``, Autodata resets the +An occasional planning failure is expected: when ``guarantee_success: true``, AutoData resets the scene and retries. If every attempt fails, first rerun the same generation command with these diagnostic options: diff --git a/docs/pages/workflows/franka_cube_stack_mimicgen/index.rst b/docs/pages/workflows/franka_cube_stack_mimicgen/index.rst index 662e1ba..01f21fa 100644 --- a/docs/pages/workflows/franka_cube_stack_mimicgen/index.rst +++ b/docs/pages/workflows/franka_cube_stack_mimicgen/index.rst @@ -1,7 +1,7 @@ Franka Cube Stacking with MimicGen ================================== -This example demonstrates the Autodata workflow using MimicGen to generate a synthetic dataset +This example demonstrates the AutoData workflow using MimicGen to generate a synthetic dataset for a Franka robot performing a cube-stacking task. The workflow covers recording source demonstrations by teleoperation, annotating their subtask boundaries, generating a large dataset with MimicGen, and validating the result. diff --git a/docs/pages/workflows/franka_cube_stack_mimicgen/step_1_record_demonstrations.rst b/docs/pages/workflows/franka_cube_stack_mimicgen/step_1_record_demonstrations.rst index dd8329c..a435825 100644 --- a/docs/pages/workflows/franka_cube_stack_mimicgen/step_1_record_demonstrations.rst +++ b/docs/pages/workflows/franka_cube_stack_mimicgen/step_1_record_demonstrations.rst @@ -1,7 +1,7 @@ Step 1: Record Source Demonstrations ------------------------------------ -Autodata consumes source demonstrations recorded as Isaac Lab HDF5 datasets (per-episode +AutoData consumes source demonstrations recorded as Isaac Lab HDF5 datasets (per-episode actions, initial state, and observations). This step collects a small set of successful teleoperated demonstrations of the cube-stacking task. diff --git a/docs/pages/workflows/franka_cube_stack_mimicgen/step_2_annotate_demonstrations.rst b/docs/pages/workflows/franka_cube_stack_mimicgen/step_2_annotate_demonstrations.rst index 1950132..df876e2 100644 --- a/docs/pages/workflows/franka_cube_stack_mimicgen/step_2_annotate_demonstrations.rst +++ b/docs/pages/workflows/franka_cube_stack_mimicgen/step_2_annotate_demonstrations.rst @@ -1,13 +1,13 @@ Step 2: Annotate Demonstrations ------------------------------- -Before Autodata generation, each source demonstration must be annotated with **subtask termination +Before AutoData generation, each source demonstration must be annotated with **subtask termination signals**: the action indices where one subtask ends and the next begins. The subtasks and their termination signal names are declared by the task descriptor (see :doc:`../../concepts/task_descriptors`). For this task, the subtasks are ``grasp_1``, ``stack_1``, and ``grasp_2`` (the final subtask ends with the trajectory and needs no explicit signal). -Autodata supports two annotation modes: +AutoData supports two annotation modes: * **Manual** — replay each episode in a Kit window and mark boundaries with the keyboard. * **Automatic** (``--auto``) — sample the environment's boolean subtask-term observations diff --git a/docs/pages/workflows/humanoid_dexmimicgen/index.rst b/docs/pages/workflows/humanoid_dexmimicgen/index.rst index 55fba22..37a324f 100644 --- a/docs/pages/workflows/humanoid_dexmimicgen/index.rst +++ b/docs/pages/workflows/humanoid_dexmimicgen/index.rst @@ -1,7 +1,7 @@ Humanoid Pick-and-Place with DexMimicGen ======================================== -This example demonstrates the Autodata workflow using DexMimicGen to generate a +This example demonstrates the AutoData workflow using DexMimicGen to generate a synthetic pick-and-place dataset for a bimanual humanoid robot. It covers recording source demonstrations by Apple Vision Pro teleoperation, annotating their per-arm subtask boundaries, generating a large dataset with DexMimicGen, and validating the result. diff --git a/docs/pages/workflows/humanoid_dexmimicgen/step_1_record_demonstrations.rst b/docs/pages/workflows/humanoid_dexmimicgen/step_1_record_demonstrations.rst index c48cebc..5b9de2d 100644 --- a/docs/pages/workflows/humanoid_dexmimicgen/step_1_record_demonstrations.rst +++ b/docs/pages/workflows/humanoid_dexmimicgen/step_1_record_demonstrations.rst @@ -8,7 +8,7 @@ Step 1: Record Source Demonstrations ------------------------------------ -Autodata consumes source demonstrations recorded as Isaac Lab HDF5 datasets (per-episode +AutoData consumes source demonstrations recorded as Isaac Lab HDF5 datasets (per-episode actions, initial state, and observations). This step collects a small set of successful teleoperated demonstrations of the humanoid pick-and-place task. @@ -16,7 +16,7 @@ Unlike the single-arm Franka task — which can be teleoperated with a SpaceMous dexterous, bimanual humanoids are teleoperated with an **Apple Vision Pro** through `NVIDIA IsaacTeleop `_ and the CloudXR runtime. The headset's wrist poses drive a differential IK controller per arm, and the finger joints are retargeted onto -the robot's hands. The Autodata development container includes IsaacTeleop, so the flow below +the robot's hands. The AutoData development container includes IsaacTeleop, so the flow below works out of the box. .. note:: @@ -39,7 +39,7 @@ works out of the box. Start the CloudXR Runtime ^^^^^^^^^^^^^^^^^^^^^^^^^^ -The CloudXR runtime bridges the Apple Vision Pro and the simulator. Start it from the Autodata dev +The CloudXR runtime bridges the Apple Vision Pro and the simulator. Start it from the AutoData dev container. Leave it running in this terminal for the whole recording session. #. On the **host machine** outside of the development container, configure the firewall to allow CloudXR traffic: @@ -60,7 +60,7 @@ container. Leave it running in this terminal for the whole recording session. sudo ufw allow 48000/udp sudo ufw allow 48002/udp -#. Start the Autodata dev container: +#. Start the AutoData dev container: :docker_run_default: @@ -93,7 +93,7 @@ container. Leave it running in this terminal for the whole recording session. Start Recording ^^^^^^^^^^^^^^^ -#. In **another terminal**, attach a second shell to the running Autodata container: +#. In **another terminal**, attach a second shell to the running AutoData container: :docker_run_default: diff --git a/docs/pages/workflows/humanoid_dexmimicgen/step_2_annotate_demonstrations.rst b/docs/pages/workflows/humanoid_dexmimicgen/step_2_annotate_demonstrations.rst index 2d1463f..2a2f761 100644 --- a/docs/pages/workflows/humanoid_dexmimicgen/step_2_annotate_demonstrations.rst +++ b/docs/pages/workflows/humanoid_dexmimicgen/step_2_annotate_demonstrations.rst @@ -1,7 +1,7 @@ Step 2: Annotate Demonstrations ------------------------------- -Before Autodata generation, each source demonstration must be annotated with **subtask +Before AutoData generation, each source demonstration must be annotated with **subtask termination signals**: the action indices where one subtask ends and the next begins. The subtasks and their termination signal names are declared **per end-effector** by the task descriptor (see :doc:`../../concepts/task_descriptors`). diff --git a/docs/pages/workflows/migrate_isaac_lab_mimic.rst b/docs/pages/workflows/migrate_isaac_lab_mimic.rst index 693e340..ceb2af8 100644 --- a/docs/pages/workflows/migrate_isaac_lab_mimic.rst +++ b/docs/pages/workflows/migrate_isaac_lab_mimic.rst @@ -2,17 +2,17 @@ Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. SPDX-License-Identifier: Apache-2.0 -Migration from Isaac Lab Mimic to Autodata: Franka Cube Stacking +Migration from Isaac Lab Mimic to AutoData: Franka Cube Stacking ================================================================ -This guide migrates the Franka cube-stacking task from Isaac Lab Mimic to Autodata. Every step +This guide migrates the Franka cube-stacking task from Isaac Lab Mimic to AutoData. Every step uses the completed Franka files and commands in this repository. Apply the same mapping to the corresponding files when migrating another Isaac Lab Mimic task. -Autodata uses standard Isaac Lab ManagerBasedRLEnv environments. For Franka cube stacking, keep the normal Isaac Lab +AutoData uses standard Isaac Lab ManagerBasedRLEnv environments. For Franka cube stacking, keep the normal Isaac Lab environment and replace the Mimic environment ID ``Isaac-Stack-Cube-Franka-IK-Rel-Mimic-v0`` with -``Isaac-Stack-Cube-Franka-IK-Rel-v0``. The Mimic-specific configuration moves into Autodata's +``Isaac-Stack-Cube-Franka-IK-Rel-v0``. The Mimic-specific configuration moves into AutoData's task and embodiment descriptors. What moves where? @@ -23,7 +23,7 @@ What moves where? :header-rows: 1 * - Isaac Lab Mimic - - Autodata + - AutoData * - Normal scene, actions, reset events, observations, and success condition - Remain in the normal Isaac Lab environment config * - ``MimicEnvCfg.datagen_config`` @@ -33,11 +33,11 @@ What moves where? * - Mimic environment methods for reading EEF poses and converting actions - Embodiment YAML and its registered embodiment adapter * - Mimic environment methods for reading object poses and subtask signals - - Autodata's ``Datastream`` reads them from the normal environment + - AutoData's ``Datastream`` reads them from the normal environment * - Isaac Lab Mimic generation scripts - - Autodata's ``annotate_demos.py`` and ``generate_dataset.py`` + - AutoData's ``annotate_demos.py`` and ``generate_dataset.py`` -Autodata runs on standard Isaac Lab ``ManagerBasedRLEnv`` environments. It does not require +AutoData runs on standard Isaac Lab ``ManagerBasedRLEnv`` environments. It does not require ``ManagerBasedRLMimicEnv``, ``MimicEnvCfg``, or any other import from ``isaaclab_mimic``. Step 1: Use the normal Isaac Lab environment @@ -49,14 +49,14 @@ The Isaac Lab Mimic implementation of Franka cube stacking combines two elements * ``FrankaCubeStackIKRelMimicEnvCfg`` and ``FrankaCubeStackIKRelMimicEnv`` add data-generation configuration and adapter methods. -Autodata uses ``FrankaCubeStackEnvCfg`` directly. Change the environment ID as follows: +AutoData uses ``FrankaCubeStackEnvCfg`` directly. Change the environment ID as follows: .. list-table:: :widths: 35 65 :header-rows: 1 * - Isaac Lab Mimic - - Autodata + - AutoData * - ``Isaac-Stack-Cube-Franka-IK-Rel-Mimic-v0`` - ``Isaac-Stack-Cube-Franka-IK-Rel-v0`` @@ -72,7 +72,7 @@ The Franka cube-stacking environment already satisfies these requirements: its o ``cube_1``, ``cube_2``, and ``cube_3``; its EEF observations are ``eef_pos`` and ``eef_quat``; and its automatic annotation signals are ``grasp_1``, ``stack_1``, and ``grasp_2``. -Step 2: Move Isaac Lab Mimic ManagerBasedRLMimicEnvCfg into an Autodata task descriptor +Step 2: Move Isaac Lab Mimic ManagerBasedRLMimicEnvCfg into an AutoData task descriptor --------------------------------------------------------------------------------------- An Isaac Lab Mimic environment config contains two kinds of data-generation information: @@ -81,7 +81,7 @@ An Isaac Lab Mimic environment config contains two kinds of data-generation info * ``subtask_configs`` describes the ordered object-relative segments that MimicGen transforms and stitches together. -An Autodata task descriptor stores the same information as data rather than Python code. +An AutoData task descriptor stores the same information as data rather than Python code. Franka cube-stacking conversion ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ @@ -104,7 +104,7 @@ First, the run-wide Isaac Lab Mimic settings: self.datagen_config.generation_interpolate_from_last_target_pose = True self.datagen_config.seed = 1 -become the Autodata task descriptor's ``generation_policy``: +become the AutoData task descriptor's ``generation_policy``: .. code-block:: yaml @@ -158,7 +158,7 @@ In ``franka_cube_stack.yaml``, that same subtask is: The other three ``SubTaskConfig`` objects for Franka cube stacking are converted identically in the completed YAML. -Step 3: Move the Isaac Lab Mimic ManagerBasedRLMimicEnv into an Autodata embodiment descriptor +Step 3: Move the Isaac Lab Mimic ManagerBasedRLMimicEnv into an AutoData embodiment descriptor ---------------------------------------------------------------------------------------------- An Isaac Lab Mimic environment wrapper implements the robot-specific interface used during data @@ -168,7 +168,7 @@ generation. Its methods define: * How the environment's action vector encodes an end-effector target. * Which action dimensions control the gripper or other non-pose channels. -In Autodata, this interface is provided by an embodiment adapter configured through an embodiment +In AutoData, this interface is provided by an embodiment adapter configured through an embodiment descriptor. Franka cube-stacking conversion @@ -187,7 +187,7 @@ First, the Isaac Lab Mimic ManagerBasedRLMimicEnv reads the Franka end-effector eef_pos = self.obs_buf["policy"]["eef_pos"][env_ids] eef_quat = self.obs_buf["policy"]["eef_quat"][env_ids] -The embodiment descriptor in Autodata records those observation keys and uses the same ``franka`` EEF name +The embodiment descriptor in AutoData records those observation keys and uses the same ``franka`` EEF name as the task descriptor: .. code-block:: yaml @@ -212,7 +212,7 @@ pose action: The reverse conversion reads position from ``action[:, :3]`` and compact axis-angle rotation from ``action[:, 3:6]``. ``actions_to_gripper_actions()`` reads ``action[:, -1:]``, so the final action -dimension is the gripper command. This is exactly the action convention implemented by Autodata's +dimension is the gripper command. This is exactly the action convention implemented by AutoData's ``delta_pose_ik_single_arm`` adapter: .. code-block:: yaml @@ -227,7 +227,7 @@ The embodiment adapter now provides the pose reads and action conversions, while exposes them to the generator. The corresponding methods are no longer needed on a Mimic environment wrapper. -Putting the pieces together, the complete Autodata Franka embodiment descriptor is: +Putting the pieces together, the complete AutoData Franka embodiment descriptor is: .. code-block:: yaml @@ -252,7 +252,7 @@ Step 4: Reuse or annotate source demonstrations ----------------------------------------------- The Franka cube stacking source demonstrations recorded in Isaac Lab Mimic are already in the -correct HDF5 format. Annotate the raw dataset directly with Autodata: +correct HDF5 format. Annotate the raw dataset directly with AutoData: .. code-block:: bash @@ -269,7 +269,7 @@ correct HDF5 format. Annotate the raw dataset directly with Autodata: Step 5: Run a small generation test ----------------------------------- -Run the migrated Franka cube stacking example in Autodata using the task and embodiment descriptors +Run the migrated Franka cube stacking example in AutoData using the task and embodiment descriptors created in Steps 2 and 3: .. code-block:: bash @@ -285,7 +285,7 @@ created in Steps 2 and 3: --input_file ./datasets/dataset_franka_annotated.hdf5 \ --output_file ./datasets/generated_dataset_franka.hdf5 -This command exercises the normal Isaac Lab environment, Autodata task descriptor, and Autodata +This command exercises the normal Isaac Lab environment, AutoData task descriptor, and AutoData embodiment. Once it succeeds, the Franka migration is complete. Increase ``--num_envs`` and ``--generation_num_trials`` for the full run. See the :doc:`complete Franka cube-stacking workflow ` for recording, diff --git a/docs/pages/workflows/skillgen/preflight.rst b/docs/pages/workflows/skillgen/preflight.rst index e079aa1..04f1080 100644 --- a/docs/pages/workflows/skillgen/preflight.rst +++ b/docs/pages/workflows/skillgen/preflight.rst @@ -11,7 +11,7 @@ Complete this preflight before annotating demonstrations or starting a long Skil run. It verifies the cuRobo image, GPU build, source dataset, Nucleus asset download, and one complete end-to-end generation run. Repeat it after changing GPUs or rebuilding the development image. -This page assumes the base :doc:`Autodata installation <../../quickstart/installation>` is complete +This page assumes the base :doc:`AutoData installation <../../quickstart/installation>` is complete and the host satisfies the :doc:`support matrix <../../quickstart/support_matrix>`. Requirements at a Glance @@ -31,7 +31,7 @@ Requirements at a Glance * - GPU build - cuRobo kernels compiled for the compute capability of the GPU running generation. * - Network - - Outbound access to the Nucleus asset server during planner initialization so Autodata can + - Outbound access to the Nucleus asset server during planner initialization so AutoData can retrieve the Franka URDF. * - Source data - Git LFS objects pulled and a structurally valid SkillGen-annotated HDF5 dataset. @@ -112,7 +112,7 @@ Expected Planning Failures SkillGen plans against a newly randomized scene on every attempt. Some configurations have no path within the planner's search budget, so an individual planning failure is expected. Both shipped -SkillGen descriptors set ``guarantee_success: true``. Autodata counts the failed attempt, resets the +SkillGen descriptors set ``guarantee_success: true``. AutoData counts the failed attempt, resets the scene, and retries until it records the requested number of successful demonstrations. .. list-table:: Interpreting SkillGen failures diff --git a/pyproject.toml b/pyproject.toml index b50a035..16b6e43 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -87,7 +87,7 @@ quiet-level = 0 ignore-words-list = 'reacher' [tool.pytest.ini_options] -# Scope default collection to the Autodata test suite (keeps pytest out of submodules). +# Scope default collection to the AutoData test suite (keeps pytest out of submodules). testpaths = ["autodata_tests"] markers = [ "with_subprocess: test launches the data-generation CLI as an Isaac Sim child process", diff --git a/scripts/annotate_demos.py b/scripts/annotate_demos.py index 76794af..8b64959 100644 --- a/scripts/annotate_demos.py +++ b/scripts/annotate_demos.py @@ -34,7 +34,7 @@ from isaaclab.app import AppLauncher parser = argparse.ArgumentParser( - description="Manually annotate Autodata source demonstrations with subtask signals.", + description="Manually annotate AutoData source demonstrations with subtask signals.", formatter_class=argparse.RawDescriptionHelpFormatter, ) parser.add_argument( diff --git a/scripts/ci/run_tests.sh b/scripts/ci/run_tests.sh index c2e5db7..5ebc5ac 100755 --- a/scripts/ci/run_tests.sh +++ b/scripts/ci/run_tests.sh @@ -2,7 +2,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 # -# Run the Autodata test suite inside the repo's GPU Docker image. +# Run the AutoData test suite inside the repo's GPU Docker image. # This is the single entry point shared by local runs and CI: it wraps # ./docker/run_docker.sh so a developer reproduces a CI failure with one command: # diff --git a/scripts/validate_dataset.py b/scripts/validate_dataset.py index 0ae5769..a53db2d 100644 --- a/scripts/validate_dataset.py +++ b/scripts/validate_dataset.py @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -"""Structurally validate Autodata HDF5 datasets. +"""Structurally validate AutoData HDF5 datasets. Prints a summary table (episode count, env id, sim args) per file followed by any issues. @@ -161,7 +161,7 @@ def print_results(results: list[ValidationResult]) -> None: def main(argv: list[str] | None = None) -> int: """Validate the requested files and return nonzero if any are invalid.""" - parser = argparse.ArgumentParser(description="Validate Autodata demonstration HDF5 datasets.") + parser = argparse.ArgumentParser(description="Validate AutoData demonstration HDF5 datasets.") parser.add_argument("files", nargs="+", help="HDF5 dataset file(s) to validate.") args = parser.parse_args(argv) diff --git a/setup.py b/setup.py index f7962ae..4c412ae 100644 --- a/setup.py +++ b/setup.py @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -"""Installation script for the Autodata Python packages.""" +"""Installation script for the AutoData Python packages.""" from setuptools import find_packages, setup diff --git a/sitecustomize.py b/sitecustomize.py index e1726db..a74fcae 100644 --- a/sitecustomize.py +++ b/sitecustomize.py @@ -1,7 +1,7 @@ # Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: Apache-2.0 -"""Install temporary compatibility patches for Autodata Python workflows.""" +"""Install temporary compatibility patches for AutoData Python workflows.""" from autodata_utils.g1_avp_teleop import install_g1_avp_teleop_patch_if_requested from autodata_utils.isaaclab_compat import install_franka_asset_path_patch