InternUtopia 2.2.1

Bug Fixes

• Handle exception during rigid body initialization. (@BrigantineX )
• Fix memory leak caused by incomplete controller cleanup. (@APX103 )
• Fix rigid body filter logic. (@uzuku )

Full Changelog: v2.2.0...v2.2.1

InternUtopia 2.2.0

New Features

• Support multiple episodes. (@APX103 )
• Introduce Vector Env. (@APX103 )
• Support Multi-GPU and Multi-Node. (@BrigantineX )

Dependency

• Upgrade Isaac Sim to 4.5.0.

Breaking Changes

• Use task_configs to declare episode config instead of task_config.
• Rename packages grutopia.* -> internutopia.*.
• Use interface classes instead of Isaac Sim native classes in extension. This may break the extension code written based on an older version of InternUtopia. The extension can be fixed by following the latest documentation.

Full Changelog: v2.1.0...v2.2.0

GRUtopia 2.1.0

Bug Fixes

• Fix zero robot case. (@APX103)

Improvements

• Accelerate point cloud calculation. (@uzuku)
• Disable UJITSO collision cooking to load GRScenes properly. (@uzuku)
• Support resuming dataset download. (@uzuku)

New Features

• Layout editing with mocap. (@Keep-changing)

Miscellaneous

• Improve documentation and fix typos. (@uzuku)
• Add simple e2e tests. (@uzuku)
• Add CONTRIBUTING.md, issue templates and update CI configurations. (@uzuku)

Full Changelog: v2.0.2...v2.1.0

GRUtopia 2.0.2

Full Changelog: v2.0.1...v2.0.2

GRUtopia 2.0.1

Full Changelog: v2.0.0...v2.0.1

GRUtopia 2.0.0

We're excited to announce that GRUtopia 2.0.0 is released!

Python package of this release can be found on PyPI.

New Features in 2.0.0:

• Use standard Gym Env.
• Use pythonic config system to support clearer and extendable configuration.
• Add diverse robots and corresponding policies
• Support teleportation with Mocap and VisionPro.
• Benchmark and baseline for social navigation and mobile manipulation task.
• Add more realistic and diverse scenes for simulation.
• Provide a large number of refined interactive object models in the scenes.
• Provide semantic annotations for a large number of objects in the scenes.
• Procedural Indoor Scene Generation with GRScenes-100.