Release Notes

1.0.preview4

General

• This release aligns the PhysX implementation in standalone Preview Isaac Gym with Omniverse Isaac Sim 2022.1 to simplify migration to Omniverse for RL workloads

New Features

• PhysX backend: Added support for SDF collisions with a nut & bolt example.

• PhysX backend: Enabled gyroscopic forces by default to improve simulation (can disable with asset_options.enable_gyroscopic_forces = False).

• PhysX backend: Allow customizing rest_offset and contact_offset per asset and per individual shape.

• Added parsing of spherical (ball) joints support in URDF importer.

Bug fixes and improvements

• PhysX: Improved elastic collision behaviour.

• PhysX: Fixed a bug when resetting some fixe-base actors.

• Fixed an issue with PhysX material property caching affecting dynamically changing friction properties, such as during domain randomization.

• Fixed an issue with friction mode for triangle meshes and heightfields; in previous releases this was incorrectly set to multiply mode. With this release we use average mode, like all other shapes in the simulation.

1.0.preview3

Driver Requirement

• NVIDIA Driver version 470+ is now required.

API Changes

• Force sensors are now defined on assets, not individual actors (see force sensor documentation).

• Added optional ForceSensorProperties to fine-tune how forces are reported (see force sensor documentation).

• Added GPU mass-matrices API.

• Added API to create non-even terrain.

• Changed default contact collection mode from CC_LAST_SUBSTEP to CC_ALL_SUBSTEPS, which is more accurate but may be slower when running with multiple substeps. It can be overriden by SimParams.physx.contact_collection.

RL Environment Changes

• RL framework and all RL environments are moved to https://github.com/NVIDIA-Omniverse/IsaacGymEnvs.

• RL docs can now be found at https://github.com/NVIDIA-Omniverse/IsaacGymEnvs/blob/main/README.md and in the docs folder. Please refer to the framework docs for changes required to update existing RL environments.

Bug fixes and improvements

• Increased the limits on the number of actor bodies and joints with PhysX.

• Fixed a bug with randomization of rigid body properties with the GPU pipeline.

• Fixed a bug with setting DOF position targets with the GPU pipeline.

• Fixed inclined plane rendering.

• Updated docker base image to https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel_21-09.html

• Added viewer support when running in docker.

• Fixed instabilities related to actor scaling in GPU simulations. There are still some limitation with changing mass properties when running with the GPU pipeline, see actor scaling.

1.0.preview2

API + Command Line Parameter Changes

• Command line arguments for simulation API and device selection have changed, and are now aligned between the RL examples and general programming examples:

  • --physx_gpu command line option has been removed.

  • --sim_device is used to specify a device for simulation. It can take either cpu or cuda:n as options.

  • --pipeline is used to explicitly choose either the cpu or gpu tensor pipeline API.

• Command line arguments for setting the experiment name were improved and extended:

  • --experiment command line option replaced --experiment_name.

  • --metadata flag was added. When used with --experiment <experiment name> additional information about the physics engine, sim device, pipeline and domain randomization is added to the experiment name.

• Refactored the API for applying rigid body forces:

  • The new body force API allows applying torques and using different coordinate spaces for force, torque, and position vectors.

  • The apply_body_force function was replaced by apply_body_forces and apply_body_force_at_pos (see API docs).

  • The apply_rigid_body_force_tensor function was replaced by apply_rigid_body_force_tensors and apply_rigid_body_force_at_pos_tensors (see tensor API docs).

• Simulation determinism has been improved:

  • A --seed parameter can be set in RL examples to specify what random seed to use. Explicitly setting a seed should produce deterministic results between runs.

  • By default, the seed is set to -1, which will generate a random seed for each run.

  • The --torch_deterministic parameter can now be set to force additional determinism in PyTorch operations, at the expense of performance.

  • For more details, please refer to the Reproducibility section on the Reinforcement Learning Examples page.

• The base RL task class has been updated to prepare for future multi-gpu training support:

  • Any user environment inherited from the RL base task class should be updated as well

• Asset root and file paths can now be read in from cfg files.

• The method attach_camera_to_body now takes named enums instead of an integer argument for the camera attachment mode.

• Contents in docker image are now placed under /opt instead of /workspace.

New Features

• Asset handling

  • Added support for convex decomposition of collision meshes during asset import.

  • Now uses Assimp for mesh loading. This means that many different types of meshes can be specified in MJCF or URDF assets (including .dae, .stl, .obj, etc.).

  • Added support for loading materials and (embedded and non-embedded) textures from meshes in URDF and MJCF files.

  • Added support for overriding normals loaded from meshes with either smooth vertex normals or face normals.

  • Added flags for explicitly overriding inertia tensors and center of mass in AssetOptions.

  • Added support for visual boxes and material loading from MJCF.

  • See Assets for additional information

• Updated graphical user interface and visualization options

  • Added API support for getting mouse position and window size from the viewer

  • See Simulation Setup for additional information

• Updates related to Shadow Hand Environment and RL framework features

  • Added adaptive KL scheduling to the default rl-pytorch RL framework. This was previously available only in the rl_games RL framework.

  • Added different observation variants in the Shadow Hand environment for a closer match to OpenAI’s Learning Dexterity project: https://openai.com/blog/learning-dexterity/

  • Added support of asymmetric observations and Shadow Hand training examples using them.

  • Added examples of training with LSTM policy and value functions with rl_games.

  • Added support of setting control frequency to be lower than a simulation frequency and controlFrequencyInv parameter to the yaml configs to specify how many simulation steps per one control step should be performed.

  • Added correlated noise to Domain Randomization options.

  • Added support for custom distributions of actor parameters for domain randomization.

• Other Asset and Example updates:

  • Added new training environments: ANYmal quadruped robot, Quadcopter and NASA Ingenuity helicopter.

• Scalability related containerization updates:

  • Added support for Python 3.8.

  • Added support for headless rendering in docker with available graphics driver.

  • Improved CUDA context handling to prepare for multi-GPU training support

• Additional new features:

  • Added support for runtime scaling of actors.

  • Added support for recomputing the inertia tensors of rigid bodies when their mass is changed.

  • Added support for specifying position offset when applying force to bodies.

  • Changed the default values of max_depenetration_velocity and bounce_threshold_velocity.

  • Added CoordinateSpace enum to specify position offsets in local, env, or global space.

Bug Fixes

• Fixed issue with observation and action noise.

• Fixed joint limit ranges for CMU and NV humanoids.

• Fixed axes bug for humanoid training.

• Fixed incorrect visualization of collision meshes with PhysX backend.

• Fixed a bug with env spacing in z-up simulations.

• Fixed a bug where meshes, cylinders, or ellipsoids imported from MJCF could have incorrect friction properties.

• Fixed a bug where primitive shapes created procedurally could have incorrect thickness in Flex.

• Fixed a possible crash when getting net contact force tensor on GPU.

• Fixed submitting mixed control tensors in GPU pipeline.

• Fixed issues with z-up camera view matrix calculation, lookAt function, and mouse drag direction.

• Fixed rigid body property getter that was returning erroneous inertia tensors.

• PhysX: Fixed occasional crash with aggregates on GPU.

• PhysX: Fixed possible buffer overflow in convex-capsule collision on GPU.

• PhysX: Fixed stability issues with small meshes.

• PhysX: Improvements to TGS restitution.

• PhysX: Fixed issue with applying body forces in GPU pipeline.

• PhysX: Fixed issue with applying body torques in GPU pipeline.

• PhysX: Fixed various issues causing non-determinism.

• Fixed synchronization issues in GPU pipeline.

• Fixed issue with z-up camera view matrix calculation.

• Fixed issues with setting the rigid shape properties of an actor.

• Improved error checking for input tensors.

• Improved error reporting when CPU-only functions get called during simulation with the GPU pipeline.

• Fixed a bug in computing transform inverse.

• Fixed a Flex crash on startup caused by a driver bug.

• Fixed a bug with ground plane friction in the body_physics_props example.

1.0.preview1

New Features

• Implemented end-to-end GPU pipeline for physics simulation, which allows interacting with simulations on the GPU without copying data to or from the host.

• Added new Tensor API for physics state and control, on both CPU and GPU.

• Added Pytorch interop utilities for the Tensor API.

• Added a new simple framework for reinforcement learning and a collection of sample tasks.

• Added new configurable domain randomization features.

• Added support for fixed and spatial tendons with PhysX.

• Added support for user-defined force sensors attached to articulation links with PhysX.

• Exposed DOF forces in PhysX.

• Added Jacobian and generalized mass matrices with PhysX.

• Improved PVD support - can connect to PVD remotely or log to file.

• Improved contact handling with multiple substeps.

• Added support for multiple subscenes to parallelize computations with CPU PhysX.

• Improved contact handling performance in PhysX.

• Support for rigid dynamic actors in PhysX to increase performance with single-body actors.

• Support for custom aggregates with PhysX.

• Exposed joint armature and joint friction in PhysX.

• Added support for soft contacts with Flex.

• Added stress tensors with Flex.

• Added pneumatic pressure/target tensors with Flex.

• Added soft materials with Flex.

• Added new simulation and asset/actor options for Flex and PhysX.

• Parsing tendon definitions from MJCF.

• Loading cylinder geometry from MJCF.

• Loading visual meshes from MJCF.

• Generating filters from contact specification in MJCF.

• Improved support for multiple sensor cameras per env.

• Improved Z-up simulation support.

• Updated PhysX and FleX versions.

• Viewer sync can be toggled by pressing V (disabling sync increases performance, especially with the GPU pipeline).

• Improved setup scripts.

• Updated documentation and examples.

• Updated docker images.

Removed Features

• Removed Python multiprocessing support, which is superseded by the new Tensor API.

• Removed old rlbase module and examples, which are replaced by the new RL framework designed around the Tensor API.

• Removed old RTX renderer.

Changes

• Renamed package from carbongym to isaacgym.

• Improved quaternion-Euler conversion utilities.

Bug Fixes

• Numerous physics bug fixes and simulation stability improvements.

• Fixes and improvements to URDF and MJCF importers.

• Fixes and improvements to camera sensors.

Known Issues and Limitations

• Most of the rigid body tensor API is only available with PhysX.

• Soft body support is currently only available with FleX.

• Missing tensor API for setting all rigid body states: root and DOF state tensors can be used instead.

• Missing API for configuring properties of spatial tendons.

• When using the GPU pipeline, DOF states don’t refresh in the viewer.