Isaac Lab-Arena 0.2 (Alpha) Announcement

[sangeetas-nv]

NVIDIA Isaac Lab-Arena 0.2 Alpha Release
We are excited to announce that Isaac Lab-Arena has moved from pre-Alpha to Alpha stage in version 0.2.
Isaac Lab-Arena is an open-source framework that extends Isaac Lab for simplified task curation and policy evaluation at scale in simulation.

Key Features

• LEGO-like Composable Environments — Mix and match scenes, embodiments, and tasks independently
• On-the-fly Assembly — Environments are built at runtime; no duplicate config files to maintain.
• New Sequential Task Chaining — Chain atomic skills (e.g. Pick + Walk + Place + …) to create complex long-horizon tasks.
• New Natural Language Object Placement — Define scene layouts using semantic relationships like "on" or "next to", instead of manually specified coordinates.
• Integrated Evaluation — Extensible metrics and evaluation pipelines for policy benchmarking
• New Large-scale Parallel Evaluations with Heterogeneous Objects — Evaluate policy on multiple parallel environments, each with different objects, to maximize evaluation throughput.
• New RL Workflow Support and Seamless Interoperation with Isaac Lab — Plug Isaac Lab-Arena environments into Isaac Lab workflows for Reinforcement learning and Data generation for imitation learning.

Ecosystem
NVIDIA and partners are building Industrial and academic benchmarks on the unified Isaac Lab-Arena core, so you can reuse LEGO blocks (tasks, scenes, metrics, and datasets) for your custom evaluations.

• Lightwheel RoboFinals — high fidelity industrial benchmarks
• Lightwheel RoboCasa Tasks — 138+ open-source tasks, 50 datasets per task, 7+ robots
• Lightwheel LIBERO Tasks — Adapted LIBERO benchmarks
• RoboTwin 2.0 — Extended simulation benchmarks using Arena; Arxiv
• LeRobot Environment Hub — Share and discover Arena environments on Hugging Face
• NIST Board 1, NVIDIA Isaac GR00T Industrial Benchmarks, NVIDIA DexBench, NVIDIA RoboLab, and more are coming soon.

Developer preview branches

• A developer preview of Isaac Lab‑Arena 0.2 (based on Isaac Lab 2.3) is now available on main. This early version includes the 0.2 features and is meant for users who can accept some instability.
• Support for Isaac Lab-Arena 0.2 on Isaac Lab 3.0 is underway in a dedicated feature branch isaac_lab_3_newton. This branch is subject to significant changes and instability as Lab 3.0 (Newton) is evolving quickly.
• The official, stable, and tested release of Isaac Lab-Arena 0.2 on Isaac Lab 3.0 is coming soon in April 2026.

Collaboration
Isaac Lab-Arena is being developed as an open-source, shared evaluation framework that the community can collectively enhance and expand. We invite you to try Isaac Lab-Arena 0.2 Alpha, share feedback, and help shape its future. In the Alpha stage, development velocity is high and core features/APIs are evolving. Your input at this stage is especially valuable.

Whats Next
Future releases will focus on agentic, prompt‑first scene and task generation, non-sequential long horizon tasks, easy-to-configure sensitivity analysis with targeted environment variations and evaluation sweeps without code changes, enhanced heterogeneity across parallel evaluations, and VLM‑augmented analysis to surface insights from large‑scale evaluations. These will come with ongoing improvements to performance and usability, such as PIP packaging.