{ "dataset": "simulators", "updated": "2026-07-09", "entities": [ { "id": "nvidia-isaac-sim", "name": "NVIDIA Isaac Sim + Isaac Lab", "org": "NVIDIA", "org_country": "US", "license": "Apache-2.0 (composants Omniverse Kit propriétaires)", "open_source": true, "physics_engine": "PhysX 5 (+ moteur Newton intégré dans Isaac Lab)", "gpu_accelerated": true, "strengths_en": "Photorealistic RTX rendering, massively parallel GPU physics and an end-to-end pipeline from synthetic data generation to robot learning with Isaac Lab.", "strengths_fr": "Rendu RTX photoréaliste, physique GPU massivement parallèle et pipeline complet de la génération de données synthétiques à l'apprentissage robotique avec Isaac Lab.", "typical_use": ["RL training", "synthetic data", "sim2real", "digital twin"], "ros_support": true, "notable_en": "Isaac Sim 5.0 and Isaac Lab 2.2 reached general availability at SIGGRAPH 2025, with the Isaac Sim application open-sourced on GitHub under Apache-2.0 (the underlying Omniverse Kit remains closed source). It is the de facto industry platform for humanoid and manipulation policy training, now compatible with the Newton physics engine.", "notable_fr": "Isaac Sim 5.0 et Isaac Lab 2.2 sont en disponibilité générale depuis SIGGRAPH 2025, l'application Isaac Sim étant open source sur GitHub sous Apache-2.0 (le socle Omniverse Kit reste propriétaire). C'est la plateforme industrielle de référence pour l'entraînement de politiques humanoïdes et de manipulation, désormais compatible avec le moteur Newton.", "sources": [ { "url": "https://developer.nvidia.com/blog/isaac-sim-and-isaac-lab-are-now-available-for-early-developer-preview", "publisher": "NVIDIA Technical Blog", "date": "2025-08-11", "fields": ["open_source", "license", "notable_en", "notable_fr", "ros_support"] }, { "url": "https://github.com/isaac-sim/IsaacSim", "publisher": "GitHub (isaac-sim)", "date": "2026-07-09", "fields": ["license", "open_source", "physics_engine"] }, { "url": "https://developer.nvidia.com/isaac/sim", "publisher": "NVIDIA Developer", "date": "2026-07-09", "fields": ["strengths_en", "strengths_fr", "typical_use", "gpu_accelerated"] } ], "last_verified": "2026-07-09" }, { "id": "mujoco", "name": "MuJoCo", "org": "Google DeepMind", "org_country": "GB", "license": "Apache-2.0", "open_source": true, "physics_engine": "MuJoCo (moteur natif, contacts articulés)", "gpu_accelerated": true, "strengths_en": "Reference-grade contact dynamics accuracy and speed for control research, with GPU/TPU acceleration via MJX (JAX) and MuJoCo Warp.", "strengths_fr": "Précision et vitesse de référence pour la dynamique des contacts en recherche en contrôle, avec accélération GPU/TPU via MJX (JAX) et MuJoCo Warp.", "typical_use": ["RL training", "sim2real", "model-based control", "benchmarking"], "ros_support": null, "notable_en": "Open-sourced by DeepMind in 2022, MuJoCo remains the most cited physics engine in robot learning research; the 3.x series added MJX for accelerator hardware, and MuJoCo Warp (GPU-optimized, up to ~70x faster on humanoids) also powers a solver inside NVIDIA's Newton engine.", "notable_fr": "Passé en open source par DeepMind en 2022, MuJoCo reste le moteur physique le plus cité en recherche sur l'apprentissage robotique ; la série 3.x a ajouté MJX pour le matériel accéléré, et MuJoCo Warp (optimisé GPU, jusqu'à ~70x plus rapide sur les humanoïdes) sert aussi de solveur dans le moteur Newton de NVIDIA.", "sources": [ { "url": "https://github.com/google-deepmind/mujoco", "publisher": "GitHub (google-deepmind)", "date": "2026-07-09", "fields": ["license", "open_source", "physics_engine", "gpu_accelerated"] }, { "url": "https://github.com/google-deepmind/mujoco_warp", "publisher": "GitHub (google-deepmind)", "date": "2026-07-09", "fields": ["gpu_accelerated", "notable_en", "notable_fr"] }, { "url": "https://github.com/google-deepmind/mujoco_playground", "publisher": "GitHub (google-deepmind)", "date": "2026-07-09", "fields": ["typical_use", "strengths_en", "strengths_fr"] } ], "last_verified": "2026-07-09" }, { "id": "genesis", "name": "Genesis", "org": "Genesis AI (issu d'un projet académique multi-universités)", "org_country": "US", "license": "Apache-2.0", "open_source": true, "physics_engine": "Moteur multi-physique unifié maison (rigide, MPM, SPH, FEM, PBD)", "gpu_accelerated": true, "strengths_en": "Ultra-fast fully Pythonic multi-physics simulation (rigid, soft, fluids) with differentiable solvers and a generative data engine driven by natural language.", "strengths_fr": "Simulation multi-physique ultra-rapide et entièrement pythonique (rigide, souple, fluides), avec solveurs différentiables et moteur de génération de données piloté en langage naturel.", "typical_use": ["RL training", "synthetic data", "soft robotics", "generative simulation"], "ros_support": null, "notable_en": "Launched in December 2024 as an academic collaboration, Genesis became one of the fastest-adopted robotics simulators ever and its development is now backed by the startup Genesis AI (repackaged as genesis-world). It loads MJCF, URDF and mesh formats and targets the full spectrum from rigid robots to soft-body and fluid simulation.", "notable_fr": "Lancé en décembre 2024 comme collaboration académique, Genesis est devenu l'un des simulateurs robotiques adoptés le plus rapidement, et son développement est désormais soutenu par la startup Genesis AI (repackagé en genesis-world). Il charge les formats MJCF, URDF et maillages, et couvre le spectre complet du robot rigide à la simulation de corps souples et de fluides.", "sources": [ { "url": "https://github.com/Genesis-Embodied-AI/Genesis", "publisher": "GitHub (Genesis-Embodied-AI)", "date": "2026-07-09", "fields": ["license", "open_source", "physics_engine", "gpu_accelerated", "notable_en", "notable_fr"] }, { "url": "https://github.com/Genesis-Embodied-AI/genesis-world", "publisher": "GitHub (Genesis-Embodied-AI)", "date": "2026-07-09", "fields": ["org", "notable_en", "notable_fr"] }, { "url": "https://github.com/Genesis-Embodied-AI/Genesis/blob/main/README.md", "publisher": "GitHub (Genesis-Embodied-AI)", "date": "2026-07-09", "fields": ["strengths_en", "strengths_fr", "typical_use"] } ], "last_verified": "2026-07-09" }, { "id": "gazebo", "name": "Gazebo (nouvelle génération, ex-Ignition)", "org": "Open Source Robotics Foundation (Open Robotics)", "org_country": "US", "license": "Apache-2.0", "open_source": true, "physics_engine": "Pluggable : DART (défaut), Bullet, TPE", "gpu_accelerated": false, "strengths_en": "The standard open-source simulator of the ROS ecosystem, with pluggable physics engines, high-fidelity sensor models and mature multi-robot and CI workflows.", "strengths_fr": "Le simulateur open source standard de l'écosystème ROS, avec moteurs physiques interchangeables, capteurs haute fidélité et workflows multi-robots et CI matures.", "typical_use": ["CI testing", "ROS development", "digital twin", "multi-robot simulation"], "ros_support": true, "notable_en": "Gazebo Jetty, the 10th major release, shipped on September 30, 2025 as an LTS supported until September 2030. Physics runs on CPU (GPU is used for rendering and sensors), which keeps it more suited to systems integration and testing than to massively parallel RL training.", "notable_fr": "Gazebo Jetty, la 10e version majeure, est sortie le 30 septembre 2025 en LTS supportée jusqu'en septembre 2030. La physique tourne sur CPU (le GPU sert au rendu et aux capteurs), ce qui le destine davantage à l'intégration système et aux tests qu'à l'entraînement RL massivement parallèle.", "sources": [ { "url": "https://discourse.openrobotics.org/t/gazebo-jetty-released/50349", "publisher": "Open Robotics Discourse", "date": "2025-09-30", "fields": ["notable_en", "notable_fr"] }, { "url": "https://github.com/gazebosim/gz-sim", "publisher": "GitHub (gazebosim)", "date": "2026-07-09", "fields": ["license", "open_source", "physics_engine", "strengths_en", "strengths_fr"] }, { "url": "https://gazebosim.org/docs/latest/install/", "publisher": "gazebosim.org", "date": "2026-07-09", "fields": ["notable_en", "notable_fr", "typical_use"] } ], "last_verified": "2026-07-09" }, { "id": "pybullet", "name": "PyBullet / Bullet Physics", "org": "Communauté Bullet Physics (Erwin Coumans)", "org_country": "US", "license": "zlib", "open_source": true, "physics_engine": "Bullet", "gpu_accelerated": false, "strengths_en": "Zero-friction install and a simple Python API that made it the historical workhorse of robot RL research, still valued for prototyping and teaching.", "strengths_fr": "Installation sans friction et API Python simple qui en ont fait le cheval de trait historique de la recherche en RL robotique, toujours apprécié pour le prototypage et l'enseignement.", "typical_use": ["RL training", "prototyping", "education", "benchmarking"], "ros_support": null, "notable_en": "Development has slowed to maintenance mode (latest release 3.2.7, January 2025) and most new RL research has migrated to GPU-native simulators, but PyBullet remains widely used in education and as a lightweight CPU baseline.", "notable_fr": "Le développement est passé en mode maintenance (dernière version 3.2.7, janvier 2025) et l'essentiel de la recherche RL a migré vers des simulateurs natifs GPU, mais PyBullet reste très utilisé en enseignement et comme référence CPU légère.", "sources": [ { "url": "https://pypi.org/project/pybullet/", "publisher": "PyPI", "date": "2025-01-30", "fields": ["license", "notable_en", "notable_fr"] }, { "url": "https://github.com/bulletphysics/bullet3", "publisher": "GitHub (bulletphysics)", "date": "2026-07-09", "fields": ["license", "open_source", "physics_engine", "org"] }, { "url": "https://pybullet.org/", "publisher": "pybullet.org", "date": "2026-07-09", "fields": ["strengths_en", "strengths_fr", "typical_use"] } ], "last_verified": "2026-07-09" }, { "id": "webots", "name": "Webots", "org": "Cyberbotics", "org_country": "CH", "license": "Apache-2.0", "open_source": true, "physics_engine": "ODE (fork Open Dynamics Engine)", "gpu_accelerated": false, "strengths_en": "Turnkey cross-platform simulator with a large library of ready-made robots and sensors, strong ROS 2 support and a long track record in education and research.", "strengths_fr": "Simulateur clé en main multiplateforme avec une large bibliothèque de robots et capteurs prêts à l'emploi, un bon support ROS 2 et un long historique en enseignement et recherche.", "typical_use": ["education", "prototyping", "ROS development", "competitions"], "ros_support": true, "notable_en": "Born at EPFL in 1996 and open-sourced in December 2018, Webots is the leading European (EMEA) robot simulator; the latest stable release R2025a (February 2025) added new robot models and improved ROS 2 support, with nightly builds still active in 2026.", "notable_fr": "Né à l'EPFL en 1996 et passé en open source en décembre 2018, Webots est le premier simulateur robotique européen (EMEA) ; la dernière version stable R2025a (février 2025) a ajouté de nouveaux modèles de robots et amélioré le support ROS 2, avec des builds nightly toujours actifs en 2026.", "sources": [ { "url": "https://github.com/cyberbotics/webots/releases/tag/R2025a", "publisher": "GitHub (cyberbotics)", "date": "2025-02-04", "fields": ["notable_en", "notable_fr", "ros_support"] }, { "url": "https://github.com/cyberbotics/webots", "publisher": "GitHub (cyberbotics)", "date": "2026-07-09", "fields": ["license", "open_source", "physics_engine"] }, { "url": "https://cyberbotics.com/", "publisher": "Cyberbotics", "date": "2026-07-09", "fields": ["org", "org_country", "strengths_en", "strengths_fr", "typical_use"] } ], "last_verified": "2026-07-09" }, { "id": "coppeliasim", "name": "CoppeliaSim (ex-V-REP)", "org": "Coppelia Robotics", "org_country": "CH", "license": "Propriétaire (Edu gratuite pour l'enseignement, Pro commerciale)", "open_source": false, "physics_engine": "Multiples au choix : Bullet, ODE, Newton Dynamics, Vortex, MuJoCo", "gpu_accelerated": false, "strengths_en": "Highly scriptable all-in-one simulator with switchable physics engines and a distributed control architecture, popular for factory-cell and algorithm validation.", "strengths_fr": "Simulateur tout-en-un hautement scriptable avec moteurs physiques interchangeables et architecture de contrôle distribuée, populaire pour la validation d'îlots industriels et d'algorithmes.", "typical_use": ["prototyping", "education", "factory simulation", "algorithm validation"], "ros_support": true, "notable_en": "Successor to V-REP, developed in Switzerland; version 4.10 shipped in May 2025. The Edu edition is free but restricted to schools and universities, while commercial use requires a CoppeliaSim Pro license.", "notable_fr": "Successeur de V-REP, développé en Suisse ; la version 4.10 est sortie en mai 2025. L'édition Edu est gratuite mais réservée aux écoles et universités, tandis que l'usage commercial requiert une licence CoppeliaSim Pro.", "sources": [ { "url": "https://manual.coppeliarobotics.com/en/licensing.htm", "publisher": "Coppelia Robotics", "date": "2026-07-09", "fields": ["license", "open_source", "notable_en", "notable_fr"] }, { "url": "https://www.coppeliarobotics.com/previousVersions", "publisher": "Coppelia Robotics", "date": "2025-05-14", "fields": ["notable_en", "notable_fr"] }, { "url": "https://www.coppeliarobotics.com/", "publisher": "Coppelia Robotics", "date": "2026-07-09", "fields": ["org", "org_country", "physics_engine", "strengths_en", "strengths_fr", "typical_use"] } ], "last_verified": "2026-07-09" }, { "id": "newton-physics", "name": "Newton", "org": "NVIDIA + Google DeepMind + Disney Research (gouvernance Linux Foundation)", "org_country": "US", "license": "Apache-2.0", "open_source": true, "physics_engine": "Moteur propre construit sur NVIDIA Warp (solveurs incluant MuJoCo Warp et VBD déformables)", "gpu_accelerated": true, "strengths_en": "GPU-accelerated, differentiable and extensible physics engine built on NVIDIA Warp and OpenUSD, unifying solvers (including MuJoCo Warp) for rigid, deformable and multiphysics robot learning.", "strengths_fr": "Moteur physique GPU, différentiable et extensible construit sur NVIDIA Warp et OpenUSD, unifiant des solveurs (dont MuJoCo Warp) pour l'apprentissage robotique rigide, déformable et multi-physique.", "typical_use": ["RL training", "sim2real", "differentiable physics", "deformable simulation"], "ros_support": null, "notable_en": "Announced at GTC in March 2025 and released in open beta at CoRL 2025 under Linux Foundation governance, Newton is the major new entrant of 2025-2026: it plugs into Isaac Lab and MuJoCo Playground, and is already used by Skild AI and Samsung for fine manipulation and deformable-object training.", "notable_fr": "Annoncé à la GTC en mars 2025 et publié en bêta ouverte à CoRL 2025 sous gouvernance Linux Foundation, Newton est le grand nouvel entrant 2025-2026 : il s'intègre à Isaac Lab et MuJoCo Playground, et est déjà utilisé par Skild AI et Samsung pour la manipulation fine et l'entraînement sur objets déformables.", "sources": [ { "url": "https://developer.nvidia.com/blog/announcing-newton-an-open-source-physics-engine-for-robotics-simulation/", "publisher": "NVIDIA Technical Blog", "date": "2025-03-18", "fields": ["org", "physics_engine", "gpu_accelerated", "strengths_en", "strengths_fr"] }, { "url": "https://nvidianews.nvidia.com/news/nvidia-accelerates-robotics-research-and-development-with-new-open-models-and-simulation-libraries", "publisher": "NVIDIA Newsroom", "date": "2025-09-29", "fields": ["open_source", "notable_en", "notable_fr"] }, { "url": "https://github.com/newton-physics", "publisher": "GitHub (newton-physics)", "date": "2026-07-09", "fields": ["license", "open_source"] } ], "last_verified": "2026-07-09" } ] }