{"slug": "alloy-physics-engine-for-robot-learning", "title": "Alloy physics engine for robot learning", "summary": "Argon Robotics released Alloy, an open-source multi-material, parallel physics engine for robot learning built on Newton and MuJoCo Warp and licensed under Apache-2.0, installable via `pip install alloy-engine` on Python 3.11 or 3.12. Alloy steps robots, cloth, cables, liquids, sand and soft bodies together in one scene, runs the same script on an NVIDIA GPU or CPU, and ships with 5,423 library objects, 9 ready-made tasks and quadruped walking, plus a sim.health() check that distinguishes physics failures from policy failures. The engine is aimed at AI agents building training tasks from in-package docs, which Argon Robotics says agents used to build new tasks in minutes.", "body_md": "# A multi-material, parallel physics engine for robot learning.\n\nBuilt on Newton and MuJoCo Warp. Open source under [Apache-2.0](https://github.com/argonrobotics-ai/alloy/blob/main/LICENSE).\n\nYou\n\n`pip install alloy-engine`\nYour agent\nBuild and train \n\n*<your task>*with Alloy\n## see prompt\n\nInstall alloy-engine with pip (Python 3.11 or 3.12, in a virtual environment). Read its docs: run python -m alloy.docs, or fetch https://alloy.argonrobotics.ai/llms.txt. Then build <your task> in many parallel worlds: start from the closest ready-made environment in the docs and change it step by step. Train a policy until it succeeds, check sim.health() after every run, and report the success rate.\n\n## Built for agents.\n\n- **Docs in the package**`python -m alloy.docs` shows them on your machine.\n- **Build a new task in minutes** Agents built new training tasks from the docs in minutes.\n- **Health in every world** Tells a physics failure from a policy failure.\n- **Settings choose themselves** Solvers and steps are picked;`sim.routes()` says why.\n\n## Key features.\n\n- **Many materials, one scene** Robots, cloth, cables, liquids, sand and soft bodies, stepped together. Cloth and a cable beside a robot run slowly.\n- **GPU or CPU** The same script, renderer included, runs on an NVIDIA GPU or the CPU.\n- **A camera for every world** Colour, depth and segmentation of every world at once.\n- **Robots ready to go** Arms, a hand, a quadruped and two humanoids, with RL and demo recording.\n- **Assets and tasks included** 5,423 library objects, 9 ready-made tasks (pick-and-place, pouring, drawers, cloth folding) and quadruped walking.", "url": "https://wpnews.pro/news/alloy-physics-engine-for-robot-learning", "canonical_source": "https://alloy.argonrobotics.ai/", "published_at": "2026-11-17 21:57:48+00:00", "updated_at": "2026-11-17 22:13:23.511545+00:00", "lang": "en", "topics": ["robotics", "ai-research", "ai-tools", "developer-tools", "ai-agents"], "entities": ["Alloy", "Argon Robotics", "Newton", "MuJoCo Warp", "NVIDIA", "Apache-2.0", "Python"], "also_reported_by": [], "takeaway": "Alloy runs the same robot-learning script on NVIDIA GPU or CPU and steps robots alongside cloth, cables, liquids, sand and soft bodies in one scene, with sim.health() separating physics failures from policy failures after each run.", "recommendation": null, "alternates": {"html": "https://wpnews.pro/news/alloy-physics-engine-for-robot-learning", "markdown": "https://wpnews.pro/news/alloy-physics-engine-for-robot-learning.md", "text": "https://wpnews.pro/news/alloy-physics-engine-for-robot-learning.txt", "jsonld": "https://wpnews.pro/news/alloy-physics-engine-for-robot-learning.jsonld"}}