# Ai2 — ToyBlocks: A controlled world for robot learning and evaluation

- Company: Ai2 (allenai.org)
- Announced: 2026-11-06
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://allenai.org/blog/toyblocks
- Record: https://forck.live/items/32619-toyblocks-a-controlled-world-for-robot-learning-and-evaluation
- Subject: Olmo / Molmo / Asta
- Models affected: Molmo 2, ManiFlow, MolmoAct 2, MolmoBot, MolmoSpaces

Ai2 is introducing FoundationWorlds, a framework for building controlled simulated environments to study robot learning, with ToyBlocks as the first world. ToyBlocks is an open simulation framework built around a robot arm and blocks on a tabletop, where the robot must reason about how blocks interact while pushing, flipping and stacking them. The setting is deliberately simple, but Ai2 says its narrow scope makes it possible to meet four requirements: realistic physics and appearance, automatic generation of tasks and demonstrations at scale, challenges that combine reasoning with dexterity, and broad coverage of the tasks possible within a domain. ToyBlocks automatically generates demonstrations without teleoperation, and researchers can vary training mixtures, for example generating more pushing or flipping demonstrations and evaluating on tasks kept out of training. Skill training runs in Isaac Lab, Nvidia's open-source framework for robot learning in simulation, using reinforcement learning and motion planning to learn non-prehensile skills such as pushing a block or flipping it onto another face. A task-and-motion planner sequences skills and plans movements, and ToyBlocks generates and executes hundreds of plans in parallel on a single GPU, reaching up to 24 times real-time performance. Ai2 says it plans to extend the approach to increasingly complex environments beyond ToyBlocks.

## Evidence

Verbatim from https://allenai.org/blog/toyblocks:

> Robots are learning to fold laundry, sort parts, and handle unfamiliar objects. But those demonstrations are often rather narrow and don't explain what a robot has learned or why it fails.
> 
> To study a complex system, scientists often begin with a small, controlled environment, such as a petri dish or terrarium, where they can vary conditions and observe the effects. Robotics needs environments that make this kind of investigation possible—researchers should be able to change the training data or model design, repeat a test, and trace the cause of a failure.
> 
> Today we're introducing a framework for creating these environments; we call them FoundationWorlds. Each pairs a controlled, simulated world targeted at studying robotics in a specific environment with tools for automatically generating tasks and training demonstrations, as well as metrics for measuring a robot's behavior over time. Together, they provide a setting for studying questions such as how models improve as more training data becomes available, and how to best combine high-level reasoning with precise physical control.

---

Record: https://forck.live/items/32619-toyblocks-a-controlled-world-for-robot-learning-and-evaluation
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md
