Video World Models · 2026-11-13
BWM-Forge: Agentic World Programming for Embodied Intelligence
BWM Team
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
BWM-Forge is a code-driven framework for controllable robot video generation. It uses a coding agent to build editable Blender scene programs specifying objects, robot motion, and camera states, then renders proxy videos that provide structural and motion references for a video generation model that adds visual detail. A multi-agent harness (preprocessing, building, review, rendering) supports iterative refinement and experience reuse across tasks, robot embodiments, and viewpoints.
Why it matters
The authors target controllable, reusable robot video generation where users can edit scene geometry, robot motion, and camera viewpoints—limitations of end-to-end video generation. By keeping scene and motion in editable programs and using proxy videos as intermediate guidance, the system aims to support synchronized multi-view rendering, adaptation to different robot assets, and appearance variation from a shared scene program.
Method
- Given robot assets and a task description, BWM-Forge accepts an initial image, an image with motion data, or a human demonstration video, then uses a coding agent to construct editable Blender scene programs specifying objects, robot motion, and cameras.
- A multi-agent Harness coordinates preprocessing (asset/motion validation and visual parsing), building (Blender scene construction and animation), review (visual and numerical evidence checks with local corrections), and rendering (generate high-fidelity videos from proxy videos, reference images, and prompts).
- The framework renders proxy videos from executable scene programs and uses them as structural/motion references for a video generation model to add visual detail; it also supports iterative refinement and experience repository reuse.
Abstract (from arXiv)
Generating diverse robot videos requires control over scene geometry, robot motion, and viewpoints. End-to-end video generation offers limited direct access to these variables, making targeted edits and reuse across tasks difficult. We present BWM-Forge, a code-driven framework for controllable video generation in robotics. Given robot assets and a task description, the framework accepts an initial image, an image with motion data, or a human demonstration video. A coding agent constructs editable Blender scene programs that specify objects, robot motion, and cameras. The resulting proxy videos provide structural and motion references for a video generation model, which adds visual detail. A multi-agent harness coordinates input processing, scene construction, review, and rendering. Automated evaluation and user feedback support iterative refinement, while an experience repository stores and retrieves reusable modeling and rendering rules. This design supports multiple robot embodiments, synchronized viewpoints, and appearance variation from a shared scene program. We evaluate the framework on 15 tasks covering five robot types. Compared with LingBot-World Base Cam on these tasks, BWM-Forge achieves higher aesthetic quality, imaging quality, subject and background consistency, motion smoothness, and RoboJudge-9B physical adherence and instruction alignment scores. The baseline scores higher in dynamic degree. Qualitative cases illustrate viewpoint control, scene variation, feedback-driven edits, and experience reuse across scenes. These results demonstrate a practical approach to generating controllable robot videos through editable, executable scene representations.
Related papers
- Video2World: Benchmarking Coding Agents for Interactive World Modeling from Embodied Videos
- Video Generation Meets Agent Harness: Agentic Video Generation, Agentic World Modeling and Recursive Self-Improvement
- ReactWorld: Structuring Interaction Responses for Embodied Video Generation
- RoboRender: Robot-Oriented Video Generation for Visual Sim-to-Real Transfer
- Awomo-SimDataEngine: Agentic Simulation-ReadyWorld Generation
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-11-12 · Indexed by Pub-AI: 2026-11-13 UTC
BWM-Forge: Agentic World Programming for Embodied Intelligence
Preprint posted: 2026-11-12 · Original record ↗
- arxiv
- 2611.16221
Explicit source versions (1)
- v1 · 2026-11-12
- Preprint available
Source date: 2026-11-12 · First detected: 2026-11-13 UTC · Historical baseline
Source evidence (2)
- Official evidence ↗ · RSS guid=oai:arXiv.org:2611.16221v1; arxiv:announce_type=new; item pubDate (official metadata)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
Generating diverse robot videos requires control over scene geometry, robot motion, and viewpoints. End-to-end video generation offers limited direct access to these variables, making targeted edits and reuse across tasks difficult. We present BWM-Forge, a code-driven framework for controllable video generation in robotics. Given robot assets and a task description, the framework accepts an initial image, an image with motion data, or a human demonstration video. A coding agent constructs editable Blender scene programs that specify objects, robot motion, and cameras. The resulting proxy videos provide structural and motion references for a video generation model, which adds visual detail. A multi-agent harness coordinates input processing, scene construction, review, and rendering. Automated evaluation and user feedback support iterative refinement, while an experience repository stores and retrieves reusable modeling and rendering rules. This design supports multiple robot embodiments, synchronized viewpoints, and appearance variation from a shared scene program. We evaluate the framework on 15 tasks covering five robot types. Compared with LingBot-World Base Cam on these tasks, BWM-Forge achieves higher aesthetic quality, imaging quality, subject and background consistency, motion smoothness, and RoboJudge-9B physical adherence and instruction alignment scores. The baseline scores higher in dynamic degree. Qualitative cases illustrate viewpoint control, scene variation, feedback-driven edits, and experience reuse across scenes. These results demonstrate a practical approach to generating controllable robot videos through editable, executable scene representations.
