Boundless Dynamics Open-Sources MWA-Phoenix World Model Base, Tops Six Robot Manipulation Benchmarks
机器之心 · wechat · 2026-10-21
Boundless Dynamics (无界动力) has released and open-sourced MWA™-Phoenix, a 7B latent world-model foundation model for robotics. It tops or leads all six evaluations: LIBERO-Plus (94.63%), LIBERO-Pro (66.48%), RoboCasaGR1 (82.67%), Bench2Dex (45.6%, beating π0.5 by 17.3 points), RoboTwin2.0 (72.42%), and the real-robot RoboDojo leaderboard.
Technical approach: RGB-only input without depth ground truth, using DINOv3 as the vision backbone with dual-teacher distillation fusing VGGT geometric priors and V-JEPA temporal information. The companion Phoenix-ILA jointly encodes motion and tactile signals in one interaction latent space, lifting peeling/wiping success from 20%/75% to 40%/90%. The interactive simulation world model Phoenix-PhysCanvas enables GRPO reinforcement learning before deployment, raising real-robot test-tube success from 58.33% to 75%.
The report also notes sim scores don't guarantee real-robot safety — GPT-6 Astra's RoboDojo real-robot diagnostic run was aborted due to unsafe actions. The company has 200k+ hours of multimodal data; technical report and code are open-sourced.
More from Embodied
- MIT CSAIL's X-Hinges 3D-Prints Everyday Objects With Built-In Sensing Across Three Motion Axes — MIT News AI · 2026-10-23
- Formic raises $50M Series B to scale physical AI across 100+ US factories — garrytan · 2026-10-23
- fal starts fine-tuning a physical robot to draw, results next week — OdinLovis · 2026-10-23
- PEARL demo: an LLM agent that checks experiment results, revises plans, and re-tests — KevinKaichuang · 2026-10-23
- CU Boulder's ATLAS Institute Opens Tenure-Track Robotics Professorship — ryosuzk · 2026-10-23
- MiTaS multi-resolution tactile imitation learning hits 80% success vs 35.5% vision-only baseline — Jan_R_Peters · 2026-10-23