Physix FrontierPhysix Frontier · News Briefing Card (Leiphone · Sep 9, 2026)
GigaAI Publishes 7 Spatial Intelligence Papers at ECCV
KEY FACTS
- GigaAI had 13 papers accepted to ECCV, with seven focusing on spatial intelligence.
- In collaboration with Tsinghua University and the Chinese Academy of Sciences, the team proposed a solution spanning perception, physical simulation, and embodied decision-making.
- The VLA-R1 framework integrates reinforcement learning and chain-of-thought reasoning to enhance robotic arm robustness in complex scenarios.
- OmniNWM unifies state, action, and reward scoring within a driving world model.
- ReconPhys technology reconstructs 3D assets with physical attributes from monocular video in under one second.
KEY DATA
13Total ECCV Accepted Papers
7Spatial Intelligence Papers
13,000 entriesVLA-CoT Data Volume
36.51Affordance Region IoU
PHYSIX OBSERVATION
GigaAI’s latest achievements signal that spatial intelligence is shifting from mere visual generation toward physical interaction grounded in causal reasoning. By addressing traditional AI’s lack of physical common sense and closed-loop control, their tech stack removes key barriers to deploying embodied intelligence. This deep fusion of logical reasoning, real-world physics, and efficient reconstruction suggests robots will accelerate from 'demo units' to industrial-grade tools, warranting close attention to their subsequent commercial validation capabilities.
Source: Leiphone report