Vision-language-action models, world models, and agentic planners each advance physical intelligence, yet their composition lacks a common execution abstraction, shared state, semantic verification, and persistent experience across heterogeneous embodiments. We present PhyAgentOS…
While Multimodal Large Language Models (MLLMs) excel in general tasks, rigorous scientific reasoning remains challenging due to the limitations of monolithic, linear planning. Such sequential designs often suffer from visual-semantic misalignment, long-context hallucinations, and…
Simultaneous Localization and Mapping (SLAM) is essential for autonomous systems, yet achieving reliable, globally consistent pose estimation and dense mapping in complex environments remains challenging due to geometric degeneracy and sensor drift. While multi-sensor fusion addr…
Online reinforcement learning is widely used to align large language models (LLMs) with reward signals, yet training can be unstable under noisy or misspecified rewards. We identify a failure mode we call directional inconsistency: within a batch, a small set of high-reward rollo…