Shiyun Xiong, Dongming Wu, Peiwen Sun, Bokang Yang, Jiaxu Wang, Xiaolou Sun, Mingxing Wen, Xiangyu Yue · 2026 · arXiv
DOI: 10.48550/arXiv.2610.26751 · open-access PDF
Los sistemas encarnados recientes se basan cada vez más en modelos de gran lenguaje multimodales (MLLM) por sus capacidades de comprensión y razonamiento pre-entrenados, pero su percepción del mundo físico sigue siendo en gran medida limitada a modalidades básicas como la visión y el audio. Sin embargo, ampliar los MLLM a modalidades sensoriales más diversas del entorno físico es un reto: sus representaciones a menudo son débiles en conexión con el MLLM y varían sustancialmente en estructura y estadística, mientras que su interpretación todavía requiere una comprensión y razonamiento compartidos de alto nivel.
Machine translation (Meta M2M100); the English original follows.
Recent embodied systems increasingly build on multimodal large language models (MLLMs) for their pretrained understanding and reasoning capabilities, but their perception of the physical world remains largely limited to basic modalities such as vision and audio. However, broadening MLLMs to more diverse sensory modalities from the physical environment is challenging: their representations are often weakly connected to the MLLM and vary substantially in structure and statistics, while their interpretation still requires shared high-level understanding and reasoning. We therefore propose Unisens
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.26751, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange