With the big data and artificial intelligence explosive development, some new modal data for vision is appeared, like Lidar data, Radar data, and 3D data and so on. Multimodal fusion for vision is one new technology for higher quality human eye presentation or higher performance machine vision tasks. Multimodal fusion system for vision will be popular in the future vision system. Meanwhile the multimodal fusion system focuses on the specific methods of multimodal data fusion. This Recommendation specifies the requirements and capability framework to accommodate the existing and newly developed visual facilities for multimodal fusion for vision applications including autonomous vehicles, autonomous robots’ navigation and weather visibility prediction. This Recommendation defines the related requirements, capability framework, and reference use cases for the multimodal fusion system for vision.