With the development of artificial intelligence, automatic control and multi-agent systems, unmanned systems are evolving from isolated operations to cross-domain collaboration, and full-space unmanned systems characterized by full-domain collaboration of "sea, land and air" are gradually taking shape. This paper focuses on typical scenarios in "sea, land and air", and systematically sorts out the research progress and integration trends in key technical fields such as positioning, intelligent perception, target prediction and behavior understanding, path planning and decision control, system modeling and simulation platform, etc., around core platforms such as unmanned surface vehicles (USVs), unmanned ground vehicles (UGVs) and unmanned aerial vehicles (UAVs). At the same time, this paper compares and analyzes mainstream simulation platforms and typical application scenarios, and puts forward common challenges between multiple systems in terms of equipment, perception accuracy, intelligent control and scheduling mechanisms, and points out that full-space unmanned systems will usher in key breakthroughs in the future in the fields of digital twin fusion, cross-domain collaborative decision-making and large model interfaces. This paper provides researchers with a set of technical frameworks and theoretical references covering perception, modeling, platform and application, which helps to build a pan-space unmanned system for urban intelligent governance and extreme environment operations.