Abstract: Facing the development demands of the low-altitude economy, the large-scale application of low-altitude aircraft in scenarios such as logistics distribution, urban supervision, emergency rescue, and low-altitude operations places higher requirements on the perception, communication, computing, and intelligence of low-altitude networks. Addressing the issues of insufficient coverage, fragile links, low coordination efficiency, and limited edge resources in traditional terrestrial networks within three-dimensional high-dynamic airspace, this paper reviews the low-altitude collaborative intelligence architecture and key technologies for 6G space-air-ground integrated networks. First, it explains the operation mechanism of the dual end-cloud adaptive architecture and its parallel coupling of task flows and knowledge flows; next, it systematically reviews the research progress on key technologies such as cross-domain collaborative perception, heterogeneous dynamic networking, collaborative computing and distributed decision-making, and collaborative learning; further, by combining typical scenarios such as surveillance of unauthorized low-altitude drones, low-altitude network security coverage and resilient information assurance, and low-altitude network operations, it summarizes the application requirements and common challenges of low-altitude collaborative intelligence. The analysis indicates that the key issues faced by low-altitude collaborative intelligence mainly concentrate on perception trustworthiness, connection continuity, decision stability, and continuous learning.

Key words: Low-altitude economy, low-altitude intelligent network, unmanned aerial vehicles, collaborative intelligence, dual adaptation of edge and cloud.