Cloud–edge collaborative AI for symmetry and asymmetry in human motion understanding and collaboration

Human motion understanding has become a cornerstone for advancing intelligent systems in domains such as human–computer interaction, robotics, medical rehabilitation, and sports analysis. Among the diverse characteristics of human motion, symmetry and asymmetry represent two fundamental yet complementary dimensions. Symmetry enables simplified modeling and improved generalization, while asymmetry captures individualized patterns and adaptive variations. With the rapid progress of artificial intelligence (AI) and the rise of cloud- and edge-computing infrastructure that supports large-scale training and low-latency deployment, novel algorithms and frameworks have been developed to perceive, model, and balance these two aspects, thereby enhancing both interpretability and collaboration. This paper provides a comprehensive review of recent AI-driven approaches for modeling symmetry and asymmetry in human motion. It emphasizes how cloud platforms facilitate large-scale model training and cross-scenario knowledge sharing, while edge computing enables real-time motion perception, intent prediction, and adaptive human–robot collaboration. It examines core methodologies, perception frameworks, algorithmic paradigms, datasets, and evaluation metrics. Furthermore, it highlights representative applications in action recognition, intent prediction, pathological movement analysis, and collaborative scenarios involving humans, robots, and multi-agent systems. Finally, the paper identifies key challenges and outlines promising directions for future research, aiming to provide systematic insights for building robust, adaptive, and human-centric AI systems for motion understanding and collaboration.