Swarm intelligence, as an emerging field, has become the hot spot and frontier field of artificial intelligence as well as economic, social, biological and other interdisciplinary fields. Swarm intelligence refers to the decentralized self-organizing behaviors expressed at the collective level. Social insects or animals follow simple rules of behavior and show advanced swarm intelligence at the collective level. For instance, the complex social systems of ants and bees, and the migration of flocks of birds and fish to adapt to the air or sea. Swarm intelligence is not simply a collection of multiple individuals, but a higher performance beyond individual behavior, with more robustness, flexibility and economic advantages. The evolution from individual behavior to swarm behavior is often too complex to predict. We are interest in the combination of machine learning with complex self-organization system study. We construct multimodal sensor system to collect the data of individual and swarm behavior that regulated by multimodal (including images, sound, chemical, and physical contact, etc.) communication mechanisms, employ machine learning and computer simulation, to explore how the individual multimodal communication to achieve self-organization and emerging synergy, group decision-making and other advanced features.