Abstract: Task planning serves as the core of command and control, facing challenges such as huge state-action space, high reliability requirements and dynamic application scenarios, for which no effective solutions currently exists. Based on the analysis of the current state of research both domestically and internationally in areas including hybrid intelligence, language models and scheme evaluation, a task planning framework based on hybrid intelligence is proposed. This paper further conducts in-depth discussion on key technologies involved in the framework, including hierarchical planning of tasks and actions, representation and learning of domain knowledge, and evaluation and optimization of task and action sequences. By adopting the third-generation artificial intelligence technology driven by the integration of knowledge and data, a new generation of task planning methods is explored to solve difficulties such as low planning timeliness, poor result interpretability, and weak scenario generalization ability. The proposed method can give full play to the advantages of command decision-makers and intelligent algorithms, facilitating the exploration of new modes of task planning applications in human-machine integration scenarios, and providing theoretical basis and methodological guidance for the architecture and key technological breakthroughs of task planning systems.