针对高校多智能体博弈课程教学中存在的强化学习入门难、学生AI工程经验不足以及教学与行业需求脱节问题，分析传统教学模式的局限性，提出“兴趣驱动、场景教学、迭代开发、竞赛考核”的教学改革思路，借助腾讯开悟平台，从内容、方法和考核3个方面介绍创新改革实践，最后说明教学改革效果。

Aiming at the problems in the teaching of multi-agent game courses in colleges and universities,such as difficulties in getting started with reinforcement learning,insufficient AI engineering experience of students,and disconnection between teaching and industry needs,this paper analyzes the limitations of traditional teaching modes.It proposes a teaching reform idea of “interest-driven,scenario-based teaching,iterative development,and competition-based assessment”.With the help of Tencent Kaiwu Platform,the paper introduces innovative reform practices from three aspects:content,methods,and assessment.Finally,it explains the effects of the teaching reform.