【摘要】 自主式微直升机为了获得完成任务的行为 ,可以通过在任务环境中飞行来学习控制规则。由于增强式学习不需要精确的环境模型 ,而是采用逐次逼近的机理 ,所以微直升机需要很大的计算开销 ,难以满足实时性要求。另外由于微直升机尺寸的限制 ,不能安装功能很强的传感器来获得完全的环境信息 ,所以微直升机必须通过其他智能体协作来获得环境信息。本文利用高档台式机强大的计算和扩展功能 ,将其作为一个协作智能体与微直升机协作来完成增强式学习 ,仿真实验结果和理论分析证明这种方法的有效性 ,最后 ,给出我们今后的研究重点更多还原

【Abstract】 The autonomous micro helicopter should fly in the environment to study the behaviors accomplishing the given task.However,reinforcement learning will take much time using the trial and error mechanism without the certain environment,and not satisfy the real-time request.For the oversize of the micro-helicopter,it could not equip strong sensors to get the whole environment information.The micro-helicopter should cooperate with other agent to get the environment information.The powerful ability of slap-up computer to cooperate with the micro-helicopter to accomplish enforcement learning is used .The results of simulation prove the validity of this cooperation method.At last,the future study is given.更多还原