基于DQN算法的无人机校园安全监控路径自动规划模型

Automatic planning model of UAV campus security monitoring path based on DQN algorithm

在校园安全管理中,传统的定期定点巡逻和摄像头监控难以覆盖整个校园,而无人机监控则可以弥补上述缺点。由于当前的无人机路径规划算法难以保证信息的时效性,因此,研究提出了基于深度Q网络的路径规划算法。实验结果显示,深度Q网络的成功率随着测试次数的增加而升高,最终稳定在0.79左右,高于基于信息年龄的轨迹规划算法和Q学习算法。同时深度Q网络规划的路径拐点数量仅为16个,覆盖率趋近于1,均优于其余算法。在自由空间和建筑密集空间中,深度Q网络的成功率最终分别稳定在0.99和0.86左右,平均步数均未超过100步。上述结果表明,基于深度Q网络的无人机路径规划算法能高效稳定地实现最优路径规划,实现对校园安全的无死角实时监控。

In campus security management,traditional regular fixed point patrols and camera monitoring are difficult to cover the entire campus,while drone monitoring can make up for the aforementioned shortcomings.Due to the difficulty in ensuring the timeliness of information in current unmanned aerial vehicle path planning algorithms,a path planning algorithm based on deep Q-network has been proposed.The experimental results show that the success rate of the deep Q-network increases with the increase of testing times,and ultimately stabilizes at around 0.79,which is higher than the trajectory planning algorithm and Q-learning algorithm based on information age.At the same time,the number of path inflection points in deep Q network planning is only 16,with an average information age of 19 seconds,which is lower than other algorithms.In free space and densely built space,the success rates of the deep Q-network ultimately stabilized at around 0.99 and 0.86,respectively,with an average number of steps not exceeding 100.The above results indicate that the unmanned aerial vehicle path planning algorithm based on deep Q-network can efficiently and stably achieve optimal path planning,and achieve real-time monitoring of campus security without dead spots.