基于改进深度强化学习算法的农业机器人路径规划

赵威,张万枝,侯加林,侯瑞,李玉华,赵乐俊,程进

Path planning of agricultural robots based on improved deep reinforcement learning algorithm

Wei ZHAO,Wanzhi ZHANG,Jialin HOU,Rui HOU,Yuhua LI,Lejun ZHAO,Jin Cheng

表 3 不同路径规划方法在训练环境中的性能比较结果

Tab.3 Performance comparison results of different path planning methods in training environment

方法 N∈[1, 900] N∈[901, 1600] N∈[1601, 2000] Ctn tna/s pns/% Ctn tna/s pns/% Ctn tna/s pns/% 传统DQN 662 97.35 26.44 397 91.57 43.29 138 89.26 65.50 DDQN 613 98.64 31.89 308 89.38 56.00 114 83.59 71.50 Dueling DQN 596 99.37 33.78 303 88.45 56.71 103 82.31 74.25 本研究 409 96.72 54.56 97 83.25 86.14 0 76.77 100.00