现代农业装备

为推动名优茶叶采摘自动化，茶叶采摘机械臂快速、高质量路径规划是实现高效采摘的关键。针对传统群智能优化算法在茶园复杂环境及约束条件下存在的路径质量差、算法耗时长及规划不稳定等问题。提出一种改进豪猪优化器(Crested Porcupine Optimizer，CPO)的机械臂路径规划方法。通过引入动态种群收缩策略，在迭代过程中缩减种群规模，减少计算成本，使用末位淘汰机制及对算法结构改良提升全局寻优能力，增加个体多样性，并引入动态调整因子λ_t改进第一防御策略，平衡算法在不同阶段的探索与优化比例。通过Lindenmayer系统及UR5机械臂构建茶叶采摘仿真场景，进行仿真路径规划实验。在10个不同环境中，改进CPO算法相比原算法，平均计算时间减少4.7%，平均路径长度缩短0.78%；与灰狼优化(Grey Wolf Optimizer，GWO)、蜣螂优化(Dung Beetle Optimizer，DBO)、快速扩展随机树(Rapidly-exploring Random Trees，RRT)等算法相比，平均耗时相较GWO、DBO分别下降25%、24%，路径长度相较RRT算法减少23%、平均规划成功率高28%。改进CPO算法相较其他算法耗时更短，同时具有更好的路径质量及规划成功率，验证了其在茶叶采摘机械臂路径规划问题上的实用价值。

To promote the automation of famous tea picking,fast and high-quality path planning of tea picking robotic arms is the key to achieving efficient picking.Traditional swarm intelligence optimization algorithms face challenges in complex environments and constrained conditions,including poor path quality,long algorithm time,and unstable planning.An improved Crested Porcupine Optimizer (CPO) robotic arm path planning method is proposed.By introducing a dynamic population contraction strategy,the population size is reduced during the iterative process,reducing computational costs.Use the last-out mechanism and improve the algorithm structure to enhance the global optimization ability and increase individual diversity.And introduce a dynamic adjustment factor to improve the first defense strategy and balance the exploration and optimization ratio of the algorithm in different stages.Using the Lindenmayer system (L-System) and the UR5 robotic arm,a tea picking simulation scenario was constructed to conduct a simulation path planning experiment.In 10 different environments,the improved CPO algorithm reduces the average computation time by 4.7% and shortens the average path length by 0.78% compared to the original algorithm.Compared with Grey Wolf Optimizer (GWO),Dung Beetle Optimizer (DBO),Rapidly-exploring Random Trees (RRT) and other algorithms,the average time consumption is reduced by 25% and 24% respectively compared with GWO and DBO,the path length is reduced by 23% compared with RRT algorithm,and the average planning success rate is increased by 28%.Compared with other algorithms,the improved CPO algorithm is more time-efficient,with better path quality and planning success rate.The practical value of the method in path planning for tea picking robotic arms has been verified.