基于改进蚁群算法的农业机器人多田块路径规划方法与试验

Multi field path planning method and experiment of agricultural robot based on improved ant colony algorithm

[目的]针对丘陵地区因田块碎片化和障碍物影响农业机器人作业的情况,提出一种基于改进蚁群算法和A算法相结合的多田块路径规划方法。[方法]通过无人机构建农田遥感影像,利用栅格法对农田进行环境建模,再进行子区的划分与合并,在蚁群算法中引入区域启发函数,对不同区域内的移动代价作区分,通过人工增加最优路径中的信息素浓度并建立自适应信息素挥发系数,对信息素更新方法进行改进,以此求解作业子区间的最优遍历顺序,利用具有启发式搜索功能的A算法进行子区连接路径规划,最终实现多田块路径规划。[结果]MATLAB仿真试验结果表明,在子区数量为40时,采用改进蚁群算法进行规划的平均路径长度比传统的蚁群算法减少了3.19%,平均迭代收敛次数减少了79.5%;在路径遍历仿真试验中,农业机器人遍历路径覆盖率能达到100%,路径重复率为6.48%。利用农田无人机遥感影像和田间作业参数进行自然环境的仿真试验,进一步验证了该方法的有效性。[结论]本研究提出的方法能有效解决丘陵地区农业机器人多田块路径规划问题,可为丘陵地区农业机器人大面积作业提供路径参考与技术支持。

[Objectives]In response to the impact of fragmented fields and obstacles on agricultural robot operations in hilly areas,this study aimed to propose a multi field block path planning method based on a combination of improved ant colony algorithm and A algorithm.[Methods]By using drones to construct remote sensing images of farmland,using grid method to model the environment of farmland,and then dividing and merging sub regions,a region heuristic function was introduced into the ant colony algorithm to distinguish the movement costs in different regions.The pheromone update method was improved by manually increasing the concentration of pheromones in the optimal path and establishing an adaptive pheromone volatilization coefficient,in order to solve the optimal traversal order of job sub intervals.The A algorithm with heuristic search function was used for sub region connection path planning,ultimately achieving multi field path planning.[Results]The MATLAB simulation test results showed that when the number of sub regions was 40,the average path length planned using the improved ant colony algorithm was reduced by 3.19%compared to the traditional ant colony algorithm,and the average iteration convergence number was reduced by 79.5%.In the path traversal simulation experiment,the coverage rate of the agricultural robot’s traversal path could reach 100%,and the path repetition rate was 6.48%.The effectiveness of this method was further verified through simulation experiments of natural environment using remote sensing images of agricultural drones and field operation parameters.[Conclusions]The method proposed in this study could effectively solve the multi field path planning problem of agricultural robots in hilly areas,and provided path reference and technical support for large-scale operation of agricultural robots in hilly areas.