融合蚁群算法和差分Transformer的农业机器人路径规划研究

打开文本图片集

中图分类号：S24；TP242 文献标识码：A 文章编号：2095-5553（2025）07-0164-09

Abstract：Toaddress thechallenges of low path planning acuracyand limitedobstacleavoidance capabilitiesof agricultural robotsoperating incomplex fieldenvironments，thisstudyproposesanovelpath planningmethod that combinestheAntColonyAlgorithm（ACA）withadiferentialTransformer.Initially，theACAisforglobalpathsearch， leveraging itsdistributedandparalelsearchcapabilities togenerateaninitial feasiblepath.Toovercomethetraditional ACA's limitations，such assusceptibilityto local optimization andpooradaptability to dynamic changesof the environment，a diferentialTransformer model was introduced toreplace theconventionalpheromone updating mechanism. Byutilizing aself-attention mechanism，the diferential Transformercaptures long-rangedependenciesand nonlinear featuresbetween path nodes，therebyalowing for more precise pheromoneupdatesand betteradaptabilityin complexconditions.Experimental results showed thattheproposed methodoutperforms traditionalalgorithmsin terms of pathlength，planning time，and obstacle avoidance success rate.Specificaly，inanenvironment with agrid sizeof 50， the average path length was reduced by 16.8% ，from 150 meters to 125 meters. Planning time was shortened by 23.5% ， from 2.13 seconds to 1.63 seconds. The obstacle avoidance success rate increased by 11.2% ，reaching 96.5% ： This research providesanefective solutionforautonomous navigation inagriculturalroboticsand holdssignificant theoretical and practical value.

Keywords：agricultural robot；path planning；ant colony algorithm；diffrential Transformer;smart agriculture

0 引言

随着全球人口持续增长和农业现代化加速，智能农业成为提升生产效率和保障粮食安全的关键手段[1]。（剩余14023字）