Research On Trajectory Control Of Intelligent Lawn Mower

During the growth process of green lawn,it needs to be trimmed regularly.Pruning can effectively inhibit the heading,budding and seed bearing of lawn,block the growth of weeds,improve the ability of lawn grass to resist diseases and pests,and improve the texture of lawn.At present,most intelligent lawn mowers rely on the electromagnetic signal of the cable to identify the boundary line of the lawn and operate independently.Their operation control is basically random or a fixed way set by the program,which has the shortcomings of imperfect path planning function and poor ergodicity of mowing.To solve these problems,this paper designs a lawn intelligent mower walking trajectory control system based on deep learning.The system uses the information obtained by the visual sensor to detect the obstacles using the single-stage target detection algorithm Yolo X,and uses the semantic segmentation algorithm Deep Lab V3 + to judge the lawn boundary.The STC algorithm synthesizes the visual information to generate the best planning path.This paper mainly carries out relevant research and experiments from the following aspects:1.Taking a four-wheel differential by wire chassis module as the model machine of lawn mower,combined with the structure and motion control system of the four-wheel differential by wire chassis module,the navigation module is designed.On this basis,slam mapping and sports performance test simulating lawn environment were carried out;2.Based on the research of deep learning and semantic segmentation algorithm,the experiment of grassland boundary recognition by intelligent lawn mower is carried out;3.Based on the research of Yolo X algorithm,the research and experiment of grassland obstacle recognition are carried out;4.Combined with genetic algorithm and spanning tree covering algorithm(STC),the walking track test of intelligent lawn mower is simulated on the four-wheel differential wired chassis module,and the test results are summarized and analyzed.Based on the deep learning algorithm,the system can independently perceive the information of the external environment,dynamically plan its own walking trajectory,more accurately identify obstacles,and has high intelligence.Experiments show that the system can complete the task of mowing safely,reliably and low-carbon,and has good mowing ergodicity.