多传感器融合的温室机器人自主导航系统设计

Design of autonomous navigation system for greenhouse robot based on multi sensor fusion

目的:针对机器人在温室环境下存在扫描盲区、里程计定位误差大等问题,提出一种多传感器融合的温室机器人自主导航系统。方法:通过ROS系统,使用前后激光雷达、里程计、IMU,采用优化参数配置的Cartographer算法进行融合定位和建图;将Cartographer定位信息加入Navigation中优化导航框架,结合Dijkstra算法与DWA算法完成路径规划,实现机器人自主导航。结果:相较于单一里程计,融合IMU后的定位精度更高;Dijkstra算法相较A*算法更符合运行需求;导航系统在仿真与实际环境下自主导航验证满足温室环境下运行需求。结论:导航系统采用多种传感器融合,可以解决机器人在温室环境下存在扫描盲区、里程计定位误差大等问题,该研究可为同类研究提供参考。

Objective:To address the issues of blind spots in scanning and large odometer positioning errors of robots in greenhouse environments,a multi-sensor fusion autonomous navigation system for greenhouse robots was proposed.Methods:Through the ROS system,using front and rear LiDAR,odometer,IMU,and using the optimized parameter configuration,Cartographer algorithm for fusion were positioned and mapped.Integrating cartographer positioning information into navigation to optimize the navigation framework,and combining Dijkstra algorithm and DWA algorithm to complete path planning,achieving autonomous robot navigation.Results:Compared to a single odometer,the positioning accuracy after integrating IMU was higher.Dijkstra algorithm was more in line with operational requirements compared to A*algorithm.The autonomous navigation verification of the navigation system in both simulation and actual environments met the operational requirements of the greenhouse.Conclusion:The navigation system adopted multiple sensor fusion,which could solve the problems of robot scanning blind spots and large odometer positioning errors in greenhouse environments.This study can provide reference for similar research.