Research On Localization Of Greenhouse Robot Based On Multi-sensor Fusion

Conventional mobile robots mainly employ LIDAR for indoor global positioning and navigation,thus having strict requirements for the ground environment.Under the complex ground conditions in the greenhouse,the accumulative errors of odometer(ODOM)that arise from wheel slip are easy to occur during the long-time operation of the robot,which decreases the accuracy of robot positioning and mapping.To solve this problem,a multi-sensor fusion positioning method integrating ultra-wideband(UWB),inertial measurement unit(IMU),ODOM and LIDAR is proposed.The main research contents are as follows:(1)According to the greenhouse environment,the system structure of the positioning test platform is designed,including the chassis of the positioning test platform,the hardware and software of the control system.In the chassis,four-wheel drive and differential steering are adopted,and the respective wheel is equipped with a shock absorber,which is stable and has good trafficability.The hardware and software system are designed based on ROS system.The lower computer is used to achieve the motion control and speed feedback of the positioning test platform,and the upper computer runs the positioning system software and applies the positioning and navigation algorithm.(2)By comparing and analyzing the working performance of commonly used indoor positioning sensors in greenhouse,a multi-sensor fusion positioning method based on UWB/IMU/ODOM/LIDAR is proposed.Firstly,the appropriate navigation coordinate system is established,which is used to analyze the kinematics model of the positioning test platform,and the mathematical model of positioning sensor and map is constructed.On that basis,a multi-sensor fusion positioning framework is proposed,and a multi-sensor fusion positioning system is built in the actual greenhouse environment.(3)Combined with the characteristics of greenhouse environment,this paper analyzes the SLAM algorithm often used for indoor mapping,and selects the cartographer algorithm to build the greenhouse environment map.Firstly,UWB,IMU and ODOM is integrated by the Extended Kalman Filter(EKF)algorithm to obtain the estimated positioning information.Second,LIDAR is integrated with the established two-dimensional(2D)map by the Adaptive Monte Carlo Localization(AMCL)algorithm to achieve the global positioning of the robot.(4)The positioning system in this paper is tested.The comparison test results of greenhouse mapping and positioning accuracy demonstrate that the positioning system in this paper effectively reduces the accumulative error of greenhouse robot positioning and mapping.The maximum lateral deviation of the positioning system in this paper is less than 0.1m,and the root mean square error of the lateral deviation is reduced by 33.3% and 67% compared with UWB positioning and the conventional IMU/ODOM/LIDAR fusion positioning methods.The positioning test results of target points demonstrate that the root mean square error of the positioning system in x-axis direction,y-axis direction and overall positioning are 0.092,0.069 and 0.079 m,respectively,and the maximum positioning error is 0.102 m,thus satisfying the positioning and navigation accuracy requirements of robot operation in greenhouse environments.