Navigation Control Research For A High-Throughput Crop Phenotyping Mobile Robot With High-Clearance Configuration

Population growth and climate disasters lead to increasingly prominent food security issues.To address this crisis,many researchers have focused on breeding techniques to increase the yield of crops.At present,advanced molecular breeding technology needs to obtain high-quality and highyield crop samples,so it is particularly important to screen the required samples from a large number of crop varieties.However,the traditional artificial screening is inefficient and there is no uniform standard to measure the screening results,which cannot meet the breeding needs,so hinders the application of the new generation of molecular breeding.In contrast,the crop phenotyping based on robotics,especially mobile robot technology,enables information acquisition in a high-throughput,close-range,and high-precision manner,which becomes an interdisciplinary hotspot in the field of breeding and robotics.In this paper,we carry out the research on intelligent robot mobile platform for wheat crop field phenotyping.More specifically,our research focus on the autonomous navigation for wheat phenotyping robot.Firstly,according to the requirements of wheat field operations,the main configuration of phenotypic robot mobile platform is designed as a high-clearance structure to meet the needs of phenotypic monitoring tasks in the whole growth stage of wheat crops.The hardware integration of the robot system is completed,and the field test is implemented.In terms of hardware integration,the selection and deployment of key components are completed,including GNSS,INS,industrial computer,Arduino,brushless DC motors and motor drivers.In terms of software system,the ROS system is used as the software development platform to establish a robot software system framework,which is composed of sensing system,control system and human-machine interaction system.Secondly,this paper studies the data extracting solution for positioning sensors and the method for combined positioning.GNSS may have poor signal reception in the field environment at the edge of the farmland(the sensor signal is blocked by tall crops or canopy,which leads to the degradation of GNSS positioning accuracy).To solve this problem,this paper proposes a GNSS/INS combined positioning strategy based on adaptive Kalman filter.This strategy can fuse sensor information and deal with redundant data.In case of signal loss of GNSS,the robot positioning system can fuse the positioning information of GNSS and INS with Kalman filter,and further modify the Kalman filter coefficients through adaptive laws to change the confidence degree of each sensor,so as to improve the positioning accuracy.Then,this paper studies the robot trajectory tracking control and designs a human-machine interface.When the robot is moving forward,the robot will be disturbed by the environment and its own mechanical deviation,which lead to the decrease of trajectory tracking control accuracy.To solve this problem,this paper proposes a trajectory tracking strategy based on fuzzy PID control.The first step of this strategy is to establish the fuzzy control rules according to the robot control parameter adjustment experiences and prior knowledge.The second step is that the robot can adaptively adjust the control parameters in different states according to the control rules.On this basis,this paper also completes the design of human-machine interaction UI interface,which can show the current states of the robot,and send single-point navigation or multi-point cruise control instructions to the robot remotely.Finally,this paper completes the field experiments on the navigation control of the phenotyping robot,and tests the functions of remote control,speed adjustment,field positioning,trajectory tracking and human-machine interaction.The experimental results show that the robot performs well in various functions,and the robot can meet the basic needs of field work.