Research And Design Of A Robotic System Based On Agricultural Internet Of Things

The agricultural Internet of Things(IoT)has brought modern high-tech to agricultural production,making agricultural production more automated and intelligent.Furthermore it has been widely studied and applied in recent year.In the traditional agricultural IoT a large number of sensor nodes are needed to form a large-scale wireless sensor network for the agricultural environment monitoring.However,it is difficult to monitor the growth of the crops in all-around using this mode.By means of the agricultural robots which can inspect and monitor the agricultural environment,the growth status of crops can be monitored in a high efficiency way.Firstly,in order to realize the autonomous inspection and monitoring in the greenhouse,the position of the robot in the greenhouse needs to be clarified.Aiming at the greenhouse environment,a UWB/IMU combined positioning method based on Kalman filter fusion is proposed,and the outliers in UWB positioning signals are detected before fusion.Using this method,only the IMU update is performed if there are outliers.Otherwise,Kalman filter fusion is performed.The simulation results verify the effectiveness of the proposed positioning method.Secondly,the surface of the working track is not flat,which will make the robot slip and shake to deviate from the scheduled track during inspection.To address this problem,this paper simplifies the complex path tracking problem to a simple angular velocity tracking problem by constructing an ideal yaw angular velocity function.When the actual yaw angular velocity tends to the ideal yaw angular velocity,the lateral deviation and the heading deviation of the function converge to zero.Active Disturbance Rejection Control controller is applied to track angular velocity for the system,as well the linear expansion observer is adopted to estimate the system input and unknown disturbance.Simulation results indicate that the controller proposed has great tracking performance and disturbance suppression capability.Finally,as the greenhouse is a semi-outdoor environment,the agricultural environmental data collected by the sensors will be seriously polluted by the noise,which will affect the analysis results of the data in the user end.Therefore,a time series model based on RBF neural network is proposed to preprocess the agricultural raw data.The experimental comparison shows that,the time series model has more accurate description on agricultural data.