In agricultural plastic tunnel environments, the positioning accuracy of ultra-wideband (UWB) positioning systems is compromised due to variations in terrain elevation and ground surface irregularities, thereby affecting the navigation of picking robots and precise picking operations. A realtime correction and positioning method was proposed based on inertial measurement unit ( IMU) and UWB. Firstly, the grid map was corrected by using Rodrigues'rotation matrix and affine transformation. Subsequently, the UWB positioning data was preprocessed. By utilizing the rotation matrix obtained from the IMU that transformed the picking robot platform from its current attitude to the horizontal reference attitude, the lateral and longitudinal deviations from the UWB tag installation position to the actual position of the picking robot platform were calculated, thereby enabling real-time correction of UWB positioning errors caused by ground environmental conditions. Finally, autonomous navigation of the picking robot platform was achieved. The experimental results indicated that the maximum absolute error, maximum relative error, and root mean square error of the transformation-corrected grid map were 0.150 m, 2.27% , and 0.067 m, respectively. The real-time correction and positioning method based on IMU and UWB achieved lateral and longitudinal positioning deviations of less than or equal to 0.065 m and 0.069 m, respectively. Compared with the original UWB positioning data, the lateral positioning accuracy was improved by 26. 1% and the longitudinal positioning accuracy was improved by 13.8% . When the picking robot platform employing the aforementioned positioning method operated at autonomous navigation speeds of 0.15 m/ s, 0.30 m/ s, and 0.45 m/ s, the mean values of the lateral navigation deviation, longitudinal navigation deviation, and heading deviation were less than or equal to 0.057 m, 0.079 m, and 8.211°, respectively, with standard deviations less than or equal to 0.039 m, 0.069 m, and 4.307°, respectively.