This master’s thesis addresses the prediction of battery voltage drop of the Leo Rover mobile robot during driving. The aim of the thesis is to develop a data-driven and interpretable model that estimates the cumulative voltage drop based on motion features and the initial voltage of the drive. The data were obtained from ROS (Robot Operating System) measurement records (ROS bag files), time-aligned, and divided into segments according to the travelled distance. For individual segments, features related to linear and angular velocity, positive elevation gain, and travelled distance were calculated, while the model used their cumulative values from the beginning of the drive to the considered segment. The model was evaluated for different segment lengths and additionally assessed in terms of minimum voltage prediction and classification of drives according to the operational limit of 10.8 V. The results show that the proposed cumulative linear model enables the estimation of the minimum voltage during driving and can serve as additional information for assessing whether the voltage is approaching the operational limit.
|
