A new approach to battery power tracking and predicting for mobile robot transportation using wavelet decomposition and ANFIS networks

Abstract

An intelligent system named Laboratory Mobile Robot Transportation System (LMRTS) has been developed for the mobile robotic transportation in laboratory automation. In this paper, a new approach is presented to predict and manage the on-board battery voltages of the mobile robots for optimizing the LMRTS system. The LMRTS can select and optimize the best mobile robotic candidate for a transportation task by considering those battery forecasting results. The proposed predictor includes three components: (a) Measuring the online voltages of the robotic on-board batteries; (b) Using the wavelet method to decompose the original measured data into a series of sub-layers; (c) Building the ANFIS for all the decomposed sub-layers and make the predictions; and (d) Integrating the forecasting results of the sub-layers to have the final predictions for the original online voltage signal. Two real experimental results show that the proposed hybrid predictor has both high forecasting accuracy and fast time performance, which can provide a powerful assistance to the real mobile robotic transportation.