Energy Prediction of Cleanroom-type Differential Drive Mobile Robot Based on Recurrent Neural Network

摘要

The battery charger time is a major issue for mobile robots.The study of the power usage of each component is important for optimizing the overall power consumption.Additionally, knowing the total energy consumption before commanding a robot to execute a task is essential for effective queue management and determining which robots are ready to execute tasks or move to the charging station.In this paper, we propose an energy modeling system consisting of an energy sensing technique, logging, and a recurrent neural network prediction model.The model is configured to recognize the dynamic system of the drive unit with the support of the robot operating system.The proposed model has a prediction error of only 3.58%.The simulation and experimental results demonstrate the effectiveness of the proposed system.