Implementation of Q-Learning Algorithm on Arduino : Application to Autonomous Mobile Robot Navigation in COVID-19 Field Hospitals

ABSTRACT

The recourse to Mobile Robots (MRs) in fighting against Coronavirus pandemic (COVID-19) has today become a necessity in almost all hospitals worldwide. Indeed, for example, the Ultraviolet Disinfection (UVD) robot has been very useful, since COVID-19 pandemic began, to destroy viruses in Wuhan hospitals. However, MRs are equipped with a locomotion system, which should be capable of navigating through, generally, an unknown work environment. In such situation, MR should also have learning mechanism. In this paper, we propose a low-cost solution to deal with the problem of autonomous navigation and the trajectory optimization for MR in Field Hospitals (FHs). The architecture of these latter may be varying from one to another according to where it should be installed. Thus, a rapidly adapting to the new trajectory is then necessary to help save lives in pandemic situation. So, to conduct a successful autonomous navigation of MR particularly in COVID-19 FHs, a practical low-cost solution must then be found. To do so, a MR equipped with TCS230 Color Sensor, is used to read colored sticky notes fixed on the ground, to identify room number of the FH that serves for learning and correcting robot trajectory keypoints. Furthermore, an Android application is also developed in this work to remotely control, via a Bluetooth wireless connection, an Arduino-based MR in some specific situation. Some experiments were carried out to verify the performance of the proposed autonomous navigation approach of MR type ELEGOO Smart Robot Car V3.0. Moreover, the simulation results have confirmed that the robot can reach the goal with optimal trajectory by only exploiting the popular Q-Learning algorithm with the use of a low-cost color sensor, and sticky notes which could be easily deployed according to the architecture of the new installed FH.