International Journal of Robotics and Automation (IJRA) Vol.8, No.3, September 2019, pp. 189~193 ISSN: 2089-4856, DOI: 10.11591/ijra.v8i3.pp189-193  189 Journal homepage: http://iaescore.com/journals/index.php/IJRA Artificial pancreas techniques based on robust model predictive controller Waleed Khalid Al-Azzawi Faculty of Information Technology, Arab Open University Bahrain, Iraq Article Info ABSTRACT Article history: Received May 3, 2019 Revised Jul 5, 2019 Accepted Aug 1, 2019 Diabetes is known as the major cause of death in the world leading to kidney, retinopathy and cardiovascular diseases as well. In this paper, a Robust Model Predictive Controller (RMPC) is introduced to design artificial pancreas that solved the model uncertainty and keep the blood glucose level in the normal range by regulating the size of insulin infusion from pump based on RMPC. The simulation results will present a good performance of the proposed controller to avoid disturbance and robustness against uncertainties. Keywords: Artificial pancreas MPC techniques Robust model predictive controller (RMPC) Copyright © 2019 Institute of Advanced Engineering and Science. All rights reserved. Corresponding Author: Waleed Khalid Al-Azzawi, Faculty of Information Technology, Arab Open University Bahrain, Iraq. Email: waleedki@yahoo.com 1. INTRODUCTION Type 1 diabetes is an inveterate disease in which the pancreas does not produce sufficient insulin to control glucose levels in the blood. In the shortage of insulin, the admittance of glucose into skeletal, cardiac, smooth muscle and other tissues is reduced. When insulin is missing for a longer period of time, the muscle and tissue cells will start using fat as energy source, in place of glucose from the blood stream. The disturbance of meal eating presents great challenges to automatic blood glucose control. There are many factors must be measured to regulate blood glucose such as. Fats and proteins cause delays in absorption of glucose from carbohydrates eaten at the same time, physical exercise affects the blood glucose regulation by reducing the requirement for insulin, and furthermore, stress and physical illness are involved in the blood glucose regulation process. Thus, designing accurate predictive models are hard to achieve. In 2011, Dimitri Boiroux et al, applied a robust feedforward-feedback control strategy to people with type 1 diabetes. The feed forward controller consists of a bolus calculator which compensates the disturbance coming from meals. The feedback controller is based on a linearized description of the model describing the patient [1]. In 2008, Gianni Marchetti et al concerned with the improvement of new feed forward-feedback control techniques for actual glucose control and type 1diabetes. Enhanced post-meal responses can be achieved by a pre-prandial snack or bolus, or by reducing the glucose set-point prior to the meal [2]. In 2017, Mojgan Esna-Ashari et al proposed a technique for blood glucose level regulation in type I diabetes. The control approach is based on non-linear model predictive control. The purpose of the proposed controller optimized with genetic algorithms is to measure blood glucose level every time and estimate it for the next time interval [3]. In 2017, Lukas Ortmann et al, introduced a Model Predictive Controller that takes the periodic Insulin sensitivity into account, so as to improve blood glucose control. The upcoming effect of the Insulin sensitivity is estimated by a machine learning technique, namely, a customized Gaussian Process (GP), based on historical training data [4]. In 2017, Chiara Toffanin, proposed an adaptive Model Predictive