Professor of Computational Intelligence and Intelligent Systems .
Biography: Dr. Benyamin Kusumoputro is Professor in Computational Intelligence and Intelligent Systems since 2004. He holds a Bachelor degree in Physics from Bandung Institute of Technology, Indonesia, and Magister of Engineering from Universitas Indonesia in 1981 and 1984, respectively. Dr. Kusumoputro received his Dr. Eng. degree from Dept. of Electrical and Electronic Engineering, Tokyo Institute of Technology in 1993. He has served as Principal Investigator in various multi years reseach grants from 1995, and spent a year as a Visiting Professor at KAIST, Korea during 2006-2007 under KFAS Foundation. He also serves numerous times as a Visiting Scholars in Tokyo Institute of Technology, and Nagoya University in Japan.
Dr. Kusumoputro has received various awards as a recognition for excellence in research from Universitas Indonesia in 2002, Ministry of Research and Technology of Indonesia in 2006, and the University Award from University of Indonesia in 2016. Dr. Kusumoputro has supervised more than 12 Ph.D dissertations with 6 candidates in progress. His research interests include the development of 3D face recognition using Hemispherical Structure of Hidden Layer Neural Networks, odor recognition system, and recently, the development of autonomous control system of unmanned vehicle systems. He published more than 70 articles in academic and professional journals and serves as a frequent Invited Speaker at various academic conferences and professional meetings.
Autonomous control system for a trajectory flight dynamics of an unmanned aerial vehicle systems using artificial neural networks
Unmanned Aerial Vehicle (UAV) system is an unmanned aircraft that can be categorized into a fixed-wing UAV and a rotor wing UAV. UAV system is firstly developed to perform various dangerous military purposes of aerial missions due to its simplicity in the flying process, lower operating cost and the low safety risk to the human operator. Recently, the development of the UAV system is also designed for its utilization in many civil applications such as for land and forestry mapping, agriculture purposes, and for search and rescue (SAR) tasks. UAV system is also capable of carrying out a high-risk task, such as in the areas exposed to a nuclear radiation or a disaster area that is difficult to be reached. UAV system is a complex dynamic flying system with the characteristics of multiple input multiple output (MIMO), under actuated, nonlinear, highly coupled, time-varying and inherently unstable and varies widely across its full flight condition. The control system of a UAV could be developed using a linear control method or a nonlinear control methods, where any developed control methods required an adequate model of flight dynamics for the design, verification and analysis of the system. The UAV flight dynamics model are widely developed mathematically, derived from its physical based model to the non-parametric modeling technique, and also from the complete and complex nonlinear model to a more simplified linear model. However, as the UAV system is a very dynamic system with the ability to move at a six degree of freedom that might be prone to various disturbance within a very fast change of flight conditions, there are many problems still remain a challenge for researchers to develop a reliable robust controller system for an autonomous UAV movements. In this presentation, a neural networks control system has been developed and implemented as an autonomous trajectory flight control of a UAV system. Instead of using mathematically derived adaptive controller system, an artificial neural networks algorithm is used as the adaptive controller system for the uncertain flight dynamics and conditions of the UAV system, reducing the need for tuning offline of the usually utilized PID controller system. The artificial neural networks based autonomous controller system is developed to be able to identify and control the UAV flight dynamics, by modeling the non-linear systems and the DIC control techniqure using multi-layer neural network approximation.