Research
I am a PhD candidate under Prof. Steven Waslander supervision with the MME department of the University of Waterloo. I joined the WAVELab in January 2012. My main area of research is Modelling dynamical systems using data-driven approaches. Particularly, I am interested in the research area of Machine Learning and Computational Intelligence in Modelling and Control of mobile robots. My main goal is to devise practical methods to improve control performance over fast, complex and inherently unstable dynamical systems, such as a quadrotor, by learning the system unmodelled dynamics through applying ML methods on the observed behaviour of the dynamical system. Currently, I am working on implementing a new Recurrent Neural Network architecture, a multilayer deep RNN with global feedback and feedforward connections (the so-called MOdular DEep RNN – MODERNN framework) in parallel platformsI have already successfully employed the method in learning a full dynamic of a simulated quadrotor purely from data. The improved model will be used in a Model Predictive Control scheme.
Biography
Before joining the WAVELab, I acquired my Master of Computer Science from Örebro University, Centre for Applied Autonomous Sensor Systems (AASS), Örebro, Sweden, in December 2011 (thesis). Before that, I obtained a Master of Science in Control Engineering from Amirkabir University, Tehran, Iran. I obtained my Bachelor of Science in Electrical Engineering from Shahid Beheshti University, Tehran, Iran.
Selected Publications
- State initialization for recurrent neural network modeling of time-series data, Nima Mohajerin and Steven Waslander, International Joint Conference on Neural Networks (IJCNN), Anchorage, AK, 2017, pp. 2330-2337.
- Modelling a Quadrotor Vehicle Using a Modular Deep Recurrent Neural Network, Nima Mohajerin, Steven Waslander, IEEE International Conference on Systems, Man and Cybernetics, Hong Kong, 2015 .
- Modular Deep Recurrent Neural Network: Application to Quadrotors, Nima Mohajerin, Steven Waslander, 2014 IEEE International Conference on Systems, Man, and Cybernetics, San Diego, CA, USA.
- Feature Extraction and Radar Track Classification for Detecting UAVs in Civillian Airspace, Nima Mohajerin, Jonathan Histon, Reza Dizaji, Steven Waslander, 2014 IEEE Radar Conference (RadarCon 2014), Cincinnati, OH, USA.
- Identification and Predictive Control Using Recurrent Neural Network, Master of Science Thesis in Computer Science-Robotics and Intelligent Systems, School of Science and Technology, Örebro Universitet, Sweden, October 2011.
- A Reinforcement Learning Fuzzy Controller for the Ball and Plate System, Nima Mohajerin, Mohammad Bagher Menhadj, Ali Doustmohammadi, FUZZ-IEEE 2010 proceedings, Barcelona, Spain.
- Stabilizing Ball and Plate System with the aid of Intelligent Controller, Design and Implementation, Master of Science Thesis in Control and System Engineering, Department of Electrical Engineering , Amirkabir University of Technology, Iran, August 2007. (In Persian)
Contact
Address:
Nima Mohajerin
Department of Mechanical and Mechatronics Engineeering
University of Waterloo
Waterloo, ON, Canada
N2L 3G1
e-mail:
nima[dot]mohajerin[at]uwaterloo[dot]ca
office:
E5-3012
phone:
(519)888-4567×31404