This work proposes an obstacle motion prediction technique for sensor-based Task-Constrained Motion Planning (TCMP) problems for redundant manipulators in workspaces populated by obstacles. In particular, an accurate methodology based on Kalman filtering for estimating the movement of the obstacle was suggested, and a new extended MPC-based motion planner that generates robot motion commands according to a prediction of the obstacle motion over a specific planning horizon was formulated.
The proposed extended approach and the accuracy of the estimates were then validated through simulations, comparisons with the non-predicting approach, and experiments on the Universal Robots UR10 manipulator.
Tools: V-REP, Universal Robots UR10 manipulator, C++, TCMP (Task-Constrained Motion Planning), MPC (Model Predictive Control), ACADO Toolkit, Kalman filtering, Microsoft Kinect v2.0, CUDA Framework, MATLAB, Simulink
Design of a satellite mathematical model (based on reaction wheels) and simulation of an attitude control system in a quaternion formulation through the Direct Parametric Approach.
Tools: MATLAB, Simulink
Implementation and simulation of different nonlinear controllers for trajectory tracking, cartesian and posture regulation using GPS and Runge-Kutta localization methods.
Click HERE for slides.
Tools: MATLAB, Simulink, Trajectory Tracking Control, Regulation Control
Realization of a 7-DOF multi-body longitudinal dynamic model of a sport-bike using Lagrangian Mechanics (the model includes normal, longitudinal, rolling and drag forces, suspension systems and tire-road interaction using Pacejka Formula).
An active aerodynamics system based on winglets with variable angle of attack was implemented (the drag- and down-forces generated were retrieved by interpolating NACA airfoils database data). Two controllers were implemented in order to maximize the generated downforce (real-time optimal control) or limit the pitching motion during acceleration. Simulations and performance analysis were performed to validate the approach.
Tools: MATLAB, Simulink, Lagrangian Mechanics, Mathematical Modeling (Dynamics, Aerodynamic forces, Pacejka formula), Optimal and PID Control
Optimal control system based on the Linear Quadratic Control (LQC) problem theory for insulin delivery in a type I diabetic patient.
Click HERE for slides.
Tools: MATLAB, Simulink, LQC (Linear Quadratic Control)
Design and simulation of different models using Machine Learning techniques to predict the survival rate of a person to the Titanic disaster given its specific features.
Tools: Python (+ Scikit-Learn), Machine Learning
State of the art on FDIA in Smart Grids. Impact on power system, construction with Linear DC or nonlinear AC power flow model, full Information or partial Information, defense strategies.
Tools: Research and study of several papers about FDIA: Impact, Construction techniques (Linear Power Flow Model – Nonlinear Power Flow Model – Full information – Partial information), Defense strategies
Click HERE for slides.
Tools: MATLAB, Simulink, Multi-Robot Control
Tools: MATLAB, Simulink, Extended Kalman Filtering
A simple rotary inverted pendulum with swing-up and stabilization controller (LQR, SMC, Energy-based swing-up)
Click HERE for mathematical details, schematics, code and video [GitHub]
Tools: MATLAB, Simulink, Arduino Mega 2560, Lagrangian Mechanics, Kalman filtering, LQR (Linear Quadratic Regulator), SMC (Sliding Mode Control)
Actively controlled (thrust vectoring) model rocket
Click HERE for the flight board code [GitHub]
Tools: Arduino, IMU, Madgwick filtering, Kalman filtering, PID control