Gizem Doğa Filiz

Vision-Augmented Autonomous Surface Cleaning with Force-Impedance Control

Gizem Doğa Filiz, supervised by Prof. Sami Haddadin & Dr. Kübra Karacan

Munich Institute of Robotics and Machine Intelligence (MIRMI), TUM | May - July 2025

Developed a vision-augmented force-impedance control framework for autonomous surface cleaning using Franka Emika Panda robot and Intel RealSense D455 camera. Implemented Markov Decision Process (MDP) planner with Bayesian belief updating for intelligent surface coverage motion planning and real-time stain detection on unknown 3D surfaces.

EMG-Based Control of Hand Exoskeleton with Machine Learning

Gizem Doğa Filiz, supervised by Asst. Prof. Fanny Ficuciello & Asst. Prof. Mohammed Gohari

PRISMA Lab, University of Naples Federico II | June - September 2024

Designed, 3D printed, and assembled a hand exoskeleton with a four-bar linkage system using SolidWorks. Developed hardware for EMG-based control and implemented machine learning algorithms to classify hand states through EMG data. Integrated the algorithms for real-time actuation of the hand exoskeleton.

Bayesian Optimization for Sequential Object Placement and Rearrangement

Gizem Doğa Filiz, supervised by Prof. Volkan Patoğlu & Prof. Esra Erdem

Sabancı University, Graduation Project | 2024 - 2025

Developing an algorithm using Bayesian Optimization with transfer learning for sequential object placement and rearrangement with collision-free arrangements on cluttered surfaces. The system enables efficient planning for robotic manipulation tasks in complex environments.

Advanced Cart-Pole Swing-Up Control with JAX and MuJoCo

Sabancı University, Deep Learning for Robot Control (ME 58006) | Fall 2024

Developed a data-driven system identification framework for the Cart-Pole system, utilizing JAX, Diffrax, and Optax to estimate system parameters and analyze motion trajectories. Implemented a neural network-based controller to perform Cart-Pole swing-up and stabilization, optimizing trajectory costs and comparing performance with LQR control. Engineered a Mixture-of-Experts controller, integrating state-dependent switching for autonomous robotic motion control, improving stability.

Implementation and Analysis of Adaptive Switch Image-Based Visual Servoing for Industrial Robots

Sabancı University, Vision Based Control (EE 529) | Fall 2024

Implemented and validated the Adaptive Switch IBVS controller from Ghasemi et al. (2020) using MATLAB simulation with UR5 robot model. Developed a three-stage control strategy that switches between pure rotation, pure translation, and full IBVS based on rotation angle thresholds. Achieved 27-34% faster convergence compared to traditional IBVS, and successfully handled extreme orientation mismatches (200°+) where traditional methods failed. Comparative analysis demonstrated smooth feature trajectories and robust performance across multiple test scenarios.

Kinematic and Dynamic Analysis of the 3RRP Mechanism

Sabancı University, Kinematics and Dynamics of Machines (EE 521) | Fall 2024

Developed a MATLAB-based dynamic simulation for a 3RRP Mechanism and a Linear Delta Mechanism, implementing forward/inverse kinematics, Jacobian-based motion analysis, and Baumgarte-stabilized integration for constraint handling. Formulated and derived equations of motion using Kane's Method and Lagrange's Equations, analyzing workspace and singular configurations.

Design and Control of a Planar Elbow Manipulator

Sabancı University, Introduction to Robotics (ME 403) | Spring 2024

Engineered a 3-DoF planar elbow manipulator with optimized kinematic analysis. Implemented comprehensive forward and inverse kinematics solutions. Developed dynamic models using MATLAB/Simulink for simulation and control analysis. Designed robust PID and computed torque controllers for enhanced trajectory following and improved end-effector positioning accuracy.