Projects

• Developed a lane-keeping control module for a latency-aware teleoperation system, enabling stable vehicle behavior under delayed perception conditions.

• Implemented and evaluated a lateral control module based on Pure Pursuit within a closed-loop simulation pipeline, integrating perception and control.

• Conducted performance evaluation under varying latency conditions, analyzing trajectory tracking accuracy and system stability.

• Demonstrated real-time system behavior through integrated visualization and telemetry

Two major tasks were accomplished during the thesis project. First, a camera-LiDAR calibration method was selected and further developed, making it quick and convenient to perform during tests (Using RTMaps and Python). Second, three fusion architectures were investigated in simulations (Using MATLAB), and their comparative advantages were considered in relation to the requirements of the ongoing department project.

Two models were built:

• Using Matlab/Simulink, a Simscape model was built for a vehicle where you can switch between two transmission systems: Dual-clutch 7-speed transmission, and 6-speed manual transmission.

• Based on real data from BMW M2 (in collaboration with BME Automated Drive Lab), a mathematical model for the vehicle dynamics was built using Matlab/Simulink. Also, another model was built using Simscape components with both manual and automatic gear shifting.

Successfully completed the final project in the Complete Self-driving Car course on Udemy. Employed advanced techniques, including deep neural networks and feature extraction using convolutional neural networks, for Behavioral Cloning. The project, implemented in Python, involved continuous regression to drive a virtual car autonomously.

Developed a robust Traffic Sign Classifier as part of the Complete Self-driving Car course on Udemy. Utilized Keras and deep learning techniques to create an efficient model for traffic sign recognition. The project was implemented in Python.

Dec 2017

Developed a MATLAB/Simulink model for a fully electric vehicle, specifically the Tesla Model S. Implemented advanced control techniques, including a flux vector controller (Field-Oriented controller), to control the induction motor. The project focused on electric vehicle modeling and simulation.

Utilized MATLAB/Simulink to construct three distinct models of an elevator drive system. Conducted a comprehensive comparison of three control scenarios: single-loop controller, two-loop cascade controller, and three-loop cascade controller. Assessed the performance of these models under various disturbance conditions.

An automatic vehicle cruise control system was built using MATLAB/Simulink, to compare the performance of the system when using two different controllers: PID controller and Fuzzy logic controller.

Designed and implemented a MATLAB software application featuring a graphical user interface (GUI). The GUI facilitated the training of a neural network to learn different logic gates. Employed the backpropagation learning method to achieve this functionality.

Successfully constructed a wireless agriculture monitoring station with hardware components. Integrated multiple sensors into the system, which was controlled by two Arduinos—one located on the farm and the other in the farmhouse. This project contributed to efficient agriculture monitoring practices.

Physical model (hardware) was built for DC to DC converter: Buck converter. Similarly, the same model was built using Matlab Simulink®.