Projects

Developed a novel SLAM algorithm combining camera, lidar, IMU, and GPS sensor information for real-time UAV and aircraft navigation with over 99.5% estimation accuracy - 2x the accuracy of state-of-the-art methods. This system was tested with Bell 412 Helicopter and M600 Drone. 
(C++ in ROS framework with OpenCV and PCL libraries)

• Improved the 3D model texturing pipeline using visual, thermal, and methane concentration images of the Xplorobot desktop software.
• Developed three new algorithms to monitor methane emission in industrial plants, including a methane flowrate calculator, an interactive 3D bounding box for the GUI, and an automatic leak region detection method using methane textures.
(C++, OpenCV, PCL libraries, VTK, QT, GDB and Valgrind)

I led software development for a real-time image classification and tick detection system, achieving over 92% accuracy using the MobileNet neural network. Working closely with a diverse team, including investors, we successfully created the first prototype of an efficient tick detection device by seamlessly integrating our software with the ESP32-S3 Eye board. This project underscores my commitment to effective teamwork and the practical application of technical expertise in solving real-world problems (Python, Tensorflow, C++).

Controlling On-board arm maneuvers of a Continuous track wheeled robot to self-balance and navigate uneven terrain. To this end, I am using the principles of Zero Moment Point (ZMP) and support polygon of track wheel and ground. Applications of this research are in disaster management and field robotics (C++, simulated in ROS/Gazebo).

Autonomous end to end system to 3D reconstruct artifacts using 2D images. The system consists of a rotating mechanical platform to capture series of photographs of a given object in a hemispherical area, an algorithm to output a high-quality 3D reconstruction of the object and mobile phone application to retrieve 3D models, visualize and interact with them (C++ with OpenCV library for the 3D reconstruction algorithm, Java for the GUI).

Implementing and testing an underwater acoustics communication link between AUV and surface vehicle using sweep-spread-carrier (S2C) technology and improving underwater localization relative to the surface vehicle (C++). 

Developed a supervisory level BMS which is capable of displaying, monitoring each cell voltage, and taking intelligent actions based on battery module voltage, temperatures, and faults.

​

Built circuits and integrated battery cells to develop the 60kWh, 495V battery pack of the car.