Portfolio
Vision-Controlled Bionic Arm
Developed a real-time gesture-controlled bionic arm using Python and MediaPipe for hand gesture recognition and Arduino for precise servo control, effectively translating human hand movements into synchronized robotic responses.
Visual Odometery for Monocular Camera
Developed and implemented a Visual Odometry system for a monocular camera using Python and OpenCV, achieving precise pose estimation and trajectory mapping in real-time, thereby enhancing the understanding of camera motion within 2D and 3D environments.
Depth Completion for Self-Driving Cars
Developing a novel approach for depth completion using a combination of RGB and depth data to improve accuracy and completeness.
Image Segmentation
Developed semantic image segmentation using U-net Architecture 
Image Classification of Stroke Blood Clot Origin
Leveraged deep learning techniques, including transfer learning, to develop an algorithm that accurately classifies blood clots based on their origin, with an accuracy of 72.4% and 73.75% for CE and LAA clots, respectively. 
Bipedal Walking
Crafted a 12DoF Dynamic Walking Robot in Matlab-Simulink that overcomes the challenges of traversing on flat terrain by generating precise footstep planning, zero moment points, and applying inverse kinematics for seamless movement. 
SPEAR: Soft Robotic EMG Assisted Rehabilitation
Developed a bio-inspired soft robotic rehabilitation solution for foot paralysis in stroke patients, utilizing EMG signals for muscle activation and a custom-designed pneumatic circuit for movement control. 
Detection of Varus Thrust in Knee Osteoarthritis
Developed a wearable device with IMU sensors and applied Kalman filters to detect Varus Thrust in older adults’ gait movements, with real-time prediction updates using Pygame, as part of research work in Neuromechanics Laboratory at IIT Madras.