C++ Computer Vision

JITESH SONKUSARE

Ñ Boston, MA 857-***-**** # *********.*@************.*** ï LinkedIn profile § Github profile EDUCATION

Northeastern University, Boston, MA Dec 2024

Master of Science in Robotics, Concentration in Electrical and Computer Engineering, GPA 4.0 Courses - Advanced Computer Vision, Foundations of Artificial Intelligence, Reinforcement Learning, Autonomous Field Robotics, Robot Sensing and Navigation, Mobile Robotics, Robot Mechanics and Control, Legged Robotics Veermata Jijabai Technological Institute, Mumbai, India Aug 2018 - May 2022 Bachelor of Technology in Electrical Engineering, CGPA 9.12/10, Rank- 5/75, Top 7% of class Relevant courses - Computer Programming, Linear Algebra, Signals and Systems, Industrial automation and controller, Microprocessor and Microcontroller, Analog and Digital Circuits, Digital Signal Processing, Control Systems TECHNICAL SKILLS

Languages: Python, C++ (Data Structures and Algorithms, Solved 400+ Problems of Leetcode), C, XML, SQL, NoSQL Developer Tools: ROS, Git, OpenCV, Simulation Softwares-(Gazebo, V-rep, MATLAB, Simulink, Proteus), problem-solving Algorithms: SLAM, Path planning (A*, Dijkstra), Sensor fusion, Extended Kalman Filtering(EKF), Monte Carlo, Q-learning Hardware Skills: Lidar, Cameras, IMU, GPS, RTK GPS, Jetson Nano, Raspberry Pi, ESP-32, Arduino, PLC EXPERIENCE

Noah Medical, Pleasanton, CA Jan - Aug 2024

Robotics Software and Controls Co-op - Research and Innovation

• Augmented Reality System: Developed AR in C++ and Python for autonomous robot guidance, integrating real-time camera and raw sensor data for virtual marker placement, improving surgical precision and product safety

• IMU-based Shape Sensing: Build a high-precision shape estimation exoskeleton for Bronchoscope in C++ and Python using IMUs as control points for cubic spline interpolation thereby improving the current system by replacing expensive optical sensors with IMUs, resulting in cost savings of several hundred thousand dollars for the company

• Object Detection using Computer Vision and Deep Learning: Developed bronchoscope tip detection and angle estimation using color segmentation and deep learning in real-time, improving surgical precision and replacing costly EM sensors, leading to significant cost savings

• Outgoing Quality Control (OQC) Software Line Automation: Created a GUI software for automatic testing of the bronchoscope robot for the production team, automating 12 different tests to streamline the validation process, improving efficiency and accuracy in quality control by using Optitrack

• Error-Analysis and Calibration GUI software: Developed a Python-based GUI error analyzing and calibration tool using Tkinter that combines video and data from OptiTrack in the same window for improvement in existing robots Boosted company progress by 20% in 8 months by developing 7 projects, improving efficiency and productivity. PROJECTS

Structure From Motion (3D Reconstruction), Northeastern University, Boston, MA Nov - Dec 2023

• Developed 3D reconstruction of a Buddha sculpture in python using epipolar geometry, SIFT, and RANSAC. Calculated fundamental and essential matrices, estimated camera poses, and triangulated 3D points

• Optimized the reconstruction with bundle adjustment using GTSAM and visualized the final 3D model, successfully recovering camera poses and creating an accurate 3D scene Warehouse Robot Automation with Reinforcement Learning, Northeastern University, Boston, MA Nov - Dec 2023

• Developed and implemented RL algorithms (Monte Carlo, Q-learning, Deep Q-learning) for autonomous warehouse robots navigation, successfully demonstrating advanced path planning with A* comparisons for robotic systems Autonomous Mobile Robot for Disaster Management, Northeastern University, Boston, MA Jan - April 2023

• Utilized Cartographer ROS, move base, and explore lite for SLAM, motion planning, obstacle avoidance, and feature extraction for empowering TurtleBot3 (Raspberry Pi, camera, and lidar on-board) to navigate unexplored environments autonomously, create comprehensive maps, and precisely locate AprilTags (simulate victims) for disaster management Simultaneous Localization and Mapping, Northeastern University, Boston, MA Nov - Dec 2022

• Worked together in a group of 4 to implement 4 SLAM algorithms (OpenVSLAM, RTAB Map, SPTAM, and Cartographer) using Python, C++, and ROS. Conducted statistical data analysis to compare differences like speed, precision, reliability, memory usage, and computational power required to determine the best option for a specific application. Sensing and Navigation, Northeastern University, Boston, MA Sept - Dec 2022

• Designed individual driver nodes using Python and ROS for GPS, RTK GPS, and IMU sensors (Sensor Fusion) and performed analysis with MATLAB on dead reckoning and velocity estimation to enhance path planning and navigation Quadruped, Veermata Jijabai Technological Institute, India May - Aug 2019

• Led and collaborated in a team of 3 to design, 3D print, and build an autonomous quadrupedal robot using Python and microcontrollers (Jetson Nano, Raspberry Pi) capable of traversing rugged terrains using creep and trot gait algorithms

• Solved kinematics of 3-DOF robot legs for open-loop odometry, studied necessity for closed-loop odometry by simulating it on V-rep and Gazebo and observed that it improves overall accuracy by 50% when implemented on hardware

Contact this candidate