Pioneered the design and development of a Quadruped Robot, emulating the walking, turning, trotting, and slope-climbing
abilities of canines. Fabricated the robot using FDM and SLA 3D printing, achieving a 40% weight reduction compared to
traditional acrylic or aluminum while maintaining robust integrity and mobility. Conducted detailed motion analysis
using SolidWorks, optimizing gait patterns to achieve 90% accuracy relative to simulation models. Implemented optimized
multithreaded control for 12 servo motors using an ATmega 2560 microcontroller and PCA9685 servo control board via I2C,
leading to a first-place victory at GUJCOST Robofest and securing a design and utility patent.
Developed a software module using the Agile Iterative Process to simulate a multi-robot system with C++, ROS2, and the
Google Test Framework. Integrated CI/CD pipelines for seamless software deployment, successfully simulating over 25
robots with 99% geometric alignment accuracy and 89% code coverage.
Implemented A*, Dijkstra's, RRT, and RRT* algorithms for optimal robot navigation and obstacle avoidance using
TurtleBot3 and Gazebo. Executed the RRT* algorithm in Python, achieving efficient dynamic obstacle avoidance and optimal
path planning in complex environments. Developed a real-time replanning mechanism using the broken node approach,
enabling robots to seamlessly adapt to obstacles and generate new paths. Demonstrated the algorithm's capability with
100% obstacle avoidance and reliable path optimization across varied dynamic scenarios.
Developed a deep learning model using the Inception V3 CNN encoder to process live video feeds and generate vocal
descriptive captions, achieving 92% accuracy for real-time accessibility for visually impaired individuals. Enhanced
model confidence and reliability with Block Static Expansion and multi-headed attention vectors, significantly improving
the accuracy of real-time descriptions.
Engineered a mobile robot with autonomous navigation and obstacle detection, integrating IMU, Raspberry Pi, Arduino, Pi
Camera, and ultrasonic sensors for real-time object recognition and avoidance. Utilized PID control for precise
localization based on encoder data and programmed efficient pick-and-place operations, achieving an 80% success rate
with color-coded blocks. Soldered a custom PCB for electronics setup and power distribution.
Conducted a comprehensive evaluation of 2D mapping algorithms, including HectorSLAM, Cartographer, and Gmapping. Mapped
a 20,000 sqft hostel floor to validate the study, providing actionable insights for optimal mapping algorithm selection.