Autonomous Navigation Using Reactive Gap Following on an F1Tenth Platform

[Github]

In this project, we implemented a reactive gap following algorithm on a physical F1Tenth autonomous vehicle. The method involved processing LiDAR data to identify the closest obstacle, creating a safety bubble around it by zeroing points within a certain radius, and then detecting the largest gap free of obstacles. To ensure smooth navigation, you applied a rolling average filter to reduce sensor noise and used a disparity function to account for the vehicle's physical dimensions, effectively widening the navigable gap. 

The vehicle dynamically adjusted its speed based on the steering angle to maintain stability and responsiveness while navigating a track with obstacles. After successful simulation testing, you deployed the algorithm on the real car, tuning parameters such as bubble radius, speed, and LiDAR range to prevent oscillations and jerky movements, ultimately enabling the vehicle to complete three laps around the track autonomously without relying on a global map.

ROS-Based Vision-Guided Control System for Egg Manipulation Using uArm Swift Pro

[Github]

This project utilized the ROS (Robot Operating System) framework to implement a vision-guided control system for the uArm Swift Pro robotic manipulator. By integrating ROS with Python, the system processes real-time data from an Intel RealSense camera to detect eggs and determine their precise locations. The ROS-based software performs inverse kinematics calculations to translate these target positions into joint angles, enabling accurate and smooth movements of the robotic arm. 

The system further manages the vacuum gripper to pick up and release eggs gently. Through ROS subscribers and publishers, the software ensures coordinated communication between perception, motion planning, and actuation components, allowing for responsive and adaptive manipulation. This approach proved effective in achieving controlled pick-and-place operations, demonstrating the capability of ROS in enabling sophisticated robotic tasks in agricultural automation.

Universal Robots CORE Training

Did a three-day program building skills in robot operation. Covered programming, operating, troubleshooting, optimizing, and more, this hands-on course equips individuals with the skills they need to navigate the world of collaborative and industrial robotics.

EggBot: Autonomous Egg Detection and Retrieval System.

The project focuses on designing and developing an autonomous egg detection and collection system on ROS2. The system uses a combination of LIDAR sensors, a camera, and a Mini PC to navigate, detect, and interact with the environment. LIDAR provides environmental understanding for path planning, while a camera employs a YOLOv11s model to identify and track eggs, ensuring they remain centered in the camera frame. The Arduino, receiving data from LIDAR and camera nodes, controls the machine’s movements via PID controllers. 

Ultrasonic Sensor based Navigation

[GitHub]

This work was to develop a robot using two motors (right and left) and three ultrasonic sensors (front, right, left) for obstacle avoidance and wall-following. 

Implemented PID controller to adjust motor speeds based on sensor readings, ensuring the robot maintains a target distance from walls and stops when obstacles are detected in front.. This work was done in the AGEN 836 course under Prof. Santosh Pitla.

Path Planning

[GitHub]

Implemented modified A* Algo for path planning of an Ag Bot moving in a field with trees at various locations and implemented a controller using the ikpy library in python for the Robotic Manipulation in the virtual environment in the Webots. I did this work in my Bachelor's Thesis.

Autonomous Apple Harvesting Robot-Manipulation

[GitHub]

Implemented modified A* Algo for path planning of an Ag Bot moving in a field with trees at various locations and implemented a controller using the ikpy library in python for the Robotic Manipulation in the virtual environment in the Webots. I did this work in my Bachelor's Thesis.

It Is Not the Journey but the Destination: Endpoint Conditioned Trajectory Prediction

[Link] [code]

Undertaken as part of the Machine Learning Reproducibility Challenge 2020, we reviewed the above-accepted ECCV 2020 publication for verification of its claims through computational experiments. We reproduced all the results which are within 5% error from the ones claimed in the paper.

University Rover Challenge

[challenge]

Worked in the Mars Rover Team of the Autonomous Ground Vehicle Research Group, IIT Kharagpur

Developed the wheel, chassis and suspension for rover prototype with 15 deg gradeability and max speed 20cm/s.

Designed a 5-DOF modular robotic manipulator with 2-finger grip for semi-autonomous on-board equipment repair.