Pratik Acharya

Intro

I am a graduate student in Robotics at the University of Maryland, currently working towards learning how to use robots to solve real world problems. Starting as a curiosity to know how machines work, I developed an early interest in robotics which later directed me towards choosing my undergrad in Instrumentation and Control Engineering. Having learnt the basic building blocks that enable a robot to function as it does, I decided to continue my education and improve my skills while working with robots directly. I joined the Indian Institute of Bombay, as a Project Research Assistant and worked on developing autonomous solutions for everyday problems.

Apart for robotics, I also enjoy reading, sketching, dancing and playing musical Instruments.

Projects

Musketeers: Autonomous pickup and delivery fleet

Working in a team of three, we created an autonmoous delivery fleet of 20 robots using ROS, which could access delivery and pickup locations via a graphical user interface or a command line input. Each robot was given independent navigation that worked on generating paths to the end goal while avoiding static and dynamic obstacles. We utilized ROS's "move_base" framwork to enable path planning, localization and obstacle avoidance. A custom made environment was mapped using ROS's "gmapping" package and was utlizied to demonstrate it's autonomous navigation capabilities. More information about the project can be access at this github repository.

Lane and Curvature detection for Autonomous Driving

Created an OpenCV pipeline to detect lanes on a highway using Canny Edge Detector and Hough Transform. The detected lane lines were classified into "Dotted" and "Solid" to aid autonomous driving. Lane detection was further evolved into detecting the curvature of the road.

Path Planning and Patrolling for a team of Car-like Robots in a Campus Environment

I was inducted into a team of reasearch professionals working on autonmous mutli-vehicle patrolling. Here, I worked on developing obstacle avoidance and road intersection behaviour for the vehicles, and converted the developed MATLAB code into python to be implemented with ROS.

Leader Follower Robot

Implemented Leader Follower behavior on a TurtleBot 3 robot. Programmed a Deep Learning based human and AprilTag detection pipeline using YOLO with ROS.

Conceptual Design of a Robotic Vehicle Refuelling System

After completing the multi-robot patrolling project, I was assigned to create an autonomous solution for refuelling cars. Having almost no experience in CAD modelling, it proved to be a challenge to design a complete system, but I used this opportunity to learn modelling and designed a solution in, three weeks. Two designs were created for the system and were later presented at an International IEEE conference in Robotics and Automation. The video links for the model demostrations can be be found here, and the paper can be access here.

ACME TX9 Mobile Robot: Perception Module

In my master's course I, along with my teammate, worked on a perceoption module for a mobile robot, which could detect humans using momcular camera as input. We used YOLO for detecting humans in a camera frame, and then used transformations to provide the location of the detected humans to the robot. More details can be found here.

About

I currently work part time as a Researcher at the Bio-Imaging and Machine Vision Lab at the University of Maryland. I have been working here since May 2022 and during which period I worked on two projects, one involving Imitation Learning and Machine Vision and other which required networking multiple industial instruments for mushroom processing.

Contact

• Twitter
• GitHub

Elements

i = 0;

while (!deck.isInOrder()) {
    print 'Iteration ' + i;
    deck.shuffle();
    i++;
}

print 'It took ' + i + ' iterations to sort the deck.';