Prateek Arora

PhD Student

I'm

About Me

Prateek Arora

Robotics & Computer Vision Engineer

I am a second-year PhD student in Robotic Workers Lab at the University of Nevada, Reno advised by Prof. Christos Papachristos. My research focuses on active and interactive perception for autonomous robots. I am also involved in aerial robotics research, focusing particularly on the computer vision aspect of creating resilient migrtory MAV systems.

Prior to pursing my PhD, I completed my Masters in Robotics from the University of Maryland, College Park (UMD). I was a member of the Perception and Robotics Group (PRG), advised by Prof. Yiannis Aloimonos where I worked on autonomous vision-based navigation for quadrotors.

I’m originally from Delhi, India. Prior to joining UMD, I worked as a Research Associate with Dr P.B. Sujit and Dr Sanjit Kaul on self-driving car project in Indian Urban scenario at IIIT-Delhi. Before that, I completed my Bachelors of Technology in Electrical Engineering at Bharati Vidyapeeth’s College of Engineering (BVP), GGSIPU. At BVP, I worked with Asst. Prof. Gargi Mishra on locomotion of legged robots.

In my spare time, I like to fly my own custom-built FPV drone and occasionally go on hikes.

Projects

Here are a few projects that I completed in my graduate and undergraduate years. My projects pertain to different areas like computer vision, quadrotors and mobile robots. For more information, visit my Github page .

  • All
  • Computer Vision
  • Aerial Robotics
  • Path Planning
Unscented Kalman Filter (UKF)
Unscented Kalman Filter

Estimation of IMU orientation using Unscented Kalman Filter.

Face Swap like snapchat
Face Swap

End-to-end pipeline to swap faces in a video just like Snapchat’s face swap filter.

Path Planning with differential constraints
Path planning with differential constraints

Implemented A-star path planning algorithm with differential constraints on turtlebot robot in gazebo.

Drone Quadrotor Navigation
Landing on Circular Bullseye

Estimated 3D pose of CCTag marker in real-time to land a quadrotor on it.

Panorama Stiching Computer Vision
Auto Panorama

Panorama stitching using Deep leaning and Traditional approach.

A * path (motion) Planning
A-star algorithm

Implemented A* algo to find path from start to goal location in a static map.

Dijkstra's algorithm path planning
Dijkstra's Algorithm

Implemented Dijkstra's algo to find path from start to goal location in a static map.

madgwick filter Quaternion
Madgwick Filter

Estimation of IMU orientation using Madgwick filter.

PB lite boundary detection
PB-Lite Boundary detection

Boundary Detection using a modified Probability of Boundary Method.

GapFlyt Flying through gaps
Flying Through Gaps

Implemented Gaussian-Mixture-Model to detect colored windows and used it as a feedback to autonomously navigate a drone through it.

simultaneous localization and mapping SLAM
Structure from Motion

Reconstruction of a 3D scene and simultaneously obtaining the camera poses using multiple images.

Optical flow Moving object detection
Avoid Wall using Optical Flow

The aim of this project was to detect a wall in real-time using a monocular camera mounted on a quadrotor and then avoid it.

Autonomous Navigation
Safe Autonomous Navigation

Detect a bridge that provides passage across a river. Navigate a quadrotor to the other side of river by going over the bridge.

Stereo Visual Odometry
Stereo Pose Estimation

Estimated 3D trajectory of a quadrotor equipped with a stereo camera using the optical flow equation.

Convolutional layer network Classification
Image classification using CNN

Implemented and evaluated different CNN architecture for image classification.

Camera Calibration Computer vision Zhang
Camera calibration

Estimating intrinsic camera parameters focal length, distortion coefficients and principle point.

Baxter Motion Planning
Pick and Place using Baxter Robot

Implemented simple pick and place using baxter’s ROS API using in-built kinematics and inverse kinematics engine.

Eight Piece Puzzle
8 Piece Puzzle

Implemented the solution to 8 Piece Puzzle using brute force to match numbers on the tile to a goal configuration.

Resume

Contact

Location:

1555 N Sierra St, Reno, NV 89503