Research

Cognitive Optimization and Relational Robotics Lab

The Wheelchair Tennis Robot team

We were developing a real-time tennis robot. The robot is comprised of a robotic arm and an RC wheelchair. I worked on the perception team designing a distributed, GPU-optimized computer vision algorithm for real-time tennis ball detection. We deployed the algorithm on a group of 6 Jetson Nanos with ZED 2 cameras to localize tennis balls and predict their trajectories. I also performed extensive testing to ensure sufficient range and detection frequency to enable a robot arm to be able to swing and hit the ball on a tennis court.

The Airplane Safety team

We were developing a system to safely recover the altered dynamics of a damaged aircraft. I successfully trained a network to encode aircraft system dynamics. I also deployed an MPPI algorithm to control this system in a simulation environment. This work builds on previous research from our lab.

Autonomous Control and Decision Systems Lab

In 2018-2019 we developed an early warning system for a ROS simulation-based car. We also tested the algorithm on a remote-controlled vehicle. I designed a computer vision algorithm that we trained to focus on regions of interest in a test track video. We monitored the uncertainty of the control algorithm and observed its much more gradual increase. This can be expressed as an early warning detection, which could be a trigger to switch to manual control mode in a situation, where time is of the essence. The paper was published in CoRL 2019

Work experience

Blue River Technology

This summer I am working at Blue River on designing and running experiments to evaluate the stereo camera system of John Deere autonomous tractors. I am also redesigning elements of the ROS2 communication platform for the onboard VPUs

Deutsche Bank

In 2019 I spent a summer working on two projects at DB. I was very excited to develop an ML model for a project to analyze financial news sentiment for stock market investors.

Major projects

pyxy.io

Last year I designed a recommender algorithm with a TSP solver to determine optimal personalized urban travel itineraries.

I built a server written in Go and deployed it with Kubernetes on Google Cloud Platform and created a React Native app to serve it.

Work on this project is currently paused, however, I wrote a Medium article about it!

Other recent projects

2022

I am starting 2 new and exciting projects, I’ll update you as soon as I make the repo public.

2021

I worked on improving a deep learning hate speech detection algorithm as part of a team.

I am looking to build a lip-reading system aiming to transcribe mute speech videos. Check out the beginnings of the project here.

Last semester I was part of a team where we compared DAgger and Actor-Critic methods in driving a simple model vehicle in varying environments. Check it out here.

2020

As part of a team, I built a price prediction model to help Airbnb hosts improve their listings.

I was part of a senior design project aiming to design a low-cost Mars mission to gather climate variation data.

Pre-2020

As part of a team, I outlined a design for an electrical grid drone inspection system.

I participated in a research project aiming to build a digital twin model of a university water cooling plant and optimize its operating schedule.

I performed image processing to determine a correlation between the drone pictures of Icelandic volcanic soil and localized ADP molecule concentration.

Other achievements

I was part of a team that won the 2020 AIAA Engine Design Competition, the report can be found here.

TL;DR here’s my resume