Undergrad Projects

Lidar SLAM

My favorite project in this past semester; I've put together an algorithm that maps rooms by integrating lidar and radio time-of-flight data from multiple scan locations. Because the scan locations are unknown, my algorithm has to move and rotate sets of measurements to come up with a valid map. 

The above images show the results of a scan using seven ring measurements around a complex room. The left image is the data coming in from the lidar scanners, and the right image is the output map pieced together. 

 


SEER

What moves stocks? News! But who has the time to read the news and trade stocks all day? If only we had an AI that could do it for us! This was first experiment with AI language processing. I had a program read news articles and CEO twitter feeds, then trained my neural net off DJI performance. The plot shows its performance on test data over December-January 2020. It read and traded based on approximately 1000 news articles, gaining about 3.5% over DJI in a single month. 

 SEER was shut off in February 2020 after it became fixated on COVID-19 news.


Aerial Robotics Autopilot

In the IEEE Aerial Robotics team, I wrote the piloting algorithm that would steer our autonomous plane. Above is an image of the flight path taken when the plane is instructed to hit every x and avoid every circle, called "waypoints" and "skyscrapers". The plane wove between the obstacles to hit all waypoints without colliding.

Sadly, my algorithm didn't get much use time. The plane set to run it was destroyed in a crash-- ironically, piloted by a human at the time.