About me

TL;DR

I am interested in the idea of making multi robot team more efficient and reliable than their single agent counterparts. This involves the study of multi robot path planning, dynamic task assignment, interaction with other agent and humans. My plan is to contribute to these subfields in graduate school and in my career beyond that.

Why I chose multi agent systems?

I realized that organic farming is bad for the environment. According to a study published in Nature, organic farms have 15-20% lower yield which leads to more land use and causes deforestation. On the other hand, non-organic farming's excessive use of chemicals causes environmental damage. One way to tackle this problem is to deliver pesticides and fertilizers more precisely, thus reducing their environmental impact. This idea, and my love for Unmanned Aerial Vehicles (UAVs), led me to imagine a future where drones monitor fields, and a team of ground robots mitigate the problem. As I explored this idea more, I realized that naive scheduling and static planning would not work for such an application. Questions of, what happens if one of the robots breaks down? How do we navigate safely amongst humans? And lastly, how do we make robots behave like robust human teams? kept creeping up.
While exploring these questions I got introduced to the field of multi agent system. Since then, I have realized that answering the above question would not only help agricultural robot teams but many other applications such as wildfire management, disaster response etc. Thus, I have chosen to pursue graduate studies in this topic. It will allow me to gain expertise to make multi agent systems a reality in consumer hands.

As a photographer Lytro was revolutionary. Now we could focus after taking the image. It also introduced me to the concept of light field cameras and the field of computational imaging. The idea of using physics and computation to derive more data than a 2d planar image has many applications including in robotics.
One of the first computational imaging project I worked on was an app to remove people from crowded scenes. While neural network based people removal relies on pervious data and a prediction of what could be behind the person, my technique uses physics to complete the picture hence you don’t have to worry about faulty predictions.
I believe computational photography is set to make a great impact in robotics as it addresses challenges like depth prediction with more ease and accuracy than neural networks. Also new type of cameras like Event cameras will allows us to have a different way data is represented and can provide only critical information needed. Thus, allowing us to use lower compute to achieve better results.

The quality of physiotherapy determines the rate of recovery from major operations like knee and hip replacement or from medical conditions like stroke. The demand for physiotherapist is rising but one of the banes of physiotherapy is that most times the exercise requires constant monitoring and assistance by the therapist. Thus, rendering them to be able to care for less people.
What if physiotherapist could be like doctors of internal medicine, who asses the patient condition, prescribes drugs, and moves on to the next patient. This allows them to see more patients. What if there were exoskeletons which allow patients mobility even when their own muscles are too weak for them to be mobile. These are the questions that inspire my interest in robotics based rehab.