GSoC’22 RoboComp project: Extending the Social Navigation component to realistic scenarios

30th June 2022

About me

Hey! I am Sushant, an undergraduate from BITS Pilani (Goa Campus), India. I am currently pursuing my degree in Computer Science. Since my high school days, I have loved programming, and that craze has grown over the years. I was introduced to this fascinating field of robotics by the seniors of my college, and I worked on exciting projects. I also developed an interest in Deep Learning and Deep Reinforcement Learning and completed courses to enhance my understanding. My current research interests lie in the applications of DL and DRL in robotics, Intelligent Robots, Autonomous Vehicles, and Computer Vision. The current opportunity at RoboComp lies well in my interests, and I would love to make the most of it!

About the Project

Socially aware path planning enables a robot to navigate through a crowded environment causing the least amount of discomfort to the surrounding people. Building upon the work done in GSoC ‘21, the aim of the project is to account for the interactions between the different entities in the environment by adding interaction nodes between entities in the graph representation. Since the environment posed a bit of a problem in training last time, the project also aims to develop a robust and light weight OpenAI Gym style environment. The current environments provide fairly simple scenarios, thus making our environment a better environment to benchmark various DRL models for the Social Navigation problem. The project also aims to implement DRL models that are able to do well in the environment.

Sushant Sreeram Swamy