Ryan Matthew Sander
MIT EECS | Lincoln Laboratory Undergraduate Research and Innovation Scholar
Deep Reinforcement Learning and Game-Theoretic Control for Autonomous Navigation in Multi-Agent Settings
- Artificial Intelligence and Machine Learning
Daniela L. Rus
As autonomous vehicle technologies continue to mature and more of these vehicles enter our roads, coordination between these vehicles is becoming increasingly important. Robust coordination will enable safer and more efficient transportation in these multi-agent settings, saving lives, time, and energy. In order to extend the capabilities of autonomous vehicles in multi-agent settings, we are developing a reinforcement learning-based autonomous navigation platform based off of game-theoretic control. This simulation platform leverages state-of-the-art machine learning techniques in tandem with OpenAI’s gym and game and optimization theory. This platform will serve as a testbed for improving autonomous navigation in multi-vehicle settings.
“Developing strong research skills through SuperUROP will enable me to better affect change through my career. As a student with experience in machine learning, robotics, remote sensing, and economics, this project integrates well with my academic background. I’m excited to develop impactful solutions to pertinent problems in the autonomous vehicles space because I believe these solutions will help save lives, time, and energy on our roads.”