●Working on incorporating safety in deep-learning and providing assurances during design and deployment-time

● Built a custom 1/10 scale car platform capable of drifting(aggressive maneuvers) with localization (SLAM) pipeline using lidar, sensor fusion(EKF), and developed models for simulators

● Path planning controllers for the car use deep RL (TRPO / CPO) and are monitored using reachability analysis, and KeYmaera X (work by other collaborators)

● Working on a using runtime safety monitors for safe-exploration in deep RL

● Working on improving developing better models and Sim-to-Real transfer for aggressive maneuvers

Mentor: ​ Prof. John Dolan Collaborators: ​ Prof. David Held, Prof. Stefan Mitsch Dr. Aleksey Nogin (Hughes Research Lab)

●My work with Prof. John M. Dolan involved behavioral planning (decision making) for self-driving vehicles at unsupervised intersections.

● Graduate Thesis:​ Behavior planning at roundabouts

● Developed planners using deep-reinforcement learning, improving performance over

traditional rule-based techniques

● Developed a new method to enable generalization across

different traffic densities

● Currently extending policy transfer of behaviors from

right-hand drive to left-hand drive scenarios

● Mentored summer intern working on RRT-iLQR and NMPC

based planners for vehicles