"I can't find the way, but my robots can!"
SRISHTI GUPTA
I am a first-year MRSD student at Carnegie Mellon University with a background in classical planning. I am currently focused on learning-based methods for robotics. My work sits at the intersection of planning and learning, where I aim to combine structure with adaptability to build robust robotic systems. I enjoy working on problems that involve real-world uncertainty and require closing the gap between simulation and hardware.
Object-Centric Dexterous Manipulation
Learning wrist motions from humans and executing them as stable, whole-body humanoid behavior.
Semantic SLAM & 3D Object Mapping
Turning visual detections into globally consistent, object-aware semantic maps through learning-based perception and SLAM.
Imitation & Reinforcement Learning for Robotic Control & Locomotion
Tested various learning algorithms on MuJoCo control tasks.
Probabilistic State Estimation for Mobile Robots
Custom-built for environmental data collection.
Code that worked!
Deep Q-Learning for Autonomous Snake Gameplay
Dynamics, Control, & Trajectory Planning for Quadrotor Aerial Robots
Trained a Deep Q-Network to learn collision-free Snake gameplay from state inputs.
Developed nonlinear quadrotor dynamics and control frameworks.
Pure Pursuit Path Tracking for Autonomous Vehicles
Sim2Real: Obstacle Avoidance Differential Drive Robot
Implemented and tuned a pure pursuit controller on a kinematic bicycle model.
Developed a ROS 2–based UGV for sim-to-real deployment.
F E A T U R E D P R O J E C T S
F E A T U R E D P R O J E C T S
P U B L I C A T I O N S
A few iterations later
DPSO-Q: A Reinforcement Learning–Enhanced Swarm Algorithm for Solving the Traveling Salesman Problem
