Autonomous TurtleBot

I developed an autonomous navigation system for a TurtleBot as part of Carnegie Mellon’s Intelligent Robot Systems course, focusing on full-stack robotics integration from perception to decision-making. The robot was designed to operate autonomously in a structured environment, completing tasks such as line following and executing U-turns.

The system was built in ROS2 and integrated multiple sensing modalities, including camera and odometry data. I implemented sensor fusion and state estimation techniques to maintain a reliable pose estimate, and used computer vision pipelines for line detection and environmental awareness. Motion planning and control were handled through classical planners and local controllers, ensuring smooth and stable navigation.

At a higher level, I designed the robot’s decision logic using a Behavior Tree architecture, allowing modular, interpretable, and extensible control over autonomous behaviors. The project emphasized real-world robotics challenges such as noisy sensors, imperfect perception, and system-level debugging, and culminated in a fully autonomous demonstration.

Key tools & concepts: ROS2, Python/C++, computer vision, sensor fusion, motion planning, control, Behavior Trees, systems integration.