Embedded Files
LLM-Driven Task Planning for Autonomous Robots
LLM-Driven Task Planning for Autonomous Robots
This project highlights the application of natural language-driven control to both a Tello drone and a quadruped robot using a Large Language Model (LLM). The system interprets user commands in natural language, translates them into actionable instructions, and executes them sequentially. Powered by OpenAI's LLM (e.g., GPT-4), this approach enables drones to perform tasks such as takeoff, landing, hovering, and directional movements, while quadruped robots can execute locomotion and task-specific behaviors with precision.
Experimental Field Robotics
Experimental Field Robotics
Developed GPS-based navigation systems for autonomous ground robots (UGVs) and aerial vehicles (UAVs) using ROS 2 and sensor fusion. Integrated real-time GPS and IMU data to enable waypoint-based path tracking, precision landing with ArUco markers, and dynamic re-planning. Validated the frameworks extensively in both simulation and real-world field experiments.
Energy-aware UAV-UGV Collaborative systems
Energy-aware UAV-UGV Collaborative systems
In today’s dynamic operational environments, leveraging the complementary strengths of robotic systems can drive groundbreaking innovations. This energy-aware mission planning system integrates the capabilities of Unmanned Aerial Vehicles (UAVs) and Unmanned Ground Vehicles (UGVs) for seamless cooperation. While UAVs are adept at fast aerial tasks, they are limited by battery life. UGVs, moving steadily along roads, provide the solution by offering on-the-go recharging of UAVs. This coordinated approach optimizes energy usage, improves mission efficiency, and extends the operational range of both robots. With potential applications in truck-drone delivery, last-mile logistics, persistent surveillance and reconnaissance (PSR), and more, this system offers a versatile solution for a range of real-world challenges.
Automated industrial inspection using heterogeneous multi-robot system
Automated industrial inspection using heterogeneous multi-robot system
This project focuses on developing a multi-agent collaboration strategy between a UAV and a quadruped robot to perform autonomous industrial inspection tasks. The UAV utilizes AprilTag markers and OpenCV for precise visual-inertial localization, generating navigation maps. These maps are then used by the quadruped robot, equipped with a 4-DOF robotic manipulator, to execute path planning algorithms and perform autonomous manipulations.
Computer vision based fire detection and automated suppression system
Computer vision based fire detection and automated suppression system
This project developed an image processing-based multilayered algorithm using OpenCV for real-time fire detection. An Arduino-powered autonomous fire suppression system was engineered to integrate seamlessly with the fire detection algorithm. The system's robustness and reliability were validated through a lab-scale prototype demonstration, showcasing its potential for effective fire mitigation.
Page updated
Google Sites
Report abuse