Payload Delivery through Acrobatic Quadrotor Flip-and-Throw Maneuver Using Model Predictive Control

This paper presents a novel approach for delivering payloads using quadrotor drones, by leveraging acrobatic flip or 'belly-flop' maneuver. This method not only leverages the agility of quadrotors but also offers a practical solution for deploying sensor probes into inaccessible or hazardous areas, such as volcanic vents, without the need for additional mechanical devices such as spring-based or compressed-air projectile launchers. By employing Model Predictive Control (MPC), our methodology enables real-time trajectory planning and precise execution of flip maneuvers, taking into account the dynamics of the quadrotor, payload, rotors, and thrusters. This approach ensures stable and dynamic flights even during aggressive maneuvers. The effectiveness of our method has been tested through simulations in ROS and Gazebo environments, as well as real-world experiments conducted at the ASU Drone Studio using a low-SWaP-C agile quadrotor drone equipped with a Pixhawk flight controller and an Odroid M1 companion computer. These tests demonstrated the feasibility of flip-and-throw maneuvers for payload delivery and highlighted the system's accuracy in trajectory tracking and response times. The findings from this research can advance autonomous aerial robotics in tasks requiring high precision and agility, such as delivering probes into inaccessible zones and optimizing image capture angles through acrobatic flight paths.