Trajectory Planning for Racing Drones

Project report can be accessed here. Code can be accessed on the GitHub here.

Research in racing drones is gaining popularity mainly due its immense application in agile drone flights. Previous works can be classified into four categories gate detection, localization, trajectory generation and control. In this project, a realistic simulation framework is set up that can be utilized to focus on all four aspects. Commonly used minimum snap trajectory is used, combined with differential flatness-based control or nonlinear model predictive control, to demonstrate the basic drone racing tasks and effectiveness of the simulation framework.

The above figures depict photorealistic graphics from Flightmare, dynamics simulation by Gazebo using Robot Operating System (ROS).