Planování trajektorie pro skupinu helikoptér v neznamém prostředí s omezením na množství sdílených informací

Abstract

A system composed of multiple UAVs can solve several complex tasks which are either difficult or unfeasible with a single UAV. Safe and fast operation are crucial when multiple UAVs are simultaneously working in a shared environment. This thesis focuses on studying and developing a method for collision-free trajectory planning by a multi-UAV system. The proposed method uses sensors on-board the UAV to create an online map of the environment and detect the motion of other UAVs. Theses detections are employed to predict the future trajectory of the UAVs. The map along with the predicted trajectories is used to design a collision-free trajectory to a pre-specified goal. The proposed method is decentralized and does not require any prior information about the environment or other UAVs which makes it easier to scale to large multi-UAV systems. The method is also independent of any external localization infrastructure making it ideal for real-world applications.

A system composed of multiple UAVs can solve several complex tasks which are either difficult or unfeasible with a single UAV. Safe and fast operation are crucial when multiple UAVs are simultaneously working in a shared environment. This thesis focuses on studying and developing a method for collision-free trajectory planning by a multi-UAV system. The proposed method uses sensors on-board the UAV to create an online map of the environment and detect the motion of other UAVs. Theses detections are employed to predict the future trajectory of the UAVs. The map along with the predicted trajectories is used to design a collision-free trajectory to a pre-specified goal. The proposed method is decentralized and does not require any prior information about the environment or other UAVs which makes it easier to scale to large multi-UAV systems. The method is also independent of any external localization infrastructure making it ideal for real-world applications.