Signal Source Traversal Based on Quadrotors in Environments With Obstacles: A Data-Driven Receding Horizon Control Approach

Abstract

This paper deals with the problem of signal source traversal based on a quadrotor in an environments with obstacles by proposing a data-driven receding horizon control approach. The proposed control approach includes two parts: One part is a data-driven learning method while the other part is a receding horizon control method. Gaussian process as a kind of data-driven learning method is used to provide the efficient posterior probability distribution of the possible positions for all the signal sources based on the received signal strength data. It should be pointed out that the obstacles in the environment obstruct signal detection and the antennas diverge the direction of signal transmission such that the positions of signal sources are frequently updated. In order to cope with this kind of dynamic events, a receding horizon control approach in the light of a new cost function is proposed to enables the quadrotor to traverse signal sources, where the generated trajectory is smooth and the quadrotor can be guided to efficiently avoid “L” shape obstacles. Finally, simulation and experimental results show that the quadrotor controlled by the proposed data-driven receding horizon control approach can move along the smooth trajectory to avoid obstacles and traverse all the signal sources.

Keywords

• Quadrotor
• signal source traversal
• Gaussian process regression
• receding horizon control
• data-driven learning