Study of swarming algorithms for coordination of multiple UAVS

Abstract

This project aims to research on swarming algorithms for coordination of multiple UAVs. Based on the swarming behaviours observed in nature, the feasibility of this

phenomenon being applied to UAVs is tested out in this project. Over the years, numerous researches have been done. Modifications and various methods have been employed. However one method called Artificial Potential Field (APF) stood out as it can be easily applied due to its intuitiveness. Its principle is based on projecting an artificial potential field onto the domain in which the UAV works in. By

differentiating this potential and commanding the UAV to move in the direction of negative gradient, the UAV will move to a lower potential in the shortest time possible. In this project, Marr wavelet and shallow parabola were being utilised to demonstrate the objective. However a problem arose. The simulations were unable to fulfill basic requirements of a swarm of UAVs. Upon troubleshooting, one of the reasons for the failure of the simulation was that the algorithm used did not account for heading feedback of the UAVs. Hence an alternative approach was implemented. Frenet‐Serret system of equations and various control laws described later in the report accounted for the error previously encountered. With the aid of these potential functions and manipulation of simple algorithms and control laws, obstacle avoidance and target seeking capabilities was realised.