An intelligent, heuristic path planner for multiple agent unmanned air systems

Abstract

In situ observations of atmospheric variables are currently obtained with radiosondes, collecting data along uncontrolled trajectories. As an alternative, we propose an unmanned air system comprising a swarm of unmanned aerial vehicles, released from high altitude weather balloons. Their trajectories are optimised for efficient sampling, with an objective function measuring the space-filling properties of the entire swarm. The dynamics of the aircraft swarm are captured in a number of primitive manoeuvres simulated with JSBSim. A combination of these paths forms a set of moves available to a greedy heuristic algorithm, which determines which of the flyable paths is optimal according to the objective function. At each heuristic optimisation step a decision is made upon the next single move and, once the move is complete, the heuristic repeated, resulting in a stitching together of optimal moves from the flyable set. The proposed path planner comprises of a centralised algorithm, which executes offline, thus, each aircraft executes the heuristic synchronously and the result is a cloud of waypoints to be flown by the swarm. A case study based on orographic flows over South Georgia is used to test the performance of the algorithm in windy environments. The results indicate good performance of the algorithm, even in high, unsteady wind fields.

More information

Identifiers

ORCID for Andras Sobester: orcid.org/0000-0002-8997-4375

Catalogue record

Export record

Altmetrics