3D Path Planning for Multiple Aerial Robots using Genetic Algorithm and B-Spline Curves

Abstract

UAVs (Unmanned Aerial Vehicles) are becoming more and more affordable for many military and commercial applications. For mission execution latest trend is to use multiple vehicles for missions with multiple tasks/targets. The generation of a feasible path for these vehicles keeping in view their kinematic and environmental constraints is a challenge. In this thesis we use Genetic Algorithm and B-Spline curves for the realization of feasible paths for these vehicles. The problem is such that m UAVs start from the specified m source points, have to traverse the specified n number of targets and return to the given m destination locations safely and with minimum fuel consumption. The generated paths must satisfy vehicles’ kinematic and dynamic constraints as well some mission constraints defined by the mission’s objectives. In addition to that the algorithm must produce paths which avoid collision with operational area terrain. The kinematic constraints we are going to handle are the maximum altitude ceiling, maximum curvature, maximum torsion and maximum climb angle of the vehicle. Maximum curvature is limited by maximum yaw rate of the vehicle whereas maximum torsion is constrained by maximum roll rate of the vehicle. The mission specific constraints we shall cater for are traversal of the targets at specified orientation. The orientation is defined by the azimuth and elevation angles of the vehicle. The final generated path must consume minimum energy and meet all the above mentioned constraints and objectives. The paths shall be represented by B-Spline curves. B-Spline curves are third order differentiable and this property of these curves makes it a good candidate for path representation. The other inherent advantage we have with curves is that very complex path can be represented by relatively small number of control points.