Projects per year
Abstract
Determining how an autonomous Unmanned Aircraft System (UAS) should reach a goal position amidst obstacles is a challenging and difficult problem. This thesis treats the subject of path planning and trajectory generation for UAS, while utilizing the ability to move in all three spatial dimensions. The primary contributions of this thesis are an approximate path planner and a geodesic path planner. Both planners are model independent and operate on the surface of the configuration space to identify a length minimizing path.
The approximate planner determines an approximated shortest path by building and searching a visibility graph. This planner maintains this visibility graph to enable fast multiquery searches as well as replanning when changes occur in the work space. As paths obtained from a visibility graph are not continuously differentiable, a trajectory generation method is developed that uses the path to find a collision free trajectory that is more appropriate for flight.
The geodesic planner relates to wavefronttype planning, and identifies continuously differentiable geodesic paths as parametric equations determined by surface primitives given from the configuration space. Consequently, this planner uses a more analytical approach since it relies on combinations of optimal curves.
Both planners operate on an explicit description of the configuration space in a work space containing 3D obstacles. A method was developed that generates convex configuration space obstacles from any point clouds or geometric meshes in work space.
Two approaches are used for generating a trajectory from an existing path. The first generator use Dubins curves to find a collision free continuously differentiable trajectory. The second generator relies on formulating and solving an optimal control problem (OCP) using a Legendre pseudospectral method. The main contributions of this approach are the formulation of distance functions that constrains the trajectory. This allows finding trajectories that follows minimal length paths while optimizing the trajectory according
to a performance index.
The methods and algorithms developed in this thesis are implemented in a planning application and validated through simulated flight using a helicopter model.
The approximate planner determines an approximated shortest path by building and searching a visibility graph. This planner maintains this visibility graph to enable fast multiquery searches as well as replanning when changes occur in the work space. As paths obtained from a visibility graph are not continuously differentiable, a trajectory generation method is developed that uses the path to find a collision free trajectory that is more appropriate for flight.
The geodesic planner relates to wavefronttype planning, and identifies continuously differentiable geodesic paths as parametric equations determined by surface primitives given from the configuration space. Consequently, this planner uses a more analytical approach since it relies on combinations of optimal curves.
Both planners operate on an explicit description of the configuration space in a work space containing 3D obstacles. A method was developed that generates convex configuration space obstacles from any point clouds or geometric meshes in work space.
Two approaches are used for generating a trajectory from an existing path. The first generator use Dubins curves to find a collision free continuously differentiable trajectory. The second generator relies on formulating and solving an optimal control problem (OCP) using a Legendre pseudospectral method. The main contributions of this approach are the formulation of distance functions that constrains the trajectory. This allows finding trajectories that follows minimal length paths while optimizing the trajectory according
to a performance index.
The methods and algorithms developed in this thesis are implemented in a planning application and validated through simulated flight using a helicopter model.
Original language  English 

Publisher  Section of Automation & Control, Department of Electronic Systems, Aalborg University 

Number of pages  143 
ISBN (Print)  9788792328731 
Publication status  Published  2012 
Fingerprint Dive into the research topics of '3D Path Planning for Autonomous Aerial Vehicles in Constrained Spaces'. Together they form a unique fingerprint.
Projects
 1 Finished

EVAHC : Emergency Vehicle Aided by Helicopter
la CourHarbo, A. & Schøler, F.
15/09/2008 → 01/01/2011
Project: Research