Abstract: This talk addresses the computation of vector fields for unmanned
aerial vehicles (UAV) guidance. Two main approaches will be presented.
The first one solves the problem of guiding an UAV by constructing a
polygonal corridor in the robot workspace and computing a two
dimensional, fully continuous vector field inside the corridor. The
corridor may, for example, be defined by a set of waypoints. We present
experimental results with an autonomous airplane that illustrates and
validates the approach. The second approach is more generic and is able
to compute vector fields to guide a configuration to converge and
circulate one dimensional, time varying curves embedded in
n-dimensional spaces. Curves in three-dimensional spaces are easily
defined given a set of waypoints or a specification of the desired
robot behavior. This methodology is being tested with the GRASP
Quadrotors.