Collision-free motion planning for mobile agents is a challenging task,
especially when the robot has to move towards a target position in a dynamic
environment. The main aim of this paper is to introduce motion-planning
algorithms using the changing uncertainties of the sensor-based data of
obstacles. Two main algorithms are presented in this work. The first is
based on the well-known velocity obstacle motion-planning method. In this
method, collision-free motion must be achieved by the algorithm using a
cost-function-based optimisation method. The second algorithm is an
extension of the often-used artificial potential field. For this study, it
is assumed that some of the obstacle data (e.g. the positions of static
obstacles) are already known at the beginning of the algorithm (e.g. from a
map of the enviroment), but other information (e.g. the velocity vectors of
moving obstacles) must be measured using sensors. The algorithms are tested
in simulations and compared in different situations.