Wide Range Global Path Planning for a
Large Number of Networked Mobile Robots
based on Generalized Voronoi Diagrams
Qi Wang, Marco Langerwisch, Bernardo Wagner
Leibniz Universit¨at Hannover, Institute for S ystems Engineering, Real
Time Systems Group, D-30167 Hannover, Germany (e-mail: {wang,
langerwisch, wagner}@rts.uni-hannover.de).
Abstract: This work aims to solving global path planning for multiple mobile robots. In order
to get a collision free path, an accurate map model is always essential. A roadmap is made
based on the generalized voronoi diagram (GVD), which is created with a new proposed voronoi
based parallel thinning (VPT) algorithm. Based on the GVD roadmap, a new multiple mobile
robot path planning algorithm, electric circuit based path planning (ECPP), is proposed.
ECPP is applicable for large number of mobile robots global path planning in large space, it
can efficiently avoid the possibilities of traffic jams, so that the whole multi-robot system can
work properly.
Keywords: Thinning Algorithm; Generalized Voronoi Diagram; Multiple Mobile Robot; Path
Planning
1. INTRODUCTION
Path planning is one of the crucial problems for mobile
robots. This paper provides a new approach for the prob-
lem, how to make a plan for a networked multi-robot
system, so that a large number of robots in a wide range
environment can move in an efficient and harmonious way.
There are some existing approaches for multiple mobile
robots path planning. Bennewitz et al. [2001] introduce an
approach, by applying A* algorithm in a time-space con-
sidering the priorities of missions for each robot. The al-
gorithm gathers all information including the start points
and goal points of all robots, then a collision free path in
a precise time schedule for each robot is made. Each robot
needs to follow a hard time schedule, in other words, the
robot is expected to move to a certain place at a certain
time point, if it is too late or too soon, all the paths of
the affected robots need to be replanned. Such types of
algorithms are not applicable for wide range path planning
for a large number of robots. One of the problems is that
it would take too much time to find a solution for all
robots and since the paths are following a precise time
schedule, any tiny change may destroy the whole plan.
Another problem is that the paths are made according to
priorities. This guarantees the robot with highest priority
works the best, but the performance of the whole multi-
robot system is not guaranteed. Similar approaches are
proposed by Marchese [2006], Langerwisch and Wagner
[2011].
Clark et al. [2003] solved the problem by creating several
local networks and constructing coordinated trajectories
for robots in each network. Hui [2010]’s strategy is also a
local planner. The planner will not be activated until the
moment when the robots are close enough to each other.
Such types of algorithm can solve the multi-robot path
planning locally, it could not prevent too many robots
from choosing the same road section, while the other road
sections are still free.
In all the above approaches, different solutions are made
to find a precise collision free plan for each robot in multi-
robot systems. They are, however not applicable for a
large number mobile robots path planning in wide range
space. In this paper, we propose a new multiple robots
global path planning algorithm, electric circuit based path
planning (ECPP) algorithm. ECPP will not provide a
collision free plan by accurately coordinating all the robot
moving nicely in a precise time schedule, but minimise
the possibilities that a traffic jam might happen. ECPP is
based on a roadmap created by the voronoi based parallel
thinning (VPT) algorithm, which we propose to make a
precise topological map from a given grid map. For each
single robot, it would get a path which may not be the best
solution as such, but this largely avoids the possibility of
traffic jams, so that the whole multi-robot system works
properly. The multi-robot system is assumed to be a
networked robotic system, the robots could communicate
with each other and each robot has its own start and goal
point. Every robot preserves a state list
robot
of all the
other robots. Each robot will find a path based on the
current state of list
robot
. As the list
robot
keeps updated,
the path will sometimes also be updated accordingly, while
the robot is on the way. It is also assumed that the global
path will be implemented by combining a certain local
path planner, while the robot is on the fly.
In section 2, a new approach of extracting the generalized
voronoi diagram (GVD) from the grid map is proposed.
Section 3 introduces the path planning with the GVD