Coverage Path Planning and Room Segmentation in Indoor Environments using the Constriction Decomposition Method
Loading...
Date
2017-08-28
Authors
Brown, Stanley
Journal Title
Journal ISSN
Volume Title
Publisher
University of Waterloo
Abstract
The task of complete coverage path planning in complex 2D environments is a classic NP-
Hard problem that has been an active research topic for well over 30 years. A common
approach to solving coverage problem in such environments is to partition, or segment, the
target environment into a set of cells that have some property that allows any given cell to
be covered in an optimal or near optimal manner. If each cell of an environment is visited
and covered by some agent, then the entire environment is said to be covered.
This work proposes a novel segmentation method, called the Constriction Decomposition Method (CDM), that works by locating constriction points in indoor, 2D environment
and then partitioning the environment based on the constriction points. When the CDM
is applied to 2D maps of office or laboratory environments, the CDM produces a segmentation that closely resembles a room based decomposition.
Once the environment has been
decomposed into regions, this work demonstrates that each room can be covered using a
simple coverage path planning algorithm that exploits the fact that the resulting cells do
not contain any constriction points. The lack of constriction points in each region means
that each region, or room, can be completely covered using a series of contour following
paths followed by a series of back and forth motions. Once a set of coverage paths are pro-
duced for each cell, a tour between all path is found using a heuristic Traveling Salesman
Problem (TSP) solver.
The proposed segmentation and coverage path planning methods are tested on a set of
15 indoor environments that are derived from a set of floor plans corresponding to five office
and seven laboratory environments. The quality of the segmentation produced by the CDM
is directly compared to existing methods on a qualitative and quantitative basis using a
series of metrics proposed by other authors. The set of coverage paths for each environment
are compared to existing work based on the ratio between the total path length and the
ratio between the inter-sector path and the total coverage path length. Based on these
metrics it is demonstrated that the CDM and the CDM coverage path planner produces
both superior segmentations and coverage plans in 2D indoor environments.
Description
Keywords
Coverage Path Planning, Straight Skeleton, Exact Decomposition, Room Segmentation