With the increasing size of wildfires (National
Interagency Fire Center [21]) a need for effec-
tive ways of monitoring and containing the
fires spread has emerged. The use of un-
manned aerial vehicles (UAVs) for optical mon-
itoring of a spreading wildfire is appropriate
due to e.g., the longevity of missions and the
hazardous environment.
With the increasing size of fires more UAVs
are needed to effectively monitor an area, lead-
ing to more potential conflicts between UAVs,
thereby requiring the need for communication
if a UAV were to deviate from its planned
route. However there are many situations
where communication between UAVs is not
practical or even impossible. This project there-
fore explores zoning methods that will allow
each agent to have navigational freedom while
not needing to communicate in order to avoid
conflicts.
This problem is formulated as a 2-stage
stochastic linear programming problem, that
reflects the zoning and flight stage, where
the environmental uncertainty during flight
should be considered when constructing zones.
This is solved using two zoning methods, in-
spired by the literature (Khemakhem et al.
[18]) as well as a novel routing-based cluster-
ing method proposed by the authors. These are
then compared using cumulative route scores
for routes generated in the resulting zones.
Their applicability for dynamic environments
is further tested by considering how often each
of the zoning solutions routes can be optimally
updated while respecting zone boundaries.
Lastly, the aforementioned zoning methods are
compared with a traditional routing approach,
in order to gauge performance degradation
due to zone restriction