3.1. A
∗
algorithm
𝐴
∗
algorithm is the most effective direct search algorithm in the
static environment. It combines heuristic searching and searching based
on the shortest path (
Duchoň et al.
,
2014
). In the
𝐴
∗
algorithm, the
search area is generally divided into small squares, with each square
representing a node. The
𝐴
∗
algorithm uses an evaluation function:
𝑓
(
𝑛
) =
𝑔
(
𝑛
) +
ℎ
(
𝑛
)
, which provides guidances for the selection of nodes
in an open list. In the evaluation function,
𝑔
(
𝑛
)
is the actual cost of
node
𝑛
found so far from the initial node, and the cost estimate
ℎ
(
𝑛
)
is the Manhattan distance from node
𝑛
to the target node (
Yao et al.
,
2010
). As shown in
Fig. 3
, the
𝐴
∗
algorithm calculates the
𝑓
value of
the surrounding nodes and selects the node with the smallest
𝑓
value
as the next traversal node until the target node is found.
Due to the particularity of the marine environment,
Carroll et al.
(
1992
) considered the maximum operation depth of AUV and current
information in the heuristic function when they used the
𝐴
∗
algorithm
for path planning. To shorten the search time,
Szczerba et al.
(
2000
)
proposed a sparse
𝐴
∗
algorithm using the maximum turning angle and
the maximum path length as constraints for the basic
𝐴
∗
algorithm.
Yan et al.
(
2012a
) proposed an
𝐴
∗
algorithm based on circle search.
In their method, a virtual terrain is used to map the 3D space into
a 2D search area, which can meet the constraints of the change rate
of the AUV’s pitch angle and heading angle. In addition, the search
step size and search direction of the algorithm are flexible, so that
AUV can quickly get out of the threat and reach the destination. Based
on a sparse
𝐴
∗
algorithm,
Chen et al.
(
2012
) constructed the search
space of obstacles by randomly distributing points and combined with
a visibility inspection method under the constraint of maximum turning
radius to make AUV path smoother. According to the anisotropy of
the current flow,
Yan et al.
(
2012b
) expanded the sub nodes of
𝐴
∗
algorithm in a specific direction, which simplifies the spatial range
and greatly shortens the path planning time.
Wang and Pang
(
2019
)
applied the
𝐴
∗
search algorithm to a chemical plume tracking task with
AUV. Specifically, they get the source location according to the partially
observable Markov decision process (
Jiu et al.
,
2019
), and search the
shortest path of chemical source with the
Do'stlaringiz bilan baham: |