A Long-distance Navigation Model for AUV Based
on Multiple Parameter Searching
Xingguang Peng and Kun Liu
∗
School of Marine Science and Technology
Northwestern Polytechnical University, Youyi West Rd. Xi’an China, 710072
∗
Corresponding author (E-mail: liukunkmz@126.com)
Abstract—This paper present an active search navigation
model for AUV used geophysical field information in natal
homing mission. We consider that the animals homing behavior
can be regarded as a ability of search for destination multi-
parameter geophysical field without any priori map. Here,a
hypothesis is proposed that natal homing can be generalized as
a multi-objective problem. The AUV stored the multi-parameter
of geophysical field before navigation. Then active searching for
those parameters likely animals taxis which can be represent by
the stress evolution algorithm. The results show that this model
enable the AUV implement long-range navigation without priori
map.
I. INTRODUCTION
Autonomous underwater vehicle (AUV) is an important
tool of human exploration and utilization of ocean[1]. Nav-
igation is one of the primary challenges in AUV research
today. The geophysical field, such as geomagnetic field and
gravity field, which has been confirmed as the most reliable
resource for AUV navigation[2]. Always, the geophysical field
navigation utilized the matching way as the representative
now, which using the measured geophysical field information
data related to the prior geophysical map matching approach
for AUV location and navigation purposes[3]. However, the
priori geophysical map is hardly to build and obtain underwa-
ter.
Natal homing is a pattern of behavior in which animals
migration from position to their birth place or home[4]. In this
process, those animals always across dozens or even thousands
of kilometers, to implement homing navigation. There have
some obvious difference between this navigation models with
traditional geophysical field navigation as follow:
(1) homing navigation based on active searching model
without priori map, so this navigation model can worked in
unknown environment;
(2) the navigation error not increase over work time and
distance;
(3) The method does not aim for positioning, but in order
to reach the target point.
How animals accomplish natal homing has remained an
enduring mystery of of animal behavior[5, 6]. Some re-
searchers argued that the turtles derive long-distance naviga-
tional information from the geomagnetic field by detecting the
intensity and inclination angle[7, 8].However, some scholars
considered that animal navigation use the multiple orientation
clues navigation[9]. However, the geophysical field have more
wealth of characteristic parameters and it can provide more
reliable information for navigation.
In this paper, we aim at these problems and refer to the
methods of long-distance navigation without aprioriinfor-
mation of AUV. AUV only measuring the current geophysical
field information, by through the active search object param-
eter reach the specified target point. This process similar as
the programme of animal natal homing. Firstly, we generalize
the natal homing navigation as a multi-objective convergence
problem. Then, we use multi-objective evolution algorithms
to deal with the multi-objective search problem. Moreover, a
series of simulations are performed to test the result between
present model with steepest descent method, which demon-
strate the effectiveness of our model for navigation.
The remainder of this paper is organized as follows. In
Section 2, the problem is formulated. We give a multi-objective
function to describe the progress of animal geomagnetic
navigation.In Section 3, we propose a bio-inspired navigation
model with stress evolution method. We follow the laws of
nature survival of the fittest to model of bio-inspired navigation
model. In Section 4, series of comparison experiments are
conducted. Finally, Section 5 presents some conclusions and
suggests the further work to continue this research.
II. P
ROBLEM FORMULATION
The geophysical field can perceive as a mixture that in-
cluded variety of parameters, which can be represented as:
C = 𝐶
1
, ..., 𝐶
𝑛
(1)
Where 𝐶
1
, ⋅⋅⋅ ,𝐶
𝑛
are the parameters of geophysical field,
they may be the part or all of the geophysical field parameters.
Each of parameter has own distribution. The purpose of na-
tal homing navigation is to allow multiple parameters converge
to their target values contain by the navigation path, those can
be represent as the following formula:
{
min 𝐹 (C,𝑘)=(𝑓
1
(𝐶
1
,𝑘), ⋅⋅⋅ ,𝑓
𝑛
(𝐵𝐶
𝑛
,𝑘))
𝑠.𝑡. 𝑔(𝑆
𝑘
, C
𝑘
, C
𝑇
) ≤ 0
(2)
Where 𝐶
𝑘
is the AUV current location geophysical field
parameters; 𝐶
𝑇
is the target point geophysical field param-
eters; 𝑆 is the AUV movement path; 𝑔 is the constraint
condition.