
Trajectory Tracking of Unmanned Underwater Vehicles based on
Model Predictive Control in Two Dimension
WenYang GanˈDaqi Zhu and Bing Sun
Laboratory Of Underwater Vehicles and Intelligent Systems, Shanghai Maritime University, Haigang Avenue 1550,
Shanghai, 201306ˈChinaˈE-mail:dqzhu@shmtu.edu.cn
WenYang Gan
Shanghai Maritime University
Shanghai,China
t677hq23@qq.com
Daqi Zhu
Shanghai Maritime University
Shanghai,China
zdq367@yahoo.com.cn
Bing Sun
Shanghai Maritime University
Shanghai,China
Abstract—Trajectory tracking of unmanned
underwater vehicles (UUV) in two-dimension is
researched by analyzing and establishing a
two-dimension kinematic model of UUV. A new
approach of trajectory tracking is posed, which is
called model predictive control. Model predictive
control is used based on linear description of the
error model and it has to satisfy the control
constraints. The optimization problem of
minimizing the objective function is converted to the
a quadratic programming problem, which solves the
phenomenon of speed jump problem effectively. The
simulation experiments show that model predictive
control is valid and feasible in solving trajectory
tracking of UUV in two-dimension, when compared
with the method of backstepping.
Key words: Unmanned Underwater Vehicle,
trajectory tracking, model predictive control,
backstepping control
I. Introduction
The trajectory tracking control of UUV refers that
UUV track the reference trajectory in inertial
coordinate system. It starts from the given initial state
under the designed control laws, and achieves global
uniform asymptotic stability on the premise of meeting
the requirements of position error tracking [1]. The
research of UUV’s trajectory tracking is a very
challenging research field.
So far there are some related research reported [2].
The main methods of UUV’s trajectory tracking used
commonly are: traditional Proportion Integration
Differentiation (PID) control method, Backstepping
control method, and neural networks control method,
etc. PID control [3-4] is the most commonly used
control algorithm in industrial control, which is widely
used in single-input single-output linear control system,
but control effect is not ideal for multi-input
multi-output systems and nonlinear systems.
Backstepping control[5-6] is a common control
strategy, which is widely used in tracking control of
mobile robots, and is also suitable for UUV’s control.
However, the disadvantages of backstepping control
algorithm are also evident. When the reference
trajectory has mutations, the larger state error will
generate a larger control speed, which will cause speed
jump. Neural network has a non-linear, self-learning
function, which has made the neural network control
method
[7-8] widely used in the kinematic control
system. It does not establish a precise model of UUV.
Non-linear UUV can be fitted by neural network. But
this control method has not only difficulties in
212
978-1-4673-8644-9/16/$31.00
c
2016 IEEE