1 INTRODUCTION
Multi-agent system, which consists of several agents, is an
important research area in artificial intelligence. Through
the communication and cooperation between the agents, the
problems of the transportation task which a single agent can
not complete can be solved by using the multi-robot
collaborative systems [1]. Compared with a single-agent
system, multi-agent system has many advantages, such as
completing multiple tasks in parallel, improving the
efficiency, better fault tolerance, robustness, economic
benefits and providing more solutions. Therefore, the path
planning and motion control of the multi-agent systems are
widely concerned in the control field [2-3].
The multi-robot systems are the typical multi-agent systems,
which can be utilized in the transportation field, especially
the limited small space and the dangerous area. Due to the
flexible routing capability, the multi-robot system can adapt
to different task, workload and environmental
requirements, which can greatly relieve human efforts from
various dull and dangerous task. The transportation task
requires the multi-robot system to achieve a certain
formation in the moving process and to keep formation
along a certain path, so that the transport tasks could be
completed collaboratively. Path planning and cooperative
formation control play an important role in the multi-robot
collaborative transportation system. Several approaches
have been proposed for the multi-robot transportation
control. A leader-follower formation control scheme was
proposed to move a box with three mobile robots [4].
Ground vehicles are divided into two categories, master
vehicles and slave vehicles, and a two-step strategy is
proposed to move box cooperatively [5]. Many other
methods such as behavior-based architecture, rule-based
action selection method and dynamic task allocation
architecture were also presented in [6-8]. However, the
This work was supported by National Nature Science Foundation of
China under Grant (
61273116),
Zhejiang Provincial Nature Science
Foundation under Grant (
LY15F030015) and
National High-tech R&D
Program Foundation under Grant (
2014AA041601-05).
problem of the optimal path which can satisfy complex
environment is not considered. The aim of the path
planning of the robots is to make the robots go from the
initial point to the target point and avoid obstacles by a
specific algorithm, meanwhile accomplish some tasks at
certain points [9]. Multiple mobile robots under the
condition of known forward route can be better applied in
practice, such as disaster relief, chemical plant security,
dangerous goods transport. A path planning method was
proposed and the point-to-point control of a robot was
achieved [10]. Some other path planning methods are
presented, such as artificial potential field algorithm, A*
algorithm and grid algorithm [11-13]. The
above-mentioned research mainly focuses on the
point-to-point path planning. Artificial potential field
algorithm is only utilized to solve locally optimal problem,
which cannot guarantee the optimal path in the case that
there exist multiple minima. In additon, it is difficult to
solve the path planning in the complex environment by
using the grid algorithm. If the task requirements are
complex, path planning will be more complex, such as
moving a batch of goods from the starting point to different
places many times. However, it is difficult to complete such
the complex task by using the above-mentioned path
planning method. A path planning algorithm based on
linear temporal logic, which uses mathematical language to
describe complex tasks, can be utilized to satisfy the
complex task requirement. The complex tasks are described
by using the linear temporal logic formulas. A task feasible
network topology combining the environment information
and the linear temporal logic formula are constructed. The
Dijkstra algorithm is utilized to search the optimal path in
the task feasible network topology so as to ensure the path
is optimal. Multi-robot collaborative transportation control
system is achieved with the target tracking algorithm. It can
find the optimal path satisfying environment and task
requirements, and achieve tracking control by tracking
algorithm. Therefore, complex tasks based on temporal
ordering are completed. The system can be used in some
dangerous and complex tasks, such as factory
transportation, dangerous goods transportation. Thus, it can
Research on Multi-Robot Collaborative Transportation Control System
Cheng Cheng
1
, Xin-Yi Yu
1
, Lin-Lin Ou
1
, Yong-Kui Guo
1
1. College of Information and Engineering, Zhejiang University of Technology, Zhejiang 310023
E-mail: linlinou@zjut.edu.cn
Abstract: The collaborative transportation problem of multi-robot system is studied to satisfy complex requirements
such as task, workload and environmental requirements. Firstly, based on the theory of linear temporal logic (LTL), the
global optimal path satisfying the requirements is determined. Then, the similar formation control is proposed on the
basis of the above optimal path in order to keep a certain formation. Additionally, a target tracking control algorithm is
used in order to control multiple mobile robots to transport goods along with the planned path. Finally, the provided
simulation results demonstrate the effectiveness of the multi-robot collaborative transportation control system.
Key Words: path planning, multi-agent system, formation control, collaborative transportation
4886
978-1-4673-9714-8/16/$31.00
c
2016 IEEE