Вы находитесь на странице: 1из 4

A Robot Navigation Method Based on RFID and QR

Code in the Warehouse

Wang Xiao-Long, Wu Chun-Fu, Li Guo-Dong, Chen Qing-Xie


College of Mechanical and Electrical Engineering
Longyan University
Longyan, China
wxlwxwxlwx12@163.com

Abstract—With the rapid development of e-commerce,


warehousing and logistics industry has entered a period of rapid II. WAREHOUSE SPACE STRUCTURE LAYOUT
development. New logistics demand is not different from AND TASK FLOW DESIGNE BASED ON MINIMUM COST
traditional commodity transport, with the features of small
batches, variety, high speed and low cost. Flexible warehousing A. Warehouse Space Structure Layout
technology based on small mobile handing robots is rising Using mobile robot to carry goods, the workers are only
rapidly. The article proposes a flexible warehouse space model
responsible for the inventory of goods, greatly reducing the
based on RFID and QR code on the basis of analyzing e-
amount of labor. As robot running space needs navigation and
commerce logistics. In the warehouse space model, the article
analyses logistics processes, considering time and traffic conflict,
positioning, so the type of automated storage space structure
and uses the A* algorithm for robot path planning. The article layout is very different from traditional warehouse space [2].
verifies the feasibility of navigation based RFID and QR code The article gives an automated warehouse space model shown
through simulation. in figure 1.

Keywords—flexible warehousing, mobile robot, RFID, QR code,


path planning, A* algorithm

I. INTRODUCTION
With the rapid development of e-commerce, warehousing
and logistics industry has entered a period of rapid
development. E-commerce raises new requirements for
logistics and transport that are different form traditional
commodity transport. New features mainly include small
quantities of goods, more merchandise categories, fast logistics
and transportation speed, short logistics and warehouse
construction cycle and lower logistics cost. China’s
warehousing and logistics automation level is low and logistics
workers spend sixty to seventy percent of the time picking up
and fetching goods[1]. So improving the efficiency of pickup
and turnover has become one of the key issues in the current
logistics industry. Logistics is characterized by the distribution
of goods from large transshipment centers to local transit
centers. Large logistics companies put a lot of energy to build
many large-scale logistics and transit center with advanced
logistics equipment. However, local transshipment centers
across the country have a low level of automation and have no
conditions to purchase expensive three-dimensional warehouse
equipment. Traditional equipment is difficult to meet the
demand, so the Amazon Kiva System abroad, Geek+ and
Quicktron at home as the representative of the mobile robot Fig 1. Automatic Warehouse Space Model
and flexible warehousing is on the rise.
As shown in figure 1, the warehouse space is divided into
replenishment area, inventory area and shipping area and

‹,((( 
regardless of the path or the location are represented by unified complete the task and choose the robot of minimum cost to
grids. In figure 1, the automated warehouse makes ten goods perform the task. If some robots have no task, the logistics
units into a group surrounded by the paths. The traditional dispatch system chooses the robot from the nearest location to
warehouse does not have to set the one-way channel necessary. undertake the delivery task. If all the robots have tasks, it is
But this article adopts the horizontal channel as the one-way necessary to consider the association and the number of robot
channel in order to speed up the transport of goods and reduce tasks. There is a short time and short distance evaluation
the traffic collision problem occurred by the robotic path system, through p, q weighted sum to get the total cost.
planning. The automated warehouse marked the direction of
one-way channel. When the robots need to ship or pick up III. RFID AND QR CODE POSITION INFORMATION FUSION TO
goods quickly , they can run on the one-way channel and don’t ACHIEVE RAPID AND HIGH PRECISION POSITIONG OF THE ROBOT
worry about the collision.
The automated warehouse model in figure 1 shows that each
B. Task Flow Design based on Minimum Cost grid center has a location label and each goods shelf also has a
location label that helps the robot to quickly locate and record
The main tasks of automation logistics warehouse are shelf information. Kiva system uses QR two-dimensional code
goods storage, sorting of goods, goods out of the warehouse label to achieve position function. As shown in figure 3, QR
and goods adjustment and so on. A warehouse is required to code label has position detection blocks to help identify system
carry out a number of tasks at the same time, while a number to quickly locate the label and accommodate a certain amount
of mobile robots are used for handing goods so there is a need of information through the error correction code to ensure the
for a reasonable task allocation process. A task allocation correct identification.
process based on a minimum total cost is shown in the figure 2.

Cos
t3
Co
st

Fig 3. QR Code and RFID Tag


The robot can visually decode the QR code to obtain the
current position information and correct the attitude angle of
the robot by resolving the angle of QR code in the visual image.
Although QR code can be used to achieve the precise
positioning of the robot, because of high-speed movement of
the robot will cause the motion blur of camera, with the rising
speed, the robot may decode failure. When the QR code
appears in the camera image of the robot, the image is only
used to solve the robot attitude angle, so the image accuracy
Fig 2. Task Allocation Process of Minimum Total Cost can be lower. So it is possible to solve the attitude angle only
when the robot moves at high speed, and the positioning
All types of tasks in automated warehouse ultimately lead to information is provided by the high-speed rfid passive tags.
robots carrying different lengths of paths. In general, the longer Figure 3 shows the rfid tag is omnidirectional label read and
the robot’s handling path is, the more time it takes. But when written speed up to 50~100 times per second. With the robot
the robot is jammed, the waiting time may cause the short path airborne large area antenna, the robot obtains the location
to consume more time. Therefore, the path planning algorithm information from rfid tag when it is moving quickly in the
needs to consider the path length and the traffic jam time both. channel. The robot corrects its own attitude with the external
The paper sets the task set is T={T1,T2,…Tn}, where n is the positioning information, and the robot sends the internal
number of tasks, the robot set is R={R1,R2…Rm}, where m is odometer data back to the server through the wireless network
the number of robots. After the tasks are assigned, the subset of to provide the necessary time position node information for the
tasks for each robot is RiTj. So it is easy to draw a conclusion automated warehouse backstage. The server avoids the
N O collision of the robot at the same time through the scheduling
that 7 = ¦ 5 L7M . As shown in figure2, put all the tasks to
L = M = 
algorithm.

be assigned into the task pool. Every time the task pool only
pops up a task to allocate according to the cost which the robot


IV. ROBOT SCHEDULING OPTIMIZATION BASED ON A* time factors. The pre-designed rules are used to reduce the
ALGORITHM computational complexity of the A* algorithm and improve
the path planning speed.
A. A* Algorithm Overview This paper assumes that the robot is moving at a constant
A* algorithm is a typical heuristic algorithm in speed, ignoring its acceleration and deceleration process. The
artificial intelligence field, which is one of the commonly robot model is a differential drive type, so when the direction
used robot path planning algorithms [3]. Robot path changes the robot needs to turn and turning grid factor D is
planning is the process of finding a path from the initial larger than the straight grid. The automatic warehouse is
state to the target state. Since there are many branches shown in figure 1 and the x-axis is one-way channel which its
available, all the paths from the beginning to the end is direction is marked by the arrow. Based on the above settings,
called the state space. But in the path state space through the warehouse path planning is as follows:
exhaustive search path, the efficiency is too low, too much 1) The road priority is higher than the robot priotiy rule.
work, and ultimately may not find the required path. A* X-axis unidirectional channeal has a higher level than the Y-
algorithm as a representative of the heuristic search axis channel. When the robot move from Y-axis channel to the
algorithm searches the path in the state space. After every X-channel, the robot shulde wait until no other robot pasees.
search, the algorithm evaluates the location of the search 2) The robot determines the traffic priority rule accorfing
and gets the best position and then searches further from tuo the goods priority.Picking up robot has the highest priority.
this location until the target. Using the above approach, the The robot priority for the goods return location is medium
algorithm can find the path efficiently. The A* algorithm priority. Empty robot’s priority is the lowest.
quickly sets the target direction by including the heuristic 3) The turning grid cost coefficient D is greater than the
information through the evaluation function. The general straight grid factor. When using the A* algorithm to find a
form of the evaluation function is shown in equation 1. path on a grip map, the cost of different types of grid is
different. As the X-axis channel is unidirectinal, when the
f(n)=g(n)+h(n) (1) open list of the A* algorithm is set, the grids that don’t
conform to the direction is excluded from the list of avaliable
grids.
In equation 1, f(n) is the valuation function from the initial 4) When there are multiple robot path planning, the
point through node n to the target point, which represents the
total cost of the node n to the target point, and g(n) is the true planning system plans the robot path one by one.Using A*
cost of the path from the starting point to the node n, and h(n) algorithm to plan multiple paths at the same time, it is easy to
is the heuristic term, which indicates the estimated cost of fail due to path conflict planning [5][6]. When planning a path,
node n to the end point. It is obvious that g(n) is the actual the system calculates the time the robot reaches each grid and
existence, h(n) is the estimation function, called heuristic checks whether there is a time point confilict with the
function, where h(n) <= h*(n), h*(n) is the actual minimum previouly planned robot path. If there is a time point conflict,
path cost.
planning system sets different priorities according to the rules,
so that the robot with the higher priority passes first.
B. Improvement of A* Algorithm by Using Road Rules and 5) When the robot stops due to the avoidance of obstacles,
Time Optimization the planning system will re-plan the avaliable path of the
The robot can only move in four directions that are left, robot.
right, up and down in automated warehouse. As shown in Based on the above rules, the complexity of the planning
equation 2, the A* algorithm uses the Manhattan distance as a algorithm can be greatly reduced. So the traditional A*
valuation function [4]. The calculation of the Manhattan algorithm can be used to the multi-robot path planning of the
distance in the grid map is shown in Equation 3, where D is automated warehouse, the process flow is shown in figure 4.
the cost of the walking grid and the goal represents the target Firstly, according to the rule of planning path in the
node. warehouse, update every grid’s status. Grids that do not meet
the traffic rules are set directly as obstacles. Then, add start
D(i,j) = |Xi – Xj | + |Yi – Yj | (2) node into the open list and check if the open list is empty. If
the open list is empty, finish path planning, if not follow the
next step. Find the minimum F value in the open list and
h(n) = D*(abs(n.x – goal.x) + abs(n.y – goal.y)) (3) establish the node. Check the node if it is the end node and if
it is, finish path planning. If the node is not the end node,
The traditional A* algorithm derives the path from the task calculate all possible adjacent nodes node. Update nodes’ G
starting point to the end point according to the evaluation value, H value and F value and ignore the node that is already
function f(n) by algorithm iteration. However, in the actual in the closed list. If the node is not in the open list, put it in
automated warehouse, A* algorithm has the problems of slow and calculate new F value. If the F value in the open list
computation speed and multi-robot path planning. In this smaller than before, the current node is used as the parent
paper, A* algorithm is improved by combining road rules and node to recalculate F value and if not, put the node into the


closed list. Keep the loop until there are no nodes in the open through the common path. Path planning avoids traffic
list or have found the end. At last find the path in reverse. collisions.

Fig 5. Three Robot Paths planning


This paper gives the layout of automated warehouse model,
the fusion of RFID and QR code information to provide robot
positioning data, and finally proposed a combination of traffic
rules of path planning process. However the default robot in
this paper is constant velocity, path planning needs to further
consider the actual acceleration and deceleration process,
which in the future work to further study.
ACKNOWLEDGMENT
The research is supported by the Guide project of the
Training Plan of Longyan University for young teachers
(LQ2014006), the project on the integration of industry,
education and research of Longyan University (LC2014003),
Fig 4. A* Algorithm Path Planning Process Fujian Science and Technology Department (2016H0026), the
Natural Science Foundation of Fujian Province (no.2015J012
V. EXPERIMENTAL CONCLUSION AND CONCLUSIONS 68), the Science and Technology Project of Fujian Education
Department (JK2014049).
The paper uses the Python to write the path planning
background software to carry on the simulation experiment. REFERENCES
The simulation map uses the grip warehouse model shown in [1] Zou S. The Present and Future Warehouse robot[J]. Logistics
figure 1. Experiment results show that if there is no pre-set Engineering and Management, 2013, 35(6): 171-172.
traffic rules, directly using the A* algorithm to find the path, [2] van den Berg J㸪Overmars M㸬 Roadmap㸫based motion planning in
the robot planning path failure is very likely. If the software dynamic environments㹙J㹛㸬 IEEE Transactions on Ro-botics㸪2005,
performs pre-set traffic rules, the path planning success rate is 21( 5) : 885㸫897㸬
high and the path planning will not fail due to collision [3] Wang D. Indoor Mobile-robot Path Planning based on Improved A*
Algorithim [J]. Tsinghua Science and Technology, 2012, 52(8):1085-
problems. When the robot path extends to the one-way 1089.
channel, the planning software doesn’t consider the collision [4] Shen B, Yu N, Liu Ji. Intelligent Scheduling and Path Planning of
problem. The actual operation of a one-way channel of the Warehouse Mobile Robots[J]. CAAI Transactions on Intelligent
robot encountered, only need to stop the barrier to wait. One- Systems, 2014, 9(6): 659-664.
way channel is not allowed to retrograde, so it can be seen as [5] Qiu G, Ou Ch, Han T. Map-building and Scheduling Alrorithm for
Multi-agv Intelligent Warehouse Systems[J]. Industrial Control
an obstacle when the path is planned. Computer, 2016,29(12):117-119.
As shown in figure 5, the robots R1, R2, R3, move to the [6] Shi Y, Zhou L, Wang J. A Balanced Heristic Auction Method for Multi-
goods location A, B, C respectively and carry goods to E. The Robot Task Allocation of Intelligent Warehouse[J]. Control and
paths from X11 to X4 are same and the robots are queued Decision, 2014, 10(15): 280-285



Вам также может понравиться