ACloud-AssistedRegionMonitoringStrategyofMobileRobotin ...downloads.hindawi.com/journals/misy/2019/5846232.pdf · In smart agricultural systems, the macroinformation sensing by adopting

Research ArticleA Cloud-Assisted Region Monitoring Strategy of Mobile Robot inSmart Greenhouse

Xiaomin Li 1 Zhiyu Ma 1 Xuan Chu1 and Yongxin Liu2

1College of Mechanical and Electrical Engineering Zhongkai University of Agriculture and Engineering Guangzhou China2Department of Electrical Computer Software and Systems Engineering Embry-Riddle Aeronautical University OklahomaOK USA

Correspondence should be addressed to Zhiyu Ma mazhiyu2002hotmailcom

Received 9 May 2019 Accepted 22 July 2019 Published 1 October 2019

Academic Editor Marco Anisetti

Copyright copy 2019 Xiaomin Li et al is is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

In smart agricultural systems the macroinformation sensing by adopting a mobile robot with multiple types of sensors is a keystep for sustainable development of agriculture Also in a region monitoring strategy that meets the real-scene requirementsoptimal operation of mobile robots is necessary In this paper a cloud-assisted region monitoring strategy of mobile robots in asmart greenhouse is presented First a hybrid framework that contains a cloud a wireless network and mobile multisensor robotsis deployed to monitor a wide-region greenhouse en a novel strategy that contains two phases is designed to ensure validregion monitoring and meet the time constraints of a mobile sensing robot In the first phase candidate region monitoring pointsare selected using the improved virtual forces In the second phase a moving path for the mobile node is calculated based onEuclidean distance Subsequently the applicability of the proposed strategy is verified by the greenhouse test system everification results show that the proposed algorithm has better performance than the conventional methods e results alsodemonstrate that by applying the proposed algorithm the number of monitoring points and time consumption can reduce whilethe valid monitoring region area is enlarged

1 Introduction

As it is well-known agriculture represents the basis ofmany other fields With the development of urbaniza-tion a number of countries have faced the problem oflimited farmland and urban population growth [1 2]e model society trends require the agriculture toimprove the yield per unit of land erefore precisionagriculture or smart agriculture has become a hot re-search topic for the industry and academia as a potentialsolution which might solve the mentioned problemSmart greenhouse is one of the most typical represen-tatives of the smart agriculture that should meet thegrowing requirement for food To improve the auto-mation and efficiency of a greenhouse the solar-poweredand the water- and energy-self-sufficient method [3 4]and the information and communication technologies

(ICTs) such as wireless sensor networks (WSNs) andintelligent control [5ndash8] are merged into the agriculturalsystem erefore a smart greenhouse system denotes anexcellent option to realize intelligent and precisionagriculture

e data collection especially macrodata collectionrepresents a precondition for the realization of a smartgreenhouseWith the development of information collectiontechnologies [9ndash11] the ICTs have been gradually in-troduced in a greenhouse and the agricultural WSNs havebecome classical applications for information gatheringeagricultural WSNs have obtained a great progress [12ndash14]However on the one hand current WSNs are mostlycomposed of static nodes and constraint of energy chargingso that the monitoring systems lack the flexibility andlifespan [15] On the other hand with the increase ingreenhouse size the monitoring region is becoming wider

HindawiMobile Information SystemsVolume 2019 Article ID 5846232 10 pageshttpsdoiorg10115520195846232

[16] Accordingly traditional WSNs are not suitable forgreenhouse region monitoring

Many studies have been devoted to solving the datacollection in greenhouse namely more and more WSNnodes have been deployed to meet the trends but that hasincreased the system cost significantly especially for a large-region greenhouse [17 18] Besides many studies have beenfocused on reducing energy consumption but all proposedmethods still have a constraint of limited energy of WSNs[19 20] Accordingly to increase the mobility and flexibilitysimultaneously mobile elements (a mobile robot or a mobilesensor node) have been introduced to the systeme relatedstudies have also researched on this topic from differentperspectives such as path planning and obstacle avoidance[21ndash23] However the greenhouse region monitoring stillfaces the following problems (1) how to construct a moreefficient framework using the cloud computing and mobileelements to improve the data collection and (2) how tochoose a better moving path for the greenhouse regionmonitoring with latency and energy constraints

Recently the cyber-physical system (CPS) has beenmoreand more introduced into the agriculture Motivated by theadvantages of the CPS especially cloud technologies and thefeatures of greenhouse region monitoring we present anovel solution to solve these mentioned problems

e contributions of this work can be summarized asfollows

(1) A novel data collection system based on cloudcomputing wireless networks and mobile nodes isdeveloped

(2) To meet the requirements for region monitoringratio and time consumption a cloud-assisted regionmonitoring strategy of mobile robot is designed

(3) To evaluate the framework and strategy a real ex-periment is conducted and the obtained results arediscussed

e remainder of this paper is organized as followsSection 2 introduces related work about information col-lection and region monitoring in agriculture Section 3 givesthe cloud-assisted information collection frameworks forsmart greenhouse Section 4 proposes the method in-formation collection Section 5 evaluates the proposed al-gorithm experimentally and discusses the obtained resultsSection 6 concludes the paper

2 Related Work

21 Information Collection in Agriculture As it is well-known data collection represents the basis of smartgreenhouse erefore numerous studies have been focusedon the data collection method In [24] to achieve scientificcultivation and reduce management costs a wireless sensornetwork architecture was adopted in the vegetable green-house to monitor environmental parameters In [25] toschedule irrigation the authors proposed a multiple sensorsystem to collect soil moisture and weather parametersen using the collected data the equipment was driven for

precision irrigation In [26] a wireless sensor network wasadopted to detect the presence of snails by sensing theenvironmental data e WSN motes were also designedwith real application and deployed in the data collection Asthe climate is a significant factor for crop yield and diseasesin [27] the authors realized a distributed monitoring systemfor a greenhouse using a WSN and analyzed the raw data toexplore the condition of growing crops In [28] the authorsdesigned a farmland information collection system con-taining a portable farmland information collection device aremote server and a mobile phone APP Furthermore thissystem was implemented and the information about col-lected air temperature and humidity chlorophyll and otherfarmland-related information was collected In [29] byequipping wireless sensors on tractors and other equipmenta delay disruption tolerant information collection systemwas designed and implemented in the farmland Differentinformation collection method and applications were pro-posed in the aforementioned studies However the proposedsolutions are mostly based on static wireless sensor networksand are not suitable for region monitoring in a smartgreenhouse

22 Region Monitoring Methods With the development ofinformation analysis many studies demonstrated thatregion monitoring could be adopted to monitor the ob-jective targets [30 31] In some studies the increasingnumber of mobile nodes was used for region monitoringIn [32] the authors presented a distribute frameworkcontaining a server and clients running different strategiesto monitor the region which was impossible to install fixedcameras based on the crowdsourcing and mobile deviceIn [33] mobile nodes in the wireless sensor network wereused to monitor the uncovered region en an algorithmwas used to find out an optimal path of a mobile node tocover the target region Recently drones which are the typeof unmanned aerial vehicles (UAVs) have made progressin the field thus providing a good solution for the regionmonitoring problem In [34] a UAV was used as a mobileinformation collector for region monitoring Also theregion to be monitored was divided into cells e authorassigned to each UAV the cells to visit and optimized thepath of each UAV by considering fairness In addition in[35] the authors used drones to monitor large regionseffectively Also by combining the skyline query a newstrategy was proposed to search a more suitable UAVflying path e abovementioned studies provide a ref-erence and new solutions for the greenhouse regionmonitoring problem However to meet region monitor-ing the greenhouse characteristics and low latency ofregion monitoring need to be considered

3 Cloud-Assisted Data CollectionFramework for Smart Greenhouse

In the section a region monitoring architecture in agreenhouse is presented the working process of the

2 Mobile Information Systems

framework is explained and the mobile robot isintroduced

31 Information Collection Framework for GreenhouseWith the development of Internet of ings (IoT) andcloud computing technologies the ICTs have been provedas effective methods for information collection Since thetraditional frameworks for greenhouse lack the mobilitycomputing resources and collection the previous solutionsare not gradually suitable for the new requirement Toimprove the flexibility and scalability of a smart green-house the wireless network cloud computing softwareapplications and greenhouse equipment are merged intoan improved framework e proposed framework forsmart system sensing consists of three layers the networklayer the cloud layer and the application layer and the datasensing field as shown in Figure 1 Meanwhile all thecomponents are linked by either wired or wireless networksand internet of things (IoT) We solely agree that big dataanalysis and artificial intelligence are the future trend inagriculture e distributed cloud services as far as we cansee are the most flexible and cost-efficient infrastructurefor big data analysis in smart agriculture It is known thatcloud has become an important part for the smart agri-culture On the one hand the proposed algorithm easilygets the global information for achieving the optimal re-sults by using cloud technologies On the other hand bigdata is not in conflict with the cloud as cloud is the basicplatform of big data analysis

In the application layer different applications areconstructed to meet the real requirements such as that forgreenhouse management data visualization and mobileterminal operations Data storage and computing task areperformed in the cloud layer In network layers the datatransfer is conducted via either wired or wireless net-works Meanwhile the data sensing filed is used to collectdata from different sensors of greenhouse such as tem-perature humidity light level image of crop and otherparameters

32 Mobile Region Monitoring Robot in the FrameworkIn the paper the focus is on region monitoring in agreenhouse using a mobile robot erefore a brief reviewof a mobile region-monitoring robot (MAMR) is providedA MAMR is equipped with a wireless radio module apower unit and different sensors such camera andhumiture sensor e application layer or cloud serversconduct the region monitoring task namely the cloudserver decomposes the task into multiple steps and de-termines the key orders to the mobile nodes according tothe greenhouse global information en the MAMRfinishes the region monitoring following the cloud issuingorders Finally when a mobile node moves along theplanning path it transmits the information to the cloudserver through the IoT system

4 Mobile Robot Region MonitoringStrategy in Greenhouse

In the section we first provide the problem formulationen we present a region monitoring strategy for mobilerobots which contains two phases candidate point selectingand moving path searching

41 Problem Formulation For the sake of simplicity weconsider a greenhouse as a flat rectangular and let Λ sub R2 bea flat rectangular region of a greenhouse Also we assumethat in a certain time period there are n regions to bemonitored and let Δ δ1 δ2 δn1113864 1113865(δi sub Δ Δ sub Λ) be theset of these regions Meanwhile we assume that eachmonitoring region δi has a trajectory and starting pointx0(δi) A mobile robot equipped with different sensors isnavigated through the defined monitoring region along thetrajectory Assume that a MAMR monitors region δi atcertain points which are on a trajectory trδi

ere are m(δi)

monitoring points on the trajectory trδi Let R be the sensing

radius In this work a MAMR monitoring region is denotedas a circle e set of monitoring circles is expressed asC(δi) c1 c2 cm(δi)

1113966 1113967 and the set of circlesrsquo center lo-cations is expressed as X(δi) x1(δi) x2(δi) 1113864

xm(δi)(δi) and it is arranged according to the distance from

the starting point of a monitoring region δi

Definition 1 Real monitoring area (RMA) let ck(δi) be amonitoring region where k isin m(δi) e real monitoringarea α(δi) of δi can be defined as follows

α δi( 1113857 S c1cup c2 cup middot middot middot cm δi( )1113874 1113875cap δi1113876 1113877 (1)

where S[middot] denotes the function for getting a real monitoringarea of a mobile robot Let ck cap δi be a valid monitoringregion of monitoring circle and monitoring region δi us(1) can be simplified as follows

α δi( 1113857 S cupkisinm δi( )

ck cap δi( 1113857⎡⎣ ⎤⎦ (2)

Furthermore let Ts(xk(δi)) be the sensing time of aMAMR at the point xk(δi) of the monitoring region δi usTs(X(δi)) denotes the total sensing time of δi and it is given by

Ts X δi( 1113857( 1113857 1113944

kisinm δi( )

Ts xk δi( 1113857( 1113857(3)

Assume that sensing time at any point is constant τ so(3) can be simplified as

Ts X δi( 1113857( 1113857 m δi( 1113857 middot τ (4)

erefore the total sensing time of Δ Ts(Δ) is given by

Ts(Δ) m δ1( 1113857 + m δ2( 1113857 + middot middot middot + m δn( 1113857( 1113857τ

τ 1113944iisinn

m δi( 1113857(5)

Mobile Information Systems 3

According to the above relations let T(δi) be the timeconsumption which includes the moving time and moni-toring time Tmove(X(δi)) of a monitoring region δi and it isexpressed as

T δi( 1113857 τ middot m δi( 1113857 + Tmove_point X δi( 1113857( 1113857 (6)

Moreover let Tmove_region(Δ) be the moving timethrough a monitoring region In summary the total timeconsumption of all the monitoring regions can be calcu-lated as

T Tmove_region(Δ) + 1113944iisinn

T δi( 1113857

Tmove_region(Δ) + τ 1113944iisinn

m δi( 1113857 + 1113944iisinn

Tmove_point X δi( 1113857( 1113857

(7)

Definition 2 Region monitoring ratio (RMR) let α(δi) andc(δi) be the real area of monitoring and the area of δirespectively e region monitoring rate of a monitoringregion δi denoted as ε(δi) is a fraction of the realmonitoring area within the area of δi and it can beexpressed as

ε δi( 1113857 α δi( 1113857

c δi( 1113857 (8)

erefore the region monitoring ratio of a wholegreenhouse can be expressed as

E(Δ) 1113936iα δi( 1113857

1113936ic δi( 1113857 (9)

Assume that there is a constant area in each δi and then(9) can be approximated as

E(Δ) 1113936iα δi( 1113857

n middot c δi( 11138571n

1113944iisinn

α δi( 1113857

c δi( 11138571n

1113944iisinn

ε δi( 1113857 (10)

Definition 3 Covering path (CP) CP is the moving path of aMAMR that covers all the monitoring regions Let P be theCP set which can be formulated as

P pi pi

1113868111386811138681113868 ΥΔ1113966 1113967 (11)

where piΥΔ denotes that pi covers all the regionsEach monitoring area tries to get its RMR that should be

larger than the preset constant us the objective of thispaper is to search a path from P to meet the above re-quirement of RMR while meeting the followingrequirements

E(Δ)geE

TleTdeadline

0lt ε δi( 1113857le 1

(12)

where E ε1 ε2 εn1113864 1113865 is the preset value of the RMR andTdeadline is the deadline of the monitoring region

Access point Access point

RouterRouter

Mobile robot Sensor node

Greenhouse device

Irrigation systemData sensing field

Network layer

Cloud layer

Application layer

Computingservers

Data servers

Mobile terminalData visualizationDifferent APPsManagement

Figure 1 Information collection architecture for greenhouse


42 Proposed Method for Mobile Robot-Based RegionMonitoring Solving the problem defined by (12) is chal-lenging due to the following reasons e time con-sumption is related to the number of monitoring pointsand path length In this section we divide the problem (12)into two subproblems selecting candidate monitoringpoints and an optimal region monitoring moving pathFirst an improved virtual force is employed for selectingcandidate points en to meet the monitoring timeconsumption requirement we develop an algorithm basedon simulated annealing for determining an optimalmoving path

421 Selecting Strategy for Candidate Monitoring PointsIn the traditional virtual force model the force directions arerandom It is known that the model is not suitable for agreenhouse Meanwhile the monitoring points are on themonitoring trajectory Accordingly we use a single directionalong with the trajectory

Meanwhile a coverage coefficient β(δi) of a monitoringregion δi is introduced to the model and it is formulated as

β δi( 1113857 1 ε δi( 1113857ge εi

0 otherwise1113896 (13)

us according to Hookersquos law the virtual force isformulated as

fpq β δi( 1113857k middot |d| dlq le 2R

minus β δi( 1113857k middot |d| dlq gt 2R

⎧⎨

⎩ (14)

where k denotes the virtual force coefficient and|d| |2R minus dlp| So the total virtual force between themonitoring points (l q) can be expressed as

Fq δi( 1113857rarr

1113944lisinP

flq

rarr (15)

Furthermore the total virtual force of monitoring pointset P can be formulated as

F δi( 1113857rarr

12

1113944QisinP

Fq δi( 1113857rarr

(16)

According to the above relations we minimize thevirtual force to determine the minimal number of moni-toring points

Min F δi( 1113857rarr

1113874 1113875

subject to p isin δi

ε δi( 1113857ge εi

(17)

To solve the problem given by (17) we use the followingequations to update a new location Xq(next) according tothe current position Xq(current)

Xq(next) Xq(next) |F|leFTH

Xq(next) + dmove middot e(minus(1F)) |F|leFTH

⎧⎨

⎩

(18)

where dmove is the moving distance of each iteration and FTHis the threshold value of virtual force e proposed strategyfor selecting strategy for candidate monitoring points isshown in Algorithm 1 and it can be explained as followsfirst in the initialization step the preset values of the var-iables are set en the virtual force is calculated accordingto the corresponding equations If the virtual force is largerthan the threshold value a new location of monitoringpoints is updated and if the virtual force is less than thethreshold value the monitoring location would stopupdating Also the RMR is calculated and judged whether itsvalue meets the requirement e iteration steps repeat untilall the constraints are met

422 Hybrid Solution for Optimal Moving Path To obtainthe shortest moving path we divide the strategies ofsearching the optimal moving path into two stages inter-monitoring region and outermonitoring region In the firststage a greedy strategy is used to get the shortest movingpath In the second stage the problem is first classified as atraveling salesman problem (TSP) and then a simulatedannealing is used to solve it

423 Stage I Intermonitoring Region After Algorithm 1 isfinished in the monitoring region δi the number ofmonitoring points and their locations can be obtained emoving path of a mobile robot must contain the statingpoints so the extended monitoring point set is defined asfollows

P+ δi( 1113857 P δi( 1113857cupO δi( 1113857 (19)

where O(δi) denotes the starting point of a trajectory trδi

Furthermore the location set that corresponds to the ex-tended monitoring point set can be expressed as

X+ δi( 1113857 0 x1 δi( 1113857 x2 δi( 1113857 xm δi( ) δi( 11138571113882 1113883 (20)

where 0 denotes the location of O(δi) Let NBPP(δi) be theneighboring point of p en the Euclidean distance be-tween p and any of its neighboring point q can be expressedas

dpq

xp minus xq1113872 11138732

1113970

q isin NBPP δi( 1113857( 1113857 (21)

As all the monitoring points are on the trajectory to getan optimal moving path within the monitoring region agreedy strategy is designed based on the neighboring dis-tance namely in each iteration the nearest neighboringpoint is selected as the next moving point e process stepsare given in Algorithm 2 First using (19) and (20) theextended monitoring point set and the corresponding lo-cation set are determined respectively Next the neigh-boring point set is created based on the distance value enthe neighboring point at the minimal distance is selected asthe next moving point Finally the algorithm returns themoving path points and their locations of the monitoringregion


424 Stage II Outer Monitoring Region In stage I themoving path in the intermonitoring region is solved Nextin the second stage we focus on obtaining the shortestmoving path in the outer monitoring region

In the previous stage the monitoring points in eachmonitoring subregion are obtained Since the monitoringpointsrsquo distances are smaller than the distance between thesubregions the barycenter method is used for monitoringpointsrsquo locations normalization Specifically in a subregionδi the barycenter can be formulated as

X δi( 1113857 1113936qisinP δi( )xp

P δi( 11138571113868111386811138681113868

1113868111386811138681113868 (22)

So the barycenter of all the subregions can be expressedas

X X δ1( 1113857 X δ2( 1113857 X δn( 11138571113966 1113967 (23)

With regard to the above discussion the outer shortestmoving path selection can be transferred into a TSPproblem To solve the TSP problem the simulated annealingis employed e process of determining an optimal movingpath in the outer monitoring region is given in Algorithm 3In Algorithm 3 first the barycenter of each subregion iscalculated then simulated annealing is used to obtain theshortest moving path and finally the algorithm returns theresults to the mobile robot

5 Experimental Results

e proposed algorithm was evaluated experimentally eobtained experimental results are presented and discussed interms of different metrics Besides the proposed algorithm iscompared with the traditional methods regarding the fol-lowing three aspects the number of monitoring pointsaverage covering area and time consumption

Input εi R FTH max_iteration dmovingOutput m(δi) X(δi)

Initialize m(δi) 1 X(δi) Rrepeat

while (hltmax_interaction)if F(δi)

rarr(h)geF(δi)

rarr(h minus 1)

for l m(δi)

Calculate virtual force F(δi)rarr

if |F(δi)rarr

|leFTHupdate location according to (18)if Xq(next) isin δi

Xq(next)⟵Xq(next)else

Xq(next)⟵X(boundary)

calculate the total forcem(δi)⟵m(δi) + 1

until ε(δi)ge εi

Return m(δi) X(δi)

ALGORITHM 1 Selecting candidate monitoring points based on improved virtual force

Input P(δi) X(δi) x1(δi) x2(δi) xm(δi)(δi)1113966 1113967

Output Pmove(δi)

Initialization MP(δi) O

P+(δi)⟵P(δi)cupO(δi)

X+(δi)⟵ 0 x1(δi) x2(δi) xm(δi)(δi)1113966 1113967

for f1|P+(δi)|

Search neighboring points of p+f(δi)

Create neighboring points set NBPf(δi)

for q1NBPf(δi)

Calculate the distance according (21)Creating distance set of all neighboring pointsSelecting the minimal distance point P

f

minMP(δi) MP(δi) + P

f

minNBP

Pf

min(δi) minus Pf(δi)

return MP(δi)

ALGORITHM 2 Moving path in the intermonitoring region


51 Experimental Settings A real experiment was performedusing a mobile robot in the greenhouse that was planted withdifferent crops as shown in Figure 2 A hybrid method thatcontains the ultrasonic obstacle avoidance and automatictracking (different colors) is adopted to cope with the robotobstacle avoidance e mobile robot was equipped withmany sensors to collect air temperature humidity and lightintensity and a camera

In order to demonstrate that the proposed planningalgorithm performs correctly in a real environment anexperimental setup presented in Figure 2 was constructede experimental settings were as follows e greenhousearea was Λ 40 times 50m2 and it contained 10 subregionswith the same size of 2times 4m2 e sensing radius R was 1me other experimental parameters are given in Table 1 eproposed strategy was applied for several times and theaverage results were calculated

As already mentioned there are few works that addressthe regionmonitoring problemerefore we compared ourmethod called the CARM with the following commonmethods (1) the SSRL method which selects the samenumber of the monitoring points in a subregion as theCARM but with a with random location (2) the FMPSmethod which selects fixed monitoring location at the samedistance along the trajectory In the experiments the dis-tance of points was equal to the sensing radius

52 Result Analysis

521 Number of Monitoring Points e average number ofmonitoring points of different methods corresponding totheir best performances in the real experiment is presentedin Figure 3(a) namely the average number of monitoringpoints at the RMR of 20 70 and 90 of differentmethods is presented in Figure 3(a) In Figure 3(a) it canbe seen that the average number of monitoring pointsincreased with the increase in the RMR for all themethods However the CAMR outperformed the othermethods in all the three cases which was due to theoptimization virtual force algorithm of the CAMRmethodwhich selected a more optimal number of monitoring

points and their locations e FMPS achieved betterperformance than the SSRL because monitoring pointswere evenly distributed along the trajectory At the RMRof 70 the CARM had almost 20 and 10 smalleraverage number of monitoring points than the SSRL andthe FMPS respectively

522 Average Coverage Area e average coverage area ofdifferent methods is presented in Figure 3(b) In Figure 3(b)it can be seen that for all the values of the number ofmonitoring points the average coverage area of the pro-posed method was the best because the CARM selected themost optimal number of monitoring points and their lo-cations in order to get a large coverage area As displayed inFigure 3(b) the SSRL obtained poor average coverage areaperformance because of randommonitoring points selectionwhich was employed in this method Moreover by in-creasing the number of monitoring points the averagecoverage area performance enhanced for all the methods Inaddition when the number of points was 3 the coverage areaof the CARM was larger than 35m2 Besides the CARMcould adapt to a different number of monitoring points andhad the best average coverage area performance among allthe tested methods

523 Time Consumption e total sensing time con-sumption performance for each subregion of differentmethods at the RMR of 70 is presented in Figure 4(a) Ingeneral the total sensing time of eachmethod increased withthe increase in the sensing time of a single point e CARMachieved the best performance regarding this metric inother words the CARM took less time to finish the mon-itoring task that the other two methods When the RMR was70 and the sensing time of a single point was 5 s the totalsensing time consumption of the CARM was less than 15 swhile the other methods had longer sensing time especiallythe SSRL because this method selected more points formonitoring

e total time consumption performance at a differentmoving speed for the whole monitoring region of different


Output Pmove(Δ)Initialization Pmove(Δ) empty X emptyfor i1n

X(δi)⟵1113936qisinP(δi)xp|P(δi)|

X⟵X + X(δi)

SA initializationsimulated annealingRepeatMetropolis sampling algorithmCalculate the corresponding target function value

untile current solution Xmove is an optimal solution

Return Xmove

ALGORITHM 3 Moving path in the outer monitoring region


methods at the RMR of 70 is presented in Figure 4(b)Generally as the robot moving speed increased the totaltime consumption reduced erefore the total time con-sumption of all the methods decreased with the increase inthe robot moving speed Similarly the CARM achieved thebest performance in terms of the total time consumptionwhich means that the CARM finished the monitoring taskfaster than the other methods which was because the op-timal algorithm was used to select the number of monitoringpoints and their locations Meanwhile at the RMR of 70

the robot moving speed of 05ms and the sensing time of asingle point of 5 s the CARM had 5 and 3 shorter totalsensing time consumption compared to the SSRL and FMPSmethods respectively

524 Cost Evaluation It is known that cost evaluation isan essential part of the practical application Obviouslyaccording to Figures 3 and 4 the proposal method CARMcan reduce the number of monitoring points for the same

Figure 2 e experimental scene

Table 1 Experimental parameter

Parameter ValueVirtual force threshold 0Number of monitoring subregions 5Sensing radius 1mSensing time of a single point 05sim5 sRobot moving speed 005sim05msDeadline of the monitoring region 100sim500 sRegion monitoring ratio 40sim50Length of subregion trajectory 200m

0

10

20

30

40

20 70 90Region monitoring ratio ()

CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40

1 2 3Number of monitoring points

CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)

Figure 3 (a) Number of monitoring points and (b) coverage area of different methods


coverage area and decrease the time consumption com-paring the traditional methods SSRL and FMPS In otherwords the proposed strategy can get lower maintenancecost Indeed the presented framework because of needingto build a cloud is more expensive than the traditionalregion monitoring framework such as WSN However thepublic clouds can significantly reduce costs as publicclouds present a very low price (0023sim004 dollarGB fordata storage and 0017sim0027 dollarGB RAM of AmazonWeb Services) furthermore some companies offer freecloud services Besides private farms in China could notafford to hire professional technicians for maintenance ofprivate servers so compared with using dedicated serversmaking use of the public cloud is the most feasiblesolution

6 Conclusions

An efficient region monitoring of plants in a smart greenhouseis important but challenging In this paper we propose a novelcloud-assisted region monitoring strategy of mobile robotsbased for a smart greenhouse to increase the monitoring regionsize and reduce the time consumption First an informationcollection framework based onmobile robots wireless networkand cloud computing is introduced en the region moni-toring model of a mobile robot with cloud computing isconstructed in accordance with the greenhouse features andcloud resources Moreover based on the model two strategiesare proposed to select the candidate monitoring points byadopting the improved virtual force and a hybrid solution forthe optimal moving pathe proposed algorithm is verified byreal experiments in the greenhouse e experimental resultsdemonstrate that the proposed algorithm can provide moreefficient regionmonitoring of greenhouse than the conventionalmethods

Data Availability

e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

e authors declare there are no conflicts of interest re-garding the publication of this paper

Acknowledgments

is paper was supported by the Science and TechnologyPlanning Project of Guangdong Province China (no2014A020208132) and Science and Technology Project ofGuangzhou China (no 201607010018)

References

[1] K L Skog and M Steinnes ldquoHow do centrality populationgrowth and urban sprawl impact farmland conversion inNorwayrdquo Land Use Policy vol 59 pp 185ndash196 2016

[2] W Song B C Pijanowski and A Tayyebi ldquoUrban expansionand its consumption of high-quality farmland in BeijingChinardquo Ecological Indicators vol 54 no 54 pp 60ndash70 2015

[3] E Farrell M I Hassan R A Tufa et al ldquoReverse electro-dialysis powered greenhouse concept for water- and energy-self-sufficient agriculturerdquo Applied Energy vol 187pp 390ndash409 2017

[4] S Suryawanshi S Ramasamy S Umashankar andP Sanjeevikumar ldquoDesign and implementation of solar-powered low-cost model for greenhouse systemrdquo in Advancesin Smart Grid and Renewable Energy vol 435 pp 357ndash365Springer Berlin Germany 2018

[5] G Ping P Dusadeeringsikul and S Y Nof ldquoAgriculturalcyber physical system collaboration for greenhouse stressmanagementrdquo Computers amp Electronics in Agriculturevol 150 pp 439ndash454 2018

[6] X Bai Z Wang L Sheng and Z Wang ldquoReliable data fusionof hierarchical wireless sensor networks with asynchronousmeasurement for greenhouse monitoringrdquo IEEE Transactionson Control Systems Technology vol 27 no 3 pp 1036ndash10462018

[7] P J Kia A T Far M Omid R Alimardani and L NaderlooldquoIntelligent control based fuzzy logic for automation ofgreenhouse irrigation system and evaluation in relation to

0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)

Figure 4 Time consumption of different methods (a) total sensing time (b) total time consumption


conventional systemsrdquoWorld Applied Sciences Journal vol 6no 1 pp 16ndash23 2009

[8] A Somov D Shadrin I Fastovets et al ldquoPervasive agri-culture IoT-enabled greenhouse for plant growth controlrdquoIEEE Pervasive Computing vol 17 no 4 pp 65ndash75 2018

[9] X Li D Li J Wan C Liu and M Imran ldquoAdaptivetransmission optimization in SDN-based industrial internetof things with edge computingrdquo IEEE Internet of ingsJournal vol 5 no 3 pp 1351ndash1360 2018

[10] J Wan J Li Q Hua A Celesti and Z Wang ldquoIntelligentequipment design assisted by cognitive internet of things andindustrial big datardquo Neural Computing and Applicationspp 1ndash10 2018

[11] J Wan J Li M Imran and D Li ldquoA blockchain-basedsolution for enhancing security and privacy in smart factoryrdquoIEEE Transactions on Industrial Informatics vol 15 no 6pp 3652ndash3660 2019

[12] L D Pedrosa P Melo R M Rocha and R Neves ldquoA flexibleapproach to WSN development and deploymentrdquo In-ternational Journal of Sensor Networks vol 6 no 34pp 199ndash211 2009

[13] X-M Guo X-T Yang M-X Chen M Li and Y-A WangldquoA model with leaf area index and apple size parameters for24GHz radio propagation in apple orchardsrdquo PrecisionAgriculture vol 16 no 2 pp 180ndash200 2015

[14] D Upadhyay A K Dubey and P S ilagam ldquoApplicationof non-linear Gaussian regression-based adaptive clocksynchronization technique for wireless sensor network inagriculturerdquo IEEE Sensors Journal vol 18 no 10 pp 4328ndash4335 2018

[15] O Elijah T A Rahman I Orikumhi C Y Leow andM N Hindia ldquoAn overview of internet of things (IoT) anddata analytics in agriculture benefits and challengesrdquo IEEEInternet of ings Journal vol 5 no 5 pp 3758ndash3773 2018

[16] M Kang X-R Fan J Hua H Wang X Wang andF-Y Wang ldquoManaging traditional solar greenhouse withcpss a just-for-fit philosophyrdquo IEEE Transactions on Cy-bernetics vol 48 no 12 pp 3371ndash3380 2018

[17] X Bai Z Wang L Zou and F E Alsaadi ldquoCollaborativefusion estimation over wireless sensor networks for moni-toring CO2 concentration in a greenhouserdquo InformationFusion vol 42 pp 119ndash126 2018

[18] D-H Park B-J Kang K-R Cho et al ldquoA study ongreenhouse automatic control system based on wireless sensornetworkrdquo Wireless Personal Communications vol 56 no 1pp 117ndash130 2011

[19] Y Zhang H Gao S Cheng and J Li ldquoAn efficient EH-WSNenergy management mechanismrdquo Tsinghua Science andTechnology vol 23 no 4 pp 406ndash418 2018

[20] W Xu Y Zhang Q Shi and X Wang ldquoEnergy managementand cross layer optimization for wireless sensor networkpowered by heterogeneous energy sourcesrdquo IEEE Trans-actions on Wireless Communications vol 14 no 5pp 2814ndash2826 2015

[21] C-H Ou and W-L He ldquoPath planning algorithm for mobileanchor-based localization in wireless sensor networksrdquo IEEESensors Journal vol 13 no 2 pp 466ndash475 2013

[22] A Alomari W Phillips N Aslam and F Comeau ldquoSwarmintelligence optimization techniques for obstacle-avoidancemobility-assisted localization in wireless sensor networksrdquoIEEE Access vol 6 pp 22368ndash22385 2018

[23] P Zhou C Wang and Y Yang ldquoStatic and mobile targetk-coverage in wireless rechargeable sensor networksrdquo IEEE

Transactions on Mobile Computing vol 18 no 10 pp 2430ndash2445 2019

[24] M Srbinovska C Gavrovski V Dimcev A Krkoleva andV Borozan ldquoEnvironmental parameters monitoring in pre-cision agriculture using wireless sensor networksrdquo Journal ofCleaner Production vol 88 pp 297ndash307 2015

[25] R Sui and J Baggard ldquoWireless sensor network for moni-toring soil moisture and weather conditionsrdquo Applied Engi-neering in Agriculture vol 31 no 2 pp 193ndash200 2015

[26] D Garcia-Lesta D Cabello E Ferro P Lopez andV M Brea ldquoWireless sensor network with perpetual motesfor terrestrial snail activity monitoringrdquo IEEE Sensors Journalvol 17 no 15 pp 5008ndash5015 2017

[27] K P Ferentinos N Katsoulas A Tzounis T Bartzanas andC Kittas ldquoWireless sensor networks for greenhouse climateand plant condition assessmentrdquo Biosystems Engineeringvol 153 pp 70ndash81 2017

[28] J Zhang J Hu L Huang Z Zhang and Y Ma ldquoA portablefarmland information collection system with multiple sen-sorsrdquo Sensors vol 16 no 10 Article ID 1762 2016

[29] H Ochiai H Ishizuka Y Kawakami and H Esaki ldquoA DTN-based sensor data gathering for agricultural applicationsrdquoIEEE Sensors Journal vol 11 no 11 pp 2861ndash2868 2011

[30] D Ehrlich J E Estes J Scepan and K C McGwire ldquoCroparea monitoring within an advanced agricultural informationsystemrdquo Geocarto International vol 9 no 4 pp 31ndash42 1994

[31] T-Y Lin H A Santoso K-R Wu and G-L Wang ldquoEn-hanced deployment algorithms for heterogeneous directionalmobile sensors in a bounded monitoring areardquo IEEETransactions onMobile Computing vol 16 no 3 pp 744ndash7582017

[32] A Guerriero F Giuliani and D O Nitti ldquoCrowdsourcingand mobile device for wide areas monitoringrdquo in Proceedingsof the 2015 IEEE Workshop on Environmental Energy andStructural Monitoring Systems (EESMS) pp 29ndash32 TrentoItaly July 2015

[33] C Zygowski and A Jaekel ldquoPath Planning for maximizingarea coverage of mobile nodes in wireless sensor networksrdquo inProceedings of the 2018 IEEE 9th Annual Information Tech-nology Electronics and Mobile Communication Conference(IEMCON) pp 12ndash17 Vancouver BC Canada November2018

[34] I Khoufi P Minet and N Achir ldquoUnmanned aerial vehiclespath planning for area monitoringrdquo in Proceedings of the 2016International Conference on Performance Evaluation andModeling inWired andWireless Networks (PEMWN) pp 1ndash5Paris France November 2016

[35] Y Chen and Y Chen ldquoPath planning in large area monitoringby dronesrdquo in Proceedings of the 2018 Tenth InternationalConference on Advanced Computational Intelligence (ICACI)pp 295ndash299 Xiamen China March 2018


Computer Games Technology

International Journal of

Hindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Journal ofEngineeringVolume 2018

Advances in

FuzzySystems


Volume 2018


ReconfigurableComputing



Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

thinspArtificial Intelligence

Hindawiwwwhindawicom Volumethinsp2018


Civil EngineeringAdvances in


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia


Biomedical Imaging



Engineering Mathematics


RoboticsJournal of



Computational Intelligence and Neuroscience


Mathematical Problems in Engineering

Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018


Human-ComputerInteraction

Advances in


Scientic Programming

Submit your manuscripts atwwwhindawicom

[16] Accordingly traditional WSNs are not suitable forgreenhouse region monitoring

Many studies have been devoted to solving the datacollection in greenhouse namely more and more WSNnodes have been deployed to meet the trends but that hasincreased the system cost significantly especially for a large-region greenhouse [17 18] Besides many studies have beenfocused on reducing energy consumption but all proposedmethods still have a constraint of limited energy of WSNs[19 20] Accordingly to increase the mobility and flexibilitysimultaneously mobile elements (a mobile robot or a mobilesensor node) have been introduced to the systeme relatedstudies have also researched on this topic from differentperspectives such as path planning and obstacle avoidance[21ndash23] However the greenhouse region monitoring stillfaces the following problems (1) how to construct a moreefficient framework using the cloud computing and mobileelements to improve the data collection and (2) how tochoose a better moving path for the greenhouse regionmonitoring with latency and energy constraints

Recently the cyber-physical system (CPS) has beenmoreand more introduced into the agriculture Motivated by theadvantages of the CPS especially cloud technologies and thefeatures of greenhouse region monitoring we present anovel solution to solve these mentioned problems

e contributions of this work can be summarized asfollows

(1) A novel data collection system based on cloudcomputing wireless networks and mobile nodes isdeveloped

(2) To meet the requirements for region monitoringratio and time consumption a cloud-assisted regionmonitoring strategy of mobile robot is designed

(3) To evaluate the framework and strategy a real ex-periment is conducted and the obtained results arediscussed

e remainder of this paper is organized as followsSection 2 introduces related work about information col-lection and region monitoring in agriculture Section 3 givesthe cloud-assisted information collection frameworks forsmart greenhouse Section 4 proposes the method in-formation collection Section 5 evaluates the proposed al-gorithm experimentally and discusses the obtained resultsSection 6 concludes the paper

2 Related Work

21 Information Collection in Agriculture As it is well-known data collection represents the basis of smartgreenhouse erefore numerous studies have been focusedon the data collection method In [24] to achieve scientificcultivation and reduce management costs a wireless sensornetwork architecture was adopted in the vegetable green-house to monitor environmental parameters In [25] toschedule irrigation the authors proposed a multiple sensorsystem to collect soil moisture and weather parametersen using the collected data the equipment was driven for

precision irrigation In [26] a wireless sensor network wasadopted to detect the presence of snails by sensing theenvironmental data e WSN motes were also designedwith real application and deployed in the data collection Asthe climate is a significant factor for crop yield and diseasesin [27] the authors realized a distributed monitoring systemfor a greenhouse using a WSN and analyzed the raw data toexplore the condition of growing crops In [28] the authorsdesigned a farmland information collection system con-taining a portable farmland information collection device aremote server and a mobile phone APP Furthermore thissystem was implemented and the information about col-lected air temperature and humidity chlorophyll and otherfarmland-related information was collected In [29] byequipping wireless sensors on tractors and other equipmenta delay disruption tolerant information collection systemwas designed and implemented in the farmland Differentinformation collection method and applications were pro-posed in the aforementioned studies However the proposedsolutions are mostly based on static wireless sensor networksand are not suitable for region monitoring in a smartgreenhouse

22 Region Monitoring Methods With the development ofinformation analysis many studies demonstrated thatregion monitoring could be adopted to monitor the ob-jective targets [30 31] In some studies the increasingnumber of mobile nodes was used for region monitoringIn [32] the authors presented a distribute frameworkcontaining a server and clients running different strategiesto monitor the region which was impossible to install fixedcameras based on the crowdsourcing and mobile deviceIn [33] mobile nodes in the wireless sensor network wereused to monitor the uncovered region en an algorithmwas used to find out an optimal path of a mobile node tocover the target region Recently drones which are the typeof unmanned aerial vehicles (UAVs) have made progressin the field thus providing a good solution for the regionmonitoring problem In [34] a UAV was used as a mobileinformation collector for region monitoring Also theregion to be monitored was divided into cells e authorassigned to each UAV the cells to visit and optimized thepath of each UAV by considering fairness In addition in[35] the authors used drones to monitor large regionseffectively Also by combining the skyline query a newstrategy was proposed to search a more suitable UAVflying path e abovementioned studies provide a ref-erence and new solutions for the greenhouse regionmonitoring problem However to meet region monitor-ing the greenhouse characteristics and low latency ofregion monitoring need to be considered

3 Cloud-Assisted Data CollectionFramework for Smart Greenhouse

In the section a region monitoring architecture in agreenhouse is presented the working process of the









ere are m(δi)










ck cap δi( 1113857⎡⎣ ⎤⎦ (2)


Ts X δi( 1113857( 1113857 1113944

kisinm δi( )

Ts xk δi( 1113857( 1113857(3)





τ 1113944iisinn

m δi( 1113857(5)






T δi( 1113857


m δi( 1113857 + 1113944iisinn

Tmove_point X δi( 1113857( 1113857

(7)


ε δi( 1113857 α δi( 1113857

c δi( 1113857 (8)


E(Δ) 1113936iα δi( 1113857

1113936ic δi( 1113857 (9)


E(Δ) 1113936iα δi( 1113857


1113944iisinn

α δi( 1113857

c δi( 11138571n

1113944iisinn

ε δi( 1113857 (10)


P pi pi

1113868111386811138681113868 ΥΔ1113966 1113967 (11)



E(Δ)geE

TleTdeadline

0lt ε δi( 1113857le 1

(12)



RouterRouter


Greenhouse device


Network layer

Cloud layer

Application layer

Computingservers

Data servers







β δi( 1113857 1 ε δi( 1113857ge εi

0 otherwise1113896 (13)




⎧⎨

⎩ (14)


Fq δi( 1113857rarr

1113944lisinP

flq

rarr (15)


F δi( 1113857rarr

12

1113944QisinP

Fq δi( 1113857rarr

(16)



1113874 1113875


ε δi( 1113857ge εi

(17)




⎧⎨

⎩

(18)




P+ δi( 1113857 P δi( 1113857cupO δi( 1113857 (19)





dpq

xp minus xq1113872 11138732

1113970

q isin NBPP δi( 1113857( 1113857 (21)





X δi( 1113857 1113936qisinP δi( )xp

P δi( 11138571113868111386811138681113868

1113868111386811138681113868 (22)


X X δ1( 1113857 X δ2( 1113857 X δn( 11138571113966 1113967 (23)







rarr(h)geF(δi)

rarr(h minus 1)

for l m(δi)


if |F(δi)rarr





until ε(δi)ge εi




Output Pmove(δi)




for f1|P+(δi)|



for q1NBPf(δi)


f


f

minNBP

Pf


return MP(δi)






52 Result Analysis









X⟵X + X(δi)



Return Xmove









0

10

20

30

40


CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40


CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)




6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in











ere are m(δi)










ck cap δi( 1113857⎡⎣ ⎤⎦ (2)


Ts X δi( 1113857( 1113857 1113944

kisinm δi( )

Ts xk δi( 1113857( 1113857(3)





τ 1113944iisinn

m δi( 1113857(5)






T δi( 1113857


m δi( 1113857 + 1113944iisinn

Tmove_point X δi( 1113857( 1113857

(7)


ε δi( 1113857 α δi( 1113857

c δi( 1113857 (8)


E(Δ) 1113936iα δi( 1113857

1113936ic δi( 1113857 (9)


E(Δ) 1113936iα δi( 1113857


1113944iisinn

α δi( 1113857

c δi( 11138571n

1113944iisinn

ε δi( 1113857 (10)


P pi pi

1113868111386811138681113868 ΥΔ1113966 1113967 (11)



E(Δ)geE

TleTdeadline

0lt ε δi( 1113857le 1

(12)



RouterRouter


Greenhouse device


Network layer

Cloud layer

Application layer

Computingservers

Data servers







β δi( 1113857 1 ε δi( 1113857ge εi

0 otherwise1113896 (13)




⎧⎨

⎩ (14)


Fq δi( 1113857rarr

1113944lisinP

flq

rarr (15)


F δi( 1113857rarr

12

1113944QisinP

Fq δi( 1113857rarr

(16)



1113874 1113875


ε δi( 1113857ge εi

(17)




⎧⎨

⎩

(18)




P+ δi( 1113857 P δi( 1113857cupO δi( 1113857 (19)





dpq

xp minus xq1113872 11138732

1113970

q isin NBPP δi( 1113857( 1113857 (21)





X δi( 1113857 1113936qisinP δi( )xp

P δi( 11138571113868111386811138681113868

1113868111386811138681113868 (22)


X X δ1( 1113857 X δ2( 1113857 X δn( 11138571113966 1113967 (23)







rarr(h)geF(δi)

rarr(h minus 1)

for l m(δi)


if |F(δi)rarr





until ε(δi)ge εi




Output Pmove(δi)




for f1|P+(δi)|



for q1NBPf(δi)


f


f

minNBP

Pf


return MP(δi)






52 Result Analysis









X⟵X + X(δi)



Return Xmove









0

10

20

30

40


CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40


CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)




6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in








T δi( 1113857


m δi( 1113857 + 1113944iisinn

Tmove_point X δi( 1113857( 1113857

(7)


ε δi( 1113857 α δi( 1113857

c δi( 1113857 (8)


E(Δ) 1113936iα δi( 1113857

1113936ic δi( 1113857 (9)


E(Δ) 1113936iα δi( 1113857


1113944iisinn

α δi( 1113857

c δi( 11138571n

1113944iisinn

ε δi( 1113857 (10)


P pi pi

1113868111386811138681113868 ΥΔ1113966 1113967 (11)



E(Δ)geE

TleTdeadline

0lt ε δi( 1113857le 1

(12)



RouterRouter


Greenhouse device


Network layer

Cloud layer

Application layer

Computingservers

Data servers







β δi( 1113857 1 ε δi( 1113857ge εi

0 otherwise1113896 (13)




⎧⎨

⎩ (14)


Fq δi( 1113857rarr

1113944lisinP

flq

rarr (15)


F δi( 1113857rarr

12

1113944QisinP

Fq δi( 1113857rarr

(16)



1113874 1113875


ε δi( 1113857ge εi

(17)




⎧⎨

⎩

(18)




P+ δi( 1113857 P δi( 1113857cupO δi( 1113857 (19)





dpq

xp minus xq1113872 11138732

1113970

q isin NBPP δi( 1113857( 1113857 (21)





X δi( 1113857 1113936qisinP δi( )xp

P δi( 11138571113868111386811138681113868

1113868111386811138681113868 (22)


X X δ1( 1113857 X δ2( 1113857 X δn( 11138571113966 1113967 (23)







rarr(h)geF(δi)

rarr(h minus 1)

for l m(δi)


if |F(δi)rarr





until ε(δi)ge εi




Output Pmove(δi)




for f1|P+(δi)|



for q1NBPf(δi)


f


f

minNBP

Pf


return MP(δi)






52 Result Analysis









X⟵X + X(δi)



Return Xmove









0

10

20

30

40


CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40


CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)




6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in







β δi( 1113857 1 ε δi( 1113857ge εi

0 otherwise1113896 (13)




⎧⎨

⎩ (14)


Fq δi( 1113857rarr

1113944lisinP

flq

rarr (15)


F δi( 1113857rarr

12

1113944QisinP

Fq δi( 1113857rarr

(16)



1113874 1113875


ε δi( 1113857ge εi

(17)




⎧⎨

⎩

(18)




P+ δi( 1113857 P δi( 1113857cupO δi( 1113857 (19)





dpq

xp minus xq1113872 11138732

1113970

q isin NBPP δi( 1113857( 1113857 (21)





X δi( 1113857 1113936qisinP δi( )xp

P δi( 11138571113868111386811138681113868

1113868111386811138681113868 (22)


X X δ1( 1113857 X δ2( 1113857 X δn( 11138571113966 1113967 (23)







rarr(h)geF(δi)

rarr(h minus 1)

for l m(δi)


if |F(δi)rarr





until ε(δi)ge εi




Output Pmove(δi)




for f1|P+(δi)|



for q1NBPf(δi)


f


f

minNBP

Pf


return MP(δi)






52 Result Analysis









X⟵X + X(δi)



Return Xmove









0

10

20

30

40


CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40


CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)




6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in






X δi( 1113857 1113936qisinP δi( )xp

P δi( 11138571113868111386811138681113868

1113868111386811138681113868 (22)


X X δ1( 1113857 X δ2( 1113857 X δn( 11138571113966 1113967 (23)







rarr(h)geF(δi)

rarr(h minus 1)

for l m(δi)


if |F(δi)rarr





until ε(δi)ge εi




Output Pmove(δi)




for f1|P+(δi)|



for q1NBPf(δi)


f


f

minNBP

Pf


return MP(δi)






52 Result Analysis









X⟵X + X(δi)



Return Xmove









0

10

20

30

40


CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40


CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)




6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in







52 Result Analysis









X⟵X + X(δi)



Return Xmove









0

10

20

30

40


CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40


CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)




6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in










0

10

20

30

40


CARMSSRLFMPS

Num

ber o

f mon

itori

ng p

oint

s

(a)

0

10

20

30

40


CARMSSRLFMPS

Cov

erag

e are

a (m

2 )

(b)




6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in





6 Conclusions


Data Availability




Acknowledgments


References








0

5

10

15

20

05 1 15 2 25 3 35 4 45 5Sensing time (s)

Tota

l sen

sing

time (

s)

CARMSSRLFMPS

(a)

10

30

50

70

90

110

005 01 015 02 025 03 035 04 045 05Moving speed (ms)

CARMSSRLFMPS

Tota

l tim

e con

sum

ptio

n (s

)

(b)







































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in








































Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in









Advances in

FuzzySystems


Volume 2018













Journal of

Journal of



Hindawi


Advances in

Multimedia


Biomedical Imaging





RoboticsJournal of









Volume 2018



Advances in




Documents

ACloud-AssistedRegionMonitoringStrategyofMobileRobotin ...downloads.hindawi.com/journals/misy/2019/5846232.pdf · In smart agricultural systems, the macroinformation sensing by adopting