[IEEE 2011 6th International Conference on Broadband and Biomedical Communications (IB2Com) -...

5
Unmanned Vehicle-Aided Automated Meter Reading Gurkan Tuna #1 , V. Cagri Gungor ‡2 , Kayhan Gulez *3 # Trakya University, Department of Computer Programming, Edirne, TURKEY 1 [email protected] Bahçeşehir University, Faculty of Engineering, Department of Computer Engineering, Istanbul, TURKEY 2 [email protected] *Yildiz Technical University, Electrical-Electronics Eng. Faculty, Control and Automation Eng. Dept., Istanbul, TURKEY 3 [email protected] Abstract— This paper presents a novel approach of using unmanned vehicles for Automated Meter Reading (AMR) applications in rural areas where there are a few consumers scattered around a wide area. The proposed system does not require a fixed network infrastructure to transfer data to a central database, since data collection is carried out by unmanned vehicles. Compared to traditional meter reading systems, the use of unmanned vehicles for AMR brings several advantages, such as low cost operation, flexibility, and online system management. However, the realization of these potential gains directly depends on reliable communication capabilities of the deployed system and successful navigation of unmanned vehicles. Overall, in this paper, the design principles and challenges of using unmanned vehicles for AMR applications in rural areas have been presented. Also, the communication architecture of the proposed system has been explained and comparative simulation studies have been performed in terms of energy efficiency and navigation accuracy. Index Terms—Automated meter reading, smart grid, unmanned vehicles, navigation and localization, wireless sensor networks. I. INTRODUCTION Automatically collecting consumption, diagnostic, and status data from metering devices for billing, troubleshooting, and analyzing are called Automatic Meter Reading (AMR). AMR brings cost advantages to utility providers. Also, timely information from AMR systems help utility providers perform system analysis and better control production of electric energy, gas usage, or water consumption. AMR systems not only send usage information to utility providers, they can also be used to remotely turn a service off or on at a residence, and identify problems. AMR systems can be broadly classified into four categories as follows: Touch-based AMR systems: A utility personnel who carries a data collection device collects the readings from meters by touching or placing the read probe. Walk-by AMR systems: A utility personnel who carries a handheld computer with a built-in receiver/transceiver collects meter readings from AMR meters. Drive-by AMR systems: A utility personnel drives the vehicle while the reading device in the vehicle automatically collects the readings. Fixed Network AMR systems: In these systems, a permanent network is installed to collect meter readings without periodic trips to each physical location. Various communication technologies are used in fixed network AMR systems. Generally, the communication channel of AMR systems consists of remote and local communication channels. Remote communication channels include GSM, GPRS, CDMA, fiber networks, and PSTN [3]. Local communication channels include broadband power line communication (BPL), Narrowband power line communication (PLC), RS485 bus, short-range wireless network, and CATV network [3]. Also, an existing meter can be retrofitted with a small battery-operated RF module containing a radio and a microcontroller for reading the metering data and wirelessly transmitting the data to a central database. Despite its operational benefits, the realization of an efficient AMR system requires investments during implementation and maintenance phases [1]. Communication is an important part of AMR systems and a great portion of the investment is spent on communication infrastructure besides the cost of metering devices [1], [2]. In this paper, we propose an alternative AMR system for rural areas where there are a few consumers scattered around a wide area and considering the return of investment, utility providers might be reluctant to invest in. The proposed system does not require a fixed network infrastructure to transfer data to a central database since data collection is to be carried out by an unmanned vehicle. The proposed unmanned vehicle- aided AMR is an alternative to drive-by AMR and can be integrated to an existing AMR system. In this system, an unmanned vehicle has to navigate and localize itself using a preloaded city map, and collect metering data. For an unmanned vehicle or robot, navigation means its ability to determine its own position in its reference frame and then to plan a path towards some target location. For navigation, a Proceedings of the 6th International Conference on Broadband Communications & Biomedical Applications, November 21 - 24, 2011, Melbourne, Australia 289 IB2COM 2011

Transcript of [IEEE 2011 6th International Conference on Broadband and Biomedical Communications (IB2Com) -...

Page 1: [IEEE 2011 6th International Conference on Broadband and Biomedical Communications (IB2Com) - Melbourne, Australia (2011.11.21-2011.11.24)] 7th International Conference on Broadband

Unmanned Vehicle-Aided Automated Meter Reading

Gurkan Tuna#1

, V. Cagri Gungor‡2

, Kayhan Gulez*3

#Trakya University, Department of Computer Programming, Edirne, TURKEY

[email protected]

‡Bahçeşehir University, Faculty of Engineering,

Department of Computer Engineering, Istanbul, TURKEY

[email protected]

*Yildiz Technical University, Electrical-Electronics Eng. Faculty,

Control and Automation Eng. Dept., Istanbul, TURKEY [email protected]

Abstract— This paper presents a novel approach of using

unmanned vehicles for Automated Meter Reading (AMR)

applications in rural areas where there are a few consumers

scattered around a wide area. The proposed system does not

require a fixed network infrastructure to transfer data to a

central database, since data collection is carried out by unmanned

vehicles. Compared to traditional meter reading systems, the use

of unmanned vehicles for AMR brings several advantages, such

as low cost operation, flexibility, and online system management.

However, the realization of these potential gains directly depends

on reliable communication capabilities of the deployed system and

successful navigation of unmanned vehicles. Overall, in this

paper, the design principles and challenges of using unmanned

vehicles for AMR applications in rural areas have been presented.

Also, the communication architecture of the proposed system has

been explained and comparative simulation studies have been

performed in terms of energy efficiency and navigation accuracy.

Index Terms—Automated meter reading, smart grid,

unmanned vehicles, navigation and localization, wireless sensor

networks.

I. INTRODUCTION

Automatically collecting consumption, diagnostic, and

status data from metering devices for billing, troubleshooting,

and analyzing are called Automatic Meter Reading (AMR).

AMR brings cost advantages to utility providers. Also, timely

information from AMR systems help utility providers perform

system analysis and better control production of electric

energy, gas usage, or water consumption. AMR systems not

only send usage information to utility providers, they can also

be used to remotely turn a service off or on at a residence, and

identify problems. AMR systems can be broadly classified into

four categories as follows:

• Touch-based AMR systems: A utility personnel who

carries a data collection device collects the readings

from meters by touching or placing the read probe.

• Walk-by AMR systems: A utility personnel who

carries a handheld computer with a built-in

receiver/transceiver collects meter readings from

AMR meters.

• Drive-by AMR systems: A utility personnel drives the

vehicle while the reading device in the vehicle

automatically collects the readings.

• Fixed Network AMR systems: In these systems, a

permanent network is installed to collect meter

readings without periodic trips to each physical

location.

Various communication technologies are used in fixed

network AMR systems. Generally, the communication channel

of AMR systems consists of remote and local communication

channels. Remote communication channels include GSM,

GPRS, CDMA, fiber networks, and PSTN [3]. Local

communication channels include broadband power line

communication (BPL), Narrowband power line

communication (PLC), RS485 bus, short-range wireless

network, and CATV network [3]. Also, an existing meter can

be retrofitted with a small battery-operated RF module

containing a radio and a microcontroller for reading the

metering data and wirelessly transmitting the data to a central

database. Despite its operational benefits, the realization of an

efficient AMR system requires investments during

implementation and maintenance phases [1]. Communication

is an important part of AMR systems and a great portion of the

investment is spent on communication infrastructure besides

the cost of metering devices [1], [2].

In this paper, we propose an alternative AMR system for

rural areas where there are a few consumers scattered around a

wide area and considering the return of investment, utility

providers might be reluctant to invest in. The proposed system

does not require a fixed network infrastructure to transfer data

to a central database since data collection is to be carried out

by an unmanned vehicle. The proposed unmanned vehicle-

aided AMR is an alternative to drive-by AMR and can be

integrated to an existing AMR system. In this system, an

unmanned vehicle has to navigate and localize itself using a

preloaded city map, and collect metering data. For an

unmanned vehicle or robot, navigation means its ability to

determine its own position in its reference frame and then to

plan a path towards some target location. For navigation, a

Proceedings of the 6th International Conference on Broadband Communications & Biomedical Applications, November 21 - 24, 2011, Melbourne, Australia

289 IB2COM 2011

Page 2: [IEEE 2011 6th International Conference on Broadband and Biomedical Communications (IB2Com) - Melbourne, Australia (2011.11.21-2011.11.24)] 7th International Conference on Broadband

map of the environment and the ability to interpret that map

are required. All navigation techniques require a representation

of the environment, a localization algorithm and one or more

sensors. Uncertainty is inherent in navigation and measurement

noise and motion noise are the causes of the uncertainty.

Overall, in this paper, the navigation and localization methods

of unmanned vehicles for automated meter reading systems

have been presented. Also, communication architecture of the

proposed system has been explained and comparative

simulation studies have been performed in terms of energy

efficiency and navigation accuracy.

The paper is organized as follows. Related work is given in

Section II. An automated meter reading application with

unmanned vehicles is explained in Section III. Performance

evaluations related to navigation and localization of unmanned

vehicles are also given in this section. Conclusions of the

paper and future work are given in Section IV.

II. RELATED WORK

After realizing the operational benefits of AMR systems,

AMR deployments are underway all around the world.

Specifically, the USA, Canada, Italy, Germany, and Australia

have deployed AMR systems in their regions.

Recently, the AMR systems have also taken the attention of

the researchers. In [1], a short range wireless AMR system

with a communication range of 50 m in 2.4GHz band is

presented. In [2], a prototype wireless automatic meter reading

system based on a ZigBee-based communication infrastructure

is explained. In [6], a GSM-based wireless automatic meter

reading system is proposed. In [5], a wireless automatic meter

reading system framework including the hardware of wireless

module and software design of data acquisition is explained.

Different from the above approaches, in [4] an automated

meter reading system using RF technology with a walk-by or

drive-by data collection system using pocket PCs is proposed.

In fact, most of the researches stated above are designed for

urban areas, where there are relatively high numbers of

consumers, since AMR infrastructures require relatively high

initial startup investments and maintenance costs. When there

are a few consumers scattered around a wide area, such as in

rural areas, investment costs constitute a major problem for

utility providers. The system proposed in this paper is aimed to

solve this specific problem.

III. USING UNMANNED VEHICLES FOR AUTOMATED METER

READING APPLICATIONS

The proposed system in this paper consists of several

components, such as communication infrastructure, navigation

and localization of unmanned vehicles, database architecture,

and servers running the AMR applications, etc. In this paper,

we mainly investigate two components of the proposed system

i.e., communication architecture, and navigation and

localization of unmanned vehicles. Fig. 1 shows the proposed

AMR architecture.

Fig. 1. Proposed offline AMR architecture.

A. Communication

Communication is an important part of the proposed AMR

architecture. Communication infrastructure can be wired or

wireless. The cost of a wired AMR system is more than a

wireless AMR system. In addition to the cost advantage,

deployment of wireless AMR systems is easier and more

flexible than wired AMR systems. Recently, Wireless Sensor

Networks (WSNs) have been recognized as a promising

technology for various smart grid applications [19]. WSNs

bring several advantages over traditional utility

communication technologies, such as low-cost and low-power

sensing, aggregated intelligence, and flexibility.

Communication architecture of the proposed AMR system is

based on IEEE 802.15.4 protocol and consists of battery-

operated WSN nodes. The major consideration for a WSN-

based AMR system includes the battery life of the WSN node.

To investigate the relationship between data transmissions

and lifetimes of battery operated WSN nodes, we have

developed an evaluation environment to perform simulations

using MATSNL [11], which allows using two different

operating models; trigger-driven, and schedule-driven models.

In trigger-driven operating model, a sensor node senses the

environment and wakes up the rest of the node once an event is

detected. On the other hand, in schedule-driven operating

model, the processor directly drives the sensor nodes so that

the processor wakes up to sample the sensors according to a

predetermined schedule [11]. The parameters of our simulation

are as follows:

• Event inter-arrival rate range: 10 minutes - 12 hours

• Event Duration: 1 sec

• Duty Period: 5 sec

• Duty Cycle: 0.1 - 1

• Radio TX data rate: 250 Kbps

• Data Size: 1 Kbyte

• Unit packet size: 128 byte

• Packet header size: 12 byte

In this simulation, we have examined the lifetimes of Telos

[16] and imote2 [15] sensor nodes for -25 dBm and 0 dBm

transmit powers to show the effect of transmit power on

lifetime. Table I shows the specifications of TI CC2420 which

is the transceiver of both Telos and imote2 nodes. Fig. 2 shows

our results for Telos and imote2 nodes.

Considering the results of our simulations, we can conclude

that Telos sensor nodes are more suitable for the proposed

system due to their longer battery lifetime. Although, reducing

Proceedings of the 6th International Conference on Broadband Communications & Biomedical Applications, November 21 - 24, 2011, Melbourne, Australia

290 IB2COM 2011

Page 3: [IEEE 2011 6th International Conference on Broadband and Biomedical Communications (IB2Com) - Melbourne, Australia (2011.11.21-2011.11.24)] 7th International Conference on Broadband

transmission power increases predicted minimum and

maximum lifetime of a node for both Telos, and imote2 nodes,

it reduces transmission range for both trigger-driven operation

and schedule-driven operating models. Hence, it is necessary

to balance the trade-off between connectivity and network

lifetime. This trade-off is also related to the navigation and

localization part of the proposed system.

TABLE I

TI CC2420 TRANSCEIVER SPECIFICATIONS

TI CC2420

Platforms Telos and Imote2

Standard IEEE 802.15.4

Data rate (kbps) 250

Modulation O-QPSK

Radio frequency (MHz) 2.4 GHz

Supply voltage (V) 2.1-3.6

TX max (mA/dBm) 17.4/0

TX min (mA/dBm) 8.5/-25

RX (mA) 18.8

Sleep (µA) 0.02

Startup (ms) 0.3-0.6

Fig. 2. Lifetime comparisons of Telos and imote2 sensors for -25 dBm and 0

dBm transmit powers for trigger-driven and schedule-driven models. Solid

lines represent schedule-driven operation and dashed lines represent trigger-

driven operation.

B. Localization and Navigation of Unmanned Vehicles

Navigation is the process used by unmanned vehicles to

move from an initial location to a final location. To navigate,

an unmanned vehicle must first localize itself. For localization,

an unmanned vehicle calculates its position through

information gathered from sensors [13], [14]. Since most

unmanned vehicles operate in dynamic environments, they

need to be able to acquire knowledge through perception [12],

[14]. Sensors are used for perception and measurements

gathered by the sensors are used for localization. There are a

variety of sensors used in unmanned vehicles and most of them

are used for navigation and guidance of unmanned vehicles.

Still, most navigation and guidance systems are based on dead

reckoning and external information [12], [13]. Unmanned

vehicles require a particular combination of dead reckoning

and external sensors and the performance of these sensors

directly affects the operation of unmanned vehicle navigation

and guidance systems. The roles of sensors are not limited to

navigation and guidance. Sensors are also used for obstacle

avoidance, collision avoidance, stopping the unmanned vehicle

when traffic lights are red, etc.

Successful implementation of the proposed architecture

depends on the accuracy of navigation and localization.

Accuracy of navigation and localization depends on several

factors, such as the quality of sensors, maximum observation

range, observation noises, time interval between observations,

time interval between odometry readings, environmental

factors, Simultaneous Localization and Mapping (SLAM)

algorithm, data association algorithm, etc. There are different

approaches to evaluate a performance of a SLAM algorithm.

Most SLAM benchmarking approaches use a metric relying on

a global reference frame to compute the error.

The accuracy of SLAM may be evaluated based on the

estimated trajectory of a robot. The difference between the

absolute ground truth pose and the estimated pose of the robot

is computed by using the mean squared error and this

measurement over all ground truth points in time is referred as

the absolute trajectory error (ATE) [10]. ATE is given by:

� �2 2

1

1. ( ) ( )

GT

i

N

i i i

iGT

error x x y yN

=

= − + −∑ (1)

where NGT represents the number of ground truth

measurements.

Observation range, speed, time interval between

observations, and time interval between odometry readings

affect SLAM accuracy. The effect of these parameters on

SLAM accuracy can be summarized as follows:

• Increasing observation range reduces ATE and

increases SLAM accuracy.

• Reducing the number of observations increases ATE

and reduces SLAM accuracy.

• Reducing the number of odometry readings increases

ATE and reduces SLAM accuracy.

• Increasing robot’s speed increases ATE and reduces

SLAM accuracy.

We have conducted a set of EKF-based experiments using

the simulation environment in [8] to show the navigation and

localization performances, and calculated ATE for different

speeds. Data association algorithm used in these experiments

is Nearest Neighbor data association algorithm (NN) [9]. EKF

consists of five steps. These steps are state prediction,

observation, measurement prediction, matching and

estimation. EKF uses a landmark based map and operates

recursively in two stages: Prediction and Update. In the

prediction stage, the command u(k) and the robot motion

model are utilized to estimate the robot’s location. Then, in the

update stage, to update the landmark’s position, and to refine

Proceedings of the 6th International Conference on Broadband Communications & Biomedical Applications, November 21 - 24, 2011, Melbourne, Australia

291 IB2COM 2011

Page 4: [IEEE 2011 6th International Conference on Broadband and Biomedical Communications (IB2Com) - Melbourne, Australia (2011.11.21-2011.11.24)] 7th International Conference on Broadband

the estimation of the robot’s location, the new observation z(k)

from an exteroceptive sensor is used.

Observations have been performed by using a laser range

finder. GPS observations have been used to estimate absolute

heading at the beginning and to compare the localization

results. Fig. 3 represents the ways that the vehicle should

follow which has been obtained with GPS and the estimated

paths of the vehicle. Fig. 4 shows the model covariance (x, y,

and theta) and Fig. 5 shows beacons covariances. To show the

effect of speed on ATE, we have conducted another set of

EKF-based SLAM simulations. Table II shows the results of

our simulations. It is seen that increasing the speed of an

unmanned vehicle reduces SLAM accuracy.

TABLE II

ATE WITH DIFFERENT SPEEDS

Observation

range (m)

Time interval

between

odometry

readings (sec.)

Time interval

between

observations

(sec.)

Speed

(m/sec.)

ATE

(m)

30 0.025 0.2 0.3 0.137

30 0.025 0.2 1 0.285

30 0.025 0.2 3 2.301

Fig. 3. Navigation result.

Fig. 4. Model covariance.

Fig. 5. Beacons covariances.

Considering the results of our experiments, we can conclude

that an unmanned vehicle can navigate and localize

successfully using its onboard sensors and a pre-loaded map

and collect metering data in rural areas. Also, as shown in

Table II, we observe that increasing unmanned vehicle’s speed

increases ATE, and reduces SLAM accuracy. Hence, an

optimal speed has to be determined before the operation.

One of the fundamental ideas to increase SLAM accuracy is

multi-sensor fusion [17], [18]. In this method, environmental

information obtained by multiple sensors is combined to

provide more reliable and accurate information. Using GPS-

aided localization is another method to increase SLAM

accuracy.

IV. CONCLUSION

This paper focuses on using unmanned vehicles for

automated meter reading applications and explains its potential

advantages and design challenges. The proposed unmanned

vehicle-aided AMR system is well suited for AMR

applications when there are a few consumers scattered around

a wide area. The system eliminates initial cost and

maintenance cost of a communication infrastructure and can be

integrated to an existing AMR system. Simulation studies

related to SLAM and wireless communication have been

performed to show the effectiveness of our approach.

While our performance evaluations provide valuable

insights into design issues for unmanned vehicle-aided

automated meter reading applications, they are only first steps.

As a future work, field tests with autonomous robot platforms,

i.e., Corobot [7], will be conducted to address navigation and

localization issues, in addition to communication related

issues.

ACKNOWLEDGMENT

The work of V. C. Gungor was supported by the European

Union FP7 Marie Curie International Reintegration Grant

(IRG) under Grant PIRG05-GA-2009-249206 with the

research project entitled “Spectrum-Aware and Reliable

Proceedings of the 6th International Conference on Broadband Communications & Biomedical Applications, November 21 - 24, 2011, Melbourne, Australia

292 IB2COM 2011

Page 5: [IEEE 2011 6th International Conference on Broadband and Biomedical Communications (IB2Com) - Melbourne, Australia (2011.11.21-2011.11.24)] 7th International Conference on Broadband

Wireless Sensor Networks for Europe's Future Electricity

Networks and Power Systems”. Also, the research has been

supported by Yildiz Technical University Scientific Research

Projects Coordination Department. Project Number: 2010-04-

02-ODAP01.

REFERENCES

[1]. C. Brasek, “Urban Utilities Warm up to the Idea of Wireless Automatic

Meter Reading,” IEE Computing and Control Engineering, vol. 15, no.

6, pp. 10-14, December 2004-January 2005.

[2]. T. Jamil, “Design and Implementation of a Wireless Automatic Meter

Reading System,” in Proceedings of the World Congress on

Engineering, vol. 1.pp. 217-221, London, UK, 2008.

[3]. Q. Liu, B. Zhao, Y. Wang, and J. Hu, “Experience of AMR Systems

Based on BPL in China,” in Proc. of IEEE International Symposium on

Power Line Communications and Its Applications, Dresden, pp. 280-

284, 2009.

[4]. A. Bahramiazar, “Automated Meter Reading Using RF Technology,” in

Proceedings of the IEEE Innovative Smart Grid Technologies

Conference Europe, Gothenburg, pp. 1-5, October 2010.

[5]. D. Miao, K. Xin, Y. Wu, W. Xu, and J. Chen, “Design and

implementation of a wireless automatic meter reading system,” in

Proc. of ACM International Conference on Wireless Communications

and Mobile Computing, New York, USA, pp. 1345-1349, 2009.

[6]. S. S. Sheikh, and S. Sharma, “Design and Implementation of Wireless

Automatic Meter Reading System,” International Journal of

Engineering Science and Technology (IJEST), March 2011, Vol. 3,

No. 3, pp. 2329-2334.

[7]. (2010) Corobot. [Online]. Available: http://robotics.coroware.com/

[8]. J. E. Guivant, F. R. Masson, E. M. Nebot, “Simultaneous Localization

and Map Building Using Natural Features and Absolute Information,

Robotics and Autonomous Systems, vol. 40, no. 2-3, pp. 79-90, 2002.

[9]. L. M. Paz. J. D. Tardos and J. Neira, “Divide and Conquer: EKF SLAM

in O(n),” IEEE Transactions on Robotics, vol. 24, no. 5, pp. 1107-

1120, 2008.

[10]. E. A. Ruckert, “Simultaneous Localization and Mapping for Mobile

Robots with Recent Sensor Technologies,” M.S. Thesis, Graz

University of Technology, Graz, 2009.

[11]. D. Jung, T. Teixeira, A. Barton- Sweeney and A. Savvides, “Model-

Based Design Exploration of Wireless Sensor Node Lifetimes,” in

Proc. of the European Conference on Wireless Sensor Networks,

EWSN 2007, Netherlands, 2007.

[12]. M. Nebot, “Sensors used for autonomous navigation,” chap. 7 in

Advances in Intelligent Autonomous Systems, pp. 135-156, Kluwer,

2000.

[13]. N. C. Tsourveloudis, L. Doitsidis and K. P. Valavanis, "Autonomous

Navigation of Unmanned Vehicles: A Fuzzy Logic Perspective,"

Chapter 18 in Cutting Edge Robotics, InTech, 2005.

[14]. G. A. Demetriou, “A Survey of Sensors for Localization of Unmanned

Ground Vehicles (UGVs),” in Proc. of International Conference on

Artificial Intelligence, pp. 659-668, 2006.

[15]. (2011) Imote2 – TinyOS Documentation Wiki [Online]. Available:

http://docs.tinyos.net/index.php/Imote2

[16]. (2011) Crossbow Technology. [Online]. Available:

http://www.xbow.com/

[17]. R. Luo, ve C. Yih, “Multisensor Fusion and Integration Aspects of

Mechatronics,” IEEE Industrial Electronics Magazine, vol. 4, no. 2,

pp. 20-27, 2010.

[18]. N. Tomatis, “Hybrid, Metric-Topological, Mobile Robot Navigation,”

Ph.D. Thesis, EPFL, Lausanne, 2001.

[19]. V. C. Gungor, B. Lu, and G. P. Hancke, “Opportunities and Challenges

of Wireless Sensor Networks in Smart Grid,” IEEE Transactions on

Industrial Electronics, vol. 57, no. 10, pp. 3557-3564, 2010.

Proceedings of the 6th International Conference on Broadband Communications & Biomedical Applications, November 21 - 24, 2011, Melbourne, Australia

293 IB2COM 2011