Exploring the Role of a Smartphone as a Motion Sensing and Control

Device in the Wireless Networked Control of a Motor Test-bed

Jared A. Frank, Anthony Brill, Jonghyun Bae and Vikram Kapila

Mechanical and Aerospace Engineering, NYU Polytechnic School of Engineering, Brooklyn, NY, 11201, U.S.A.

Keywords:

Attitude Sensing, Motor Control, Networked Control System, Smartphone Sensors, Wireless.

Abstract:

The sensing, computing, and control potential of smartphones remains to be fully explored in automatic control

applications. In this paper, we control the angular position of a motor test-bed using feedback from the

embedded motion sensors of a smartphone while it is mounted to the test-bed. The smartphone hosts an

interactive user interface which students and researchers can use to quickly and easily perform experiments

with the test-bed and collect measurements using their own personal devices. Proportional-plus-derivative

(PD) controllers designed using a sampled-data model of the system are compared for different sampling rates

used on the smartphone. Results from simulations and experiments conﬁrm the feasibility of utilizing mounted

smartphones in the wireless networked control of systems with rotational degrees of freedom.

1 INTRODUCTION

Smartphones are more powerful than ever; embed-

ded sensors, open and programmable architecture,

and support for several communication protocols en-

dow them with the potential to serve as lightwight,

portable, inexpensive platforms for sensing in a wide

array of applications (Lane, 2010). Several mobile

applications have been developed that utilize the em-

bedded sensors of smartphones for real-time monitor-

ing. Some of the most popular of these sensors are the

motion sensors, including accelerometer, gyroscope,

and magnetometer. Algorithms for processing and

integrating data from these sensors yield device atti-

tude measurements accurate enough for clinical use in

measuring the Cobb angle in both kyphoses (Jacquot,

2012) and scoliosis (Shaw, 2012). The motion sen-

sors in smartphones have also been used for position-

ing applications in indoor environments where GPS

signal is weak (Schindhelm et al., 2011) and to en-

courage autonomous navigation for visually impaired

people (Moder et al., 2014). By analyzing accelerom-

eter data, mobile applications can be used for activity

recognition, e.g., to detect when users are walking,

jogging, and ascending or descending stairs (Khan,

2010). The medical applications include use of smart-

phone accelerometer data to analyze gait characteris-

tics (LeMoyne, 2010a), characterize Parkinson’s dis-

ease tremor (LeMoyne, 2010b), and notify emergency

services if a user has fallen (He et al., 2012).

Exploiting the sensing power of smartphones has

led to improvements in feasibility and reductions in

complexity in diverse applications. An artiﬁcial pan-

creas system has been developed that uses a smart-

phone to process wireless data from a continuous glu-

cose monitoring (CGM) system worn by patients with

type 1 diabetes (Kovatchev, 2013). Previous imple-

mentations consisted of the CGM and insulin pump

wired to computers, which limited free movement

and outdoor usage. A smartphone implementation

may also yield cost reductions vis-a-vis conventional

hardware. For instance, a system that processes ac-

celerometer data from a smartphone to detect potholes

has been proposed as a substitute for high-cost equip-

ment (Mednis, 2011).

Although the incorporation of smartphones to

monitor and interact with physical systems is now be-

ing investigated (Frank and Kapila, 2014), the poten-

tial role of these devices in the automatic, closed-loop

networked control of systems remains largely unex-

plored. Notable exceptions include the construction

of smartphone-controlled robots and unmanned ve-

hicles. Although traditionally microprocessors on-

board the vehicle are used to execute the robot’s con-

trol algorithms, smartphones contain processors with

multiple cores, real-time operating systems, and suf-

ﬁcient speed and memory to handle large computa-

tional loads (Aroca, 2012). Thus, smartphones have

recently been used as navigation and guidance units

for marine and aerial vehicles (El-Gaaly, 2013; Desai,

328

Frank J., Brill A., Bae J. and Kapila V..

Exploring the Role of a Smartphone as a Motion Sensing and Control Device in the Wireless Networked Control of a Motor Test-bed.

DOI: 10.5220/0005544403280335

In Proceedings of the 12th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2015), pages 328-335

ISBN: 978-989-758-123-6

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

2013). In this role, the smartphone can use embedded

sensors to detect obstacles in the environment, sense

collisions, and estimate the pose of the robot as the

processor computes vehicle velocities.

In this paper, we investigate the use of a smart-

phone as a wireless mountable motion sensing and

control device in the automatic control of a motor test-

bed. This implementation reduces the wiring, cost,

and complexity involved in installing industrial-grade

sensors and enables users to leverage the sensing,

computation, and communication power of their mo-

bile devices to quickly and easily drive simple single

degree-of-freedom systems. Students and researchers

in engineering laboratories can attach their own per-

sonal devices to the test-bed to automatically control

it and collect data. An interactive user interface to

the test-bed on the smartphone screen allows users to

set parameters, such as control gains and set points

for the system, and to save and email themselves the

collected data for post-processing. An overview of

the system is provided, including the motion sens-

ing capability of the smartphone and the design of a

3D-printed mount for attaching the smartphone to the

test-bed. PD controllers, which are to be implemented

on the smartphone, are designed for several different

sampling rates using a sampled-data model of the sys-

tem. The motor’s response to these controllers are ex-

plored in simulations and experiments. Since the ac-

curacy, drift, and calibration of sensor data, as well as

time delays associated with computation and wireless

communication, can signiﬁcantly impact the stability

and performance of the system, the effects of these

factors are measured and discussed in the context of

the proposed system.

2 SYSTEM DESCRIPTION

The test-bed used in this study consists of a geared

DC-motor attached to a power ampliﬁer (see Fig-

ure 1). An incremental optical encoder and a multi-

turn potentiometer are available to monitor the motor

orientation and a tachometer is available to measure

its angular rate. A desktop computer equipped with a

PC-based data acquisition and control (DAC) board is

used to transmit control signals to the power ampliﬁer

that drives the motor.

2.1 Mount Design

To secure the smartphone to the test-bed so that ac-

curate orientation data is obtained from the motion

sensors of the smartphone, a mount is designed and

3D-printed using polylactide (PLA) material (see Fig-

Figure 1: The motor test-bed used in this study.

ure 2(a)). Using guidelines and dimensions provided

by the smartphone manufacturer, the mount is de-

signed with four tabs to clip the phone in place and

prevent it from rotating with respect to the motor arm

(see Figure 2(b)). The mount is designed to align the

axis of rotation through the geometric center of the

smartphone. Consideration is taken to not block the

buttons or the screen of the smartphone so that a user

interface may be used to interact with the test-bed.

(a) (b)

Figure 2: Phone mount design (a) without and (b) with an

iPhone 5 secured in between its tabs.

2.2 Embedded Sensing and Calibration

The smartphone used in this study is an Apple iPhone

5, which contains a three-axis accelerometer and gy-

roscope from STMicroelectronics and a magnetome-

ter from AKM Semiconductor. Data from the ﬁrst

two sensors is used to estimate the attitude of the de-

vice. A single class is used for central access to raw

data from each axis of the motion sensors. Estimates

of device attitude and rotation rate are generated af-

ter processing raw data with sensor fusion algorithms.

Device attitude is expressed in Euler angle form and

deﬁned as the rotation between the device’s current

reference frame and the initial reference frame that is

established when the mobile application (“app”) ﬁrst

begins collecting motion updates. In this study, the

iPhone is rigidly mounted to the motor arm, which

lies in the horizontal plane. Therefore, only data as-

sociated with the axis normal to the device screen is

needed to estimate the orientation of the motor arm.

To obtain motion data, the app periodically sam-

ples the most recently available measurement at a

speciﬁed rate known as the update interval. Choos-

ing an update interval is an important part of the de-

sign of the system since it has a signiﬁcant impact on

ExploringtheRoleofaSmartphoneasaMotionSensingandControlDeviceintheWirelessNetworkedControlofaMotor

Test-bed

329

the performance of the closed-loop. This interval is

referred to as the sampling rate in the sampled-data

model that characterizes the closed-loop behavior of

the wireless networked control system. With a larger

update interval, the embedded sensors are polled less

often by the app, improving battery consumption.

While 10-20 Hz sampling frequency may be suitable

for many teleoperation tasks, Apple’s event handling

guidelines recommend a range of 30-60 Hz for col-

lecting device orientation data for real-time applica-

tions. Mobile applications developed for iOS support

update frequencies up to 100 Hz for detecting high-

frequency motions, such as impacts and quick shakes.

After performing an experiment involving the collec-

tion of 350 samples, the mean motion data computa-

tion time on the smartphone is determined to be 244.4

microseconds with a standard deviation of 91.168 mi-

croseconds. This gives a 95% conﬁdence interval of

[239.13,258.36] microseconds for the mean, which is

fast enough to support a 100Hz sampling rate.

Calibration of the embedded sensors is performed

using the potentiometer sensor on the test-bed. Ex-

periments are initiated at the orientation in which the

potentiometer outputs zero volts. Therefore, the angu-

lar position readings from the smartphone and the po-

tentiometer are aligned. However, experiments show

that the readings of device attitude drift over time with

respect to readings from the potentiometer (see Sec-

tion 5.1). In future work, the use of the embedded

magnetometers of smartphones will be investigated to

both calibrate the system and improve the accuracy of

readings by correcting against drift.

2.3 Communication

Wireless communication of data between the smart-

phone and the desktop computer is performed using

a traditional client-server architecture, with a server

running on the desktop computer and the client run-

ning on the smartphone app. Data is sent and received

using the TCP/IP protocol over a wireless local area

network, which is maintained by a Wi-Fi router. The

Nagle algorithm, a standard method of improving the

efﬁciency of the network, is built into the TCP/IP

protocol. This algorithm combines small outgoing

messages that cause a sufﬁciently large overhead and

sends them all at once in a burst (Nagle, 1995). This

process introduces latency, so the Nagle algorithm is

disabled on both sides of the communication to allow

data to be sent and received in real-time.

To test communication on the wireless network, a

command is issued on the desktop computer to ping

the smartphone and another computer. After send-

ing 100 echo requests with 32 byte packets, the av-

erage round trip times for the smartphone and the

other computer are determined to be 78.11 ms and

30.79 ms, respectively. A paired t-test is conducted

to test whether the two sets come from distributions

with equal means. With a 95% conﬁdence level,

the null hypothesis could be rejected (with t(99) =

2.324, p = 0.022 < 0.05, 95%CI = [6.927,87.713]).

Thus, the communication with the smartphone expe-

riences larger latency than a typical desktop computer

on the same network.

Network latency results in time delays between

data collected by the sensors on the test-bed and the

embedded sensors of the smartphone. These time de-

lays in the closed-loop system can vary signiﬁcantly,

which can cause degradation in system stability and

performance and complicate analysis and control de-

sign. Several control methodologies have been used

over the last several decades to compensate for such

effects, spanning the use of Smith predictors, opti-

mal stochastic methods, fuzzy logic, and queuing and

buffering. However, with a mean one-way communi-

cation delay of 39.055 ms, average delays in the sys-

tem are only between 0 and 4 sampling periods, de-

pending on the sampling rate used. As will be shown

in Section 5.2, this amount of delay is negligible when

controlling the motor test-bed, whose bandwidth is

sufﬁciently low.

3 USER INTERFACE DESIGN

Mounting the smartphone to the test-bed can add

portable sensing, computing, and control capability to

the system, as well as provide an interactive user inter-

face attached to the motor arm. Figure 3shows screen-

shots of a mobile app implemented on an iPhone 5

device, allowing users to start and stop experiments

as well as issue reference commands from the smart-

phone touch screen. A connect button on the interface

establishes wireless communication with the compu-

Figure 3: Screenshots of the mobile application’s user in-

terface on an iPhone 5.

ICINCO2015-12thInternationalConferenceonInformaticsinControl,AutomationandRobotics

330

ter, and a text ﬁeld is used to issue a reference angle

for position control. Sensor data is displayed in real-

time on the screen using text labels. A button to start

and stop the experiment is located in the center of the

screen so that the position of the button is not affected

by the rotation of the device, making it easy for users

to stop the motor in case of an emergency while the

experiment is running.

4 MODELING AND CONTROL

DESIGN

To control the orientation of the motor arm using a

mounted smartphone, the mobile app on the smart-

phone computes the control action u(t) at time t = kT ,

where T is the sampling rate and k = 0,1, 2, . .., af-

ter the angle θ(kT ) and angular rate ω(kT ) have been

sampled. This control computation is done according

to the PD control structure

u[kT ] = K

θ[kT ] + K

ω[kT ].

To design the proportional and derivative control

gains K

and K

, respectively, the standard pole-

placement technique is used in discrete-time. First,

the continuous-time model of the plant is obtained.

Next, the model of the plant is discretized to model

the ﬁxed-rate sampling of motion data and control

computation by the smartphone. PD controllers are

designed and compared for several sampling rates.

4.1 Plant Model

The plant, composed of an armature-controlled DC-

motor, gearbox, and load (i.e., the motor arm, mount,

and the smartphone), has a dynamic behavior that

can be captured by a ﬁrst-order transfer function from

voltage input to the motor V

(s) to the angular veloc-

ity of the motor Ω(s)

Ω(s)

(s)

τs + 1

where K is the steady-state gain and τ is the time-

constant. A unit-step voltage is applied to the plant

to experimentally identify the values of K and τ as

1.5981 Volts and 0.029 seconds, respectively.

4.2 Discretization

The effect of the smartphone in the loop, notably the

discrete-time sampling of data, must be considered

in the control design. A sampled-data model is used

to describe the proposed system in which the smart-

phone performs the sensing and the control computa-

tion for the test-bed. Samplers and zero-order holds

are introduced in the feedback paths and in the feed-

forward path of the closed-loop (see Figure 4). These

samplers represent the sampled attitude and angular

velocity data handled by the smartphone, as well as

the control action computed using the PD control al-

gorithm running on the backend of the mobile app.

Since the control data is transmitted to the host ma-

chine in the same TCP/IP packet as the sensor read-

ings, which are used by the host machine only for data

logging, the three samplers are assumed to share the

same uniform sampling rate T .

In choosing the sampling rate on the smartphone

such that the closed-loop system is stable, it is re-

quired that the sampling rate be sufﬁciently fast rel-

ative to the dynamics of the plant. A commonly used

rule of thumb in digital control design is to keep the

sampling rate faster than 30 times the closed-loop

bandwidth frequency, ω

, which is calculated directly

from values of damping ratio ζ and natural frequency

(Kuo, 1980)

= ω

[(1 − 2ζ

) +

4ζ

− 4ζ

+ 2]

1/2

To design PD controllers such that the step response

of the closed-loop system exhibits no overshoot and

a settling time T

of less than one second, the desired

natural frequency ω

and damping ratio ζ are obtained

(Dorf and Bishop, 2007)

ζ = 0.9959, ω

= −log(0.02)/(ζ × T

) = 3.9282.

Therefore the value of the bandwidth is calculated

to be 2.5429 rad/sec (0.4047 Hz), which means that

the sampling rate used with the smartphone should

be larger than ≈ 12.2 Hz. To compare the system

performance at various sampling rates supported by

the smartphone processor, PD-controllers will be de-

signed using a discretized state-space representation

of the closed-loop system with sampling rates set to

1,5,10,30,60, and 100 Hz. The open-loop trans-

fer function for the plant can be transformed into

a state-space representation in which the state is a

vector of the angular orientation and rate, x(t) ,



θ(t) ω(t)



, yielding

˙x = Ax + Bu, y = Cx,

A ,



0 1

0 −



, B ,





, C ,



1 0



This state-space representation can be discretized, at

each sampling instant kT , k = 0, 1, 2, . .., as follows

(Kuo, 1980)

x[(k + 1)T ] = φ(T )x[kT ] + γ(T )u[kT ],

y[kT ] = Cx[kT ],

ExploringtheRoleofaSmartphoneasaMotionSensingandControlDeviceintheWirelessNetworkedControlofaMotor

Test-bed

331

Figure 4: Sampled-data block diagram of the closed-loop system.

where φ(T ) , e

is the state transition matrix of the

state matrix A of the continuous-time state equation

and γ(T ) ,

φ(T − τ)Bdτ. Before designing con-

trollers, the controllability of this second-order sys-

tem is veriﬁed by conﬁrming that the 2 × 2 controlla-

bility matrix M



φ(T ) φ(T )γ(T )



of the sys-

tem has full rank. Indeed, the sampled-data mod-

els are found controllable and since the state vec-

tor contains the angular orientation and rate, the PD

controllers can be designed with the digital full-state

feedback control law u[kT ] = −K

x[kT ], where K





is the PD-control gain matrix.

4.3 Controller Design

The continuous-time transfer function of the DC-

motor controlled by a continuous-time PD controller

u(t) = k

θ(t)+ k

ω(t) is

Θ(s)

R(s)

(1+Kk

)

s +

where R(s) and Θ(s) denote the reference command

and angular position output, respectively. The proto-

type second-order transfer function is

Y (s)

U(s)

+ 2ζω

s + ω

Matching coefﬁcients, the nominal expressions for

the control gains are found

2ζω

τ − 1

, k

Plugging numerical values of K, τ, ζ, and ω

gives

= 0.2800 and k

= −0.4838. To design con-

trollers based on discretized models of the plant using

the pole-placement technique, the prototype second-

order system is discretized. Then, the desired charac-

teristic equation is determined in the z-domain from

the discretized state-space matrix A

λ(z,A

) =

zI − A

Using Ackermann’s formula, the controller gain ma-

trix K

(T ) for the system discretized at a sampling

rate T can be computed from the controllability ma-

trices M

(T ) and the matrix returned by plugging the

Table 1: PD Control Design Parameters.

Sampling

Rate

Open-Loop

Poles

Desired Poles Controller [K

]

1 Hz 1, 0.0000 0.0187 ± 0.0070ı [0.6025, 0.0172]

5 Hz 1, 0.0010 0.4561 ± 0.0325ı [0.9296, -0.1036]

10 Hz 1, 0.0318 0.6758 ± 0.0241ı [0.6830, -0.2574]

30 Hz 1, 0.3168 0.8777 ± 0.0104ı [0.4141, -0.4099]

60 Hz 1, 0.5629 0.9369 ± 0.0056ı [0.3450, -0.4481]

100 Hz 1, 0.7083 0.9616 ± 0.0034ı [0.3184, -0.4628]

discretized state matrix φ(T ) into the desired charac-

teristic equation (Dorf and Bishop, 2007)

(T ) = [0 1]M

−1

(T )λ(z, φ(T )).

Table 1 shows the open-loop and closed-loop poles, as

well as the controller designed, for each of the sam-

pling rates explored in the study.

5 EXPERIMENTAL RESULTS

5.1 Sensor Noise and Accuracy

Drift errors associated with gyroscope-based orienta-

tion estimates have been observed and documented

over the past decades (Oravetz and Sandberg, 1970).

Commercially available devices have been found to

provide reliable estimates of orientation when used

for periods of up to one minute (Luinge and Veltink,

2005). Fortunately, the proposed system is expected

to be stabilized within a few seconds. To investigate

the noise and drift characteristics of the smartphone

sensor data, two 20 second tests are conducted with

the smartphone sensors (sampled at 60 Hz) and the

potentiometer and tachometer sensors on the test-bed.

The ﬁrst test involved collecting sensor data while the

motor is held at the zero orientation without being

driven. The angular position data from the smart-

phone sensors and potentiometer sensor are shown

in Figure 5(a) for this static test. Smartphone atti-

tude data has an average drift rate of 1.0297x10

−4

rad/sec (0.0059

◦

/sec). Angular rotation rate data from

the gyroscope and the tachometer are shown in Fig-

ure 5(b). Data from the tachometer has a slight bias;

the mean of the gyroscope data is −1.4439x10

−4

rad/sec (−0.0083

◦

/sec) while the tachometer mean

is 0.0038 rad/sec (0.2177

◦

/sec). However, the data

from the smartphone is much noisier and causes drift

ICINCO2015-12thInternationalConferenceonInformaticsinControl,AutomationandRobotics

332

(a)

(b)

Figure 5: Comparisons between (a) the smartphone attitude

sensing and potentiometer readings and (b) the smartphone

gyroscope readings and tachometer readings while the mo-

tor is not being driven.

in smartphone attitude. The standard deviation of

the gyroscope is 0.0044 rad/sec (0.2521

◦

/sec) com-

pared to 7.0937x10

−4

rad/sec (0.0406

◦

/sec) for the

tachometer.

The second test involved collecting sensor data

while the motor is driven from its zero orientation

with a constant 0.5 Volt signal. The angular po-

sition measurements from the smartphone and from

the potentiometer of the test-bed are shown together

in Figure 6(a). At the beginning of the run, the

difference between the angular position readings is

0.0030rad (0.1719

◦

) and at the end of the run, the dif-

ference is −0.0174 rad (−0.9969

◦

). Over the course

of the run, the mean difference between the two sen-

sors is −0.0247 rad (−1.4152

◦

). This represents a

sufﬁciently small amount of error for the case when

the motor is continuously driven over a long time in-

terval. The measurements from the gyroscope of the

smartphone and from the tachometer of the test-bed

are shown together in Figure 6(b). The mean read-

ing for the angular speed reported by the gyroscope

is 0.7070 rad/sec (40.5081

◦

/sec) compared to the

tachometer’s 0.7163rad/sec (41.0410

◦

/sec), indicat-

ing a difference of just 0.0093rad/sec (0.5329

◦

/sec).

The standard deviation of the smartphone’s gyroscope

for this test is 0.0113rad/sec (0.6474

◦

/sec) and for

the tachometer 0.0107 rad/sec (0.6131

◦

/sec), show-

ing that the two sensors have noise of a similar mag-

nitude during this dynamic test.

To stabilize the motor test-bed, attitude estimates

are only needed for a few seconds. Therefore, read-

ings from embedded gyroscopes and accelerometers

are sufﬁcient to provide such estimates. For long-

term use, additional sensors such as onboard three-

axis magnetometers can provide additional sensory

(a)

(b)

Figure 6: Comparisons between (a) the smartphone attitude

sensing and potentiometer readings and (b) the smartphone

gyroscope readings and tachometer readings while the mo-

tor is being driven by a 0.5 Volt signal.

information necessary to eliminate drift. Traditional

approaches of fusing the data between the three sen-

sors make use of unscented Kalman ﬁlters (Harada,

2003), extended Kalman ﬁlters (Marins, 2001), and

other ﬁltering techniques. The implementation of

such techniques using the embedded magnetometer

on the smartphone will be explored in future work.

5.2 Controlling the Test-bed

To explore the feasibility of controlling the motor test-

bed using the attitude and speed measurements col-

lected by the smartphone and running the PD-control

algorithm directly on the smartphone, trials are con-

ducted in which the motor is given a step input com-

mand of 90

◦

(1.5708 rad). Figures 7(a)–7(f) show the

motor’s closed-loop response at each sampling rate.

By plotting the angular position data reported by the

smartphone next to the readings of the potentiometer

as it is sampled by the desktop computer at 1,000 Hz,

one can visualize the communication delay between

the smartphone and computer as well as the drift in

the smartphone sensor data. Note that the results of

simulation are also plotted alongside the experimental

results. For comparison purposes, Figure 7(g) shows

the response when the nominal PD controller is im-

plemented on the desktop computer using feedback

from the potentiometer and tachometer sensors of the

test-bed.

Table 2 shows the values of the percent overshoot

and settling time for each of the responses depicted

in Figure 7. These results show that the percent over-

shoot and settling time increase as the sampling rate

of the smartphone is decreased, indicating an im-

provement in the responsiveness of the closed-loop

ExploringtheRoleofaSmartphoneasaMotionSensingandControlDeviceintheWirelessNetworkedControlofaMotor

Test-bed

333

(a) (b)

(e) (f)

(g)

Figure 7: Closed-loop responses with the redesigned con-

troller implemented using the smartphone (a–f) and the

nominal controller implemented using the computer (g).

system as the sampling rate becomes faster. Note that

a large deterioration in the quality of the performance

is observed from the motor when the smartphone-

based controller is run slower than 10 Hz. Since 5

Hz and 1 Hz are signiﬁcantly less than 30 times the

closed-loop bandwidth frequency, sampling the motor

system at these rates leaves a large amount of inter-

sample behavior uncaptured by the smartphone, caus-

ing a signiﬁcant loss in performance. Finally, each

response exhibits varying degrees of steady-state er-

Table 2: Motor response performance using the digitally

redesigned PD-control implementation.

Sampling Rate Maximum

Overshoot (%)

Settling

Time (s)

no phone 0.00 0.798

100 Hz 0.00 1.096

60 Hz 0.00 1.128

30 Hz 0.00 1.140

10 Hz 0.00 1.216

5 Hz 8.12 2.188

1 Hz 46.23 5.986

ror, which can be attributed to mechanical properties

of the motor and gearbox, such as friction, that result

in a deadzone in the motor’s sensitivity around 0 Volt.

This steady-state error can be overcome with the in-

troduction of an integral term to the controller. Note

that by designing the controllers using the sampled-

data model, the response of the motor at each sam-

pling rate is more or less consistent and varies little

from one sampling rate to another.

To conﬁrm that time delays introduced by the

wireless network latency may be neglected in the

design of the proposed system, a simulation of

the sampled-data model was built in the MAT-

LAB/Simulink environment with a constant delay of

39.055 ms introduced (the mean one-way communi-

cation time measured in Section 2.3). As can be seen

in Figure 7, the experimental results match the sim-

ulated responses quite well, indicating that delays of

0-4 sampling periods are not signiﬁcant when control-

ling a system with such a low bandwidth. In future

work, the use of the proposed system with more time-

sensitive, high-bandwidth plants will be explored.

With a low closed-loop bandwidth, a large range

of sampling rates may be used to stabilize the pro-

posed system. However, it is observed that the battery

is consumed quicker at faster sampling rates, since the

sensors as well as the communication module con-

sume a substantial amount of a smartphone’s power.

Therefore, a sampled-data model can be used to de-

sign a controller to be implemented on the smart-

phone at a sampling rate just about 30 times the

closed-loop bandwidth frequency.

6 CONCLUSIONS

This paper explored an approach to wirelessly con-

trol the orientation of a motor test-bed using feed-

back from the embedded motion sensors of a smart-

phone mounted to the test-bed. The study explored

the case of performing the control algorithm on the

smartphone itself. A PD control algorithm was used,

with controllers running at several different sampling

ICINCO2015-12thInternationalConferenceonInformaticsinControl,AutomationandRobotics

334

rates being compared. Results from simulations and

experiments validate the feasibility of using a smart-

phone to serve as the sensing and control platform in

the automatic control of systems with a single rota-

tional degree of freedom. Future work will consider

extensions to more complex systems with both rota-

tional and translational degrees of freedom, such as an

inverted pendulum on cart system. Studies with stu-

dents will be conducted to investigate whether the use

of mobile devices in the proposed manner is engag-

ing and provides access to more effective, interaction-

based educational, training, and research experiences

in the ﬁelds of automatic and digital control.

ACKNOWLEDGEMENTS

This work is supported in part by the National Science

Foundation awards RET Site EEC-1132482, GK-12

Fellows DGE: 0741714, and DRK-12 DRL: 1417769,

and NY Space Grant Consortium grant 48240-7887.

The authors would like to thank Matthew Moor-

head for the design and fabrication of the smartphone

mount.

REFERENCES

Aroca, R.V., et al. (2012). Towards smarter robots with

smartphones. In 5th Workshop in Applied Robotics

and Automation, pages 1–6.

Desai, A., et al. (2013). Stabilization and control of

quad-rotor helicopter using a smartphone device. In

IS&T/SPIE Electronic Imaging, page 866208.

Dorf, R. and Bishop, R. (2007). Modern Control Systems

(11th ed.). New Jersey: Prentice Hall.

El-Gaaly, T. et al. (2013). Visual obstacle avoidance for

autonomous watercraft using smartphones. In Au-

tonomous Robots and Multirobot Systems Workshop.

Frank, J. and Kapila, V. (2014). Development of mobile

interfaces to interact with automatic control experi-

ments [focus on education]. IEEE Control Systems,

34(5):78–98.

Harada, T., et al. (2003). Portable orientation estimation

device based on accelerometers, magnetometers and

gyroscope sensors for sensor network. In IEEE Int.

Conf. Multisensor Fusion and Integration for Intelli-

gent Systems, pages 191–196.

He, Y., Li, Y., and Yin, C. (2012). Falling-incident detec-

tion and alarm by smartphone with multimedia mes-

saging service (MMS). E-Health Telecommunication

Systems and Networks, 1(1):1–5.

Jacquot, F., et al. (2012). Measuring the Cobb angle with

the iPhone in kyphoses: A reliability study. Interna-

tional Orthopaedics, 36(8):1655–1660.

Khan, A.M., et al. (2010). Human activity recognition

via an accelerometer-enabled smartphone using ker-

nel discriminant analysis. In Int. Conf. Future Infor-

mation Technology, pages 1–6.

Kovatchev, B.P., et al. (2013). Feasibility of outpatient fully

integrated closed-loop control ﬁrst studies of wearable

artiﬁcial pancreas. Diabetes Care, 36(7):1851–1858.

Kuo, B. (1980). Digital Control Systems (2nd ed.). Oxford

University Press.

Lane, N.D., et al. (2010). A survey of mobile phone sens-

ing. IEEE Communications Magazine, 48(9):140–

150.

LeMoyne, R., et al. (2010a). Implementation of an

iPhone as a wireless accelerometer for quantifying

gait characteristics. In IEEE Int. Conf. Engineering

in Medicine and Biology Society, pages 3847–3851.

LeMoyne, R., et al. (2010b). Implementation of an iPhone

for characterizing Parkinson’s disease tremor through

a wireless accelerometer application. In IEEE Int.

Conf. Engineering in Medicine and Biology Society.

Luinge, H. and Veltink, P. (2005). Measuring orientation

of human body segments using miniature gyroscopes

and accelerometers. Medical and Biological Engi-

neering and Computing, 43(2):273–282.

Marins, J.L., et al. (2001). An extended kalman ﬁlter for

quaternion-based orientation estimation using MARG

sensors. In IEEE/RSJ Int. Conf. Intelligent Robots and

Systems, volume 4, pages 2003–2011.

Mednis, A., et al. (2011). Real time pothole detection us-

ing Android smartphones with accelerometers. In Int.

Conf. Distributed Computing in Sensor Systems and

Workshops, pages 1–6.

Moder, T., Hafner, P., and Wieser, M. (2014). Indoor posi-

tioning for visually impaired people based on smart-

phones. In Computers Helping People with Special

Needs, pages 441–444.

Nagle, J. (1995). Congestion control in IP/TCP internet-

works. ACM SIGCOMM Computer Communication

Review, 25(1):61–65.

Oravetz, A. and Sandberg, H. (1970). Stationary and non-

stationary characteristics of gyro drift rate. AIAA Jour-

nal, 8(10):1766–1772.

Schindhelm, C., Gschwandtner, F., and Banholzer, M.

(2011). Usability of Apple iPhones for inertial naviga-

tion systems. In IEEE Int. Symp. Personal Indoor and

Mobile Radio Communications, pages 1254–1258.

Shaw, M., et al. (2012). Use of the iPhone for Cobb angle

measurement in scoliosis. European Spine Journal,

21(6):1062–1068.

ExploringtheRoleofaSmartphoneasaMotionSensingandControlDeviceintheWirelessNetworkedControlofaMotor

Test-bed

335