close

Вход

Забыли?

вход по аккаунту

?

1.4976787

код для вставкиСкачать
Development of a telepresence robot for medical consultation
Nilo T. Bugtai, Aira Patrice R. Ong, Patrick Bryan C. Angeles, John Keen P. Cervera, Rachel Ann E. Ganzon,
Carlos A. G. Villanueva, and Samuel Nazirite F. Maniquis
Citation: AIP Conference Proceedings 1817, 040002 (2017);
View online: https://doi.org/10.1063/1.4976787
View Table of Contents: http://aip.scitation.org/toc/apc/1817/1
Published by the American Institute of Physics
Articles you may be interested in
Effects of variation of chitosan concentration on the characteristics of membrane cellulose bacteria-chitosan
biocomposites as candidates for artificial dura mater
AIP Conference Proceedings 1817, 020011 (2017); 10.1063/1.4976763
Development of iPS (induced pluripotent stem cells) using natural product from extract of fish oocyte to provide
stem cell for regenerative therapy
AIP Conference Proceedings 1817, 020008 (2017); 10.1063/1.4976760
Optimization in expression and purification of modified apoptin as selective anti-cancer therapy
AIP Conference Proceedings 1817, 030014 (2017); 10.1063/1.4976783
Multifunctional microsphere formulation of fluorescent magnetic properties for drug delivery system
AIP Conference Proceedings 1817, 030011 (2017); 10.1063/1.4976780
Correlation between calcium and phosphate levels to calculus accumulation on coronary heart disease patients
AIP Conference Proceedings 1817, 030007 (2017); 10.1063/1.4976776
Study of homogeneity and inhomogeneity phantom in CUDA EGS for small field dosimetry
AIP Conference Proceedings 1817, 040011 (2017); 10.1063/1.4976796
Development of a Telepresence Robot for Medical
Consultation
Nilo T. Bugtai1, a), Aira Patrice R. Ong1, Patrick Bryan C. Angeles1, John Keen P.
Cervera1, Rachel Ann E. Ganzon1, Carlos A. G. Villanueva1, Samuel Nazirite F.
Maniquis1
1
Manufacturing Engineering and Management Department, De La Salle University - Manila
2401 Taft Avenue, Manila, Philippines 1004
a)
Corresponding author: nilo.bugtai@dlsu.edu.ph
Abstract. There are numerous efforts to add value for telehealth applications in the country. In this study, the design of a
telepresence doctor to facilitate remote medical consultations in the wards of Philippine General Hospital is proposed.
This includes the design of a robot capable of performing a medical consultation with clear audio and video information
for both ends. It also provides the operating doctor full control of the telepresence robot and gives a user-friendly
interface for the controlling doctor. The results have shown that it provides a stable and reliable mobile medical service
through the use of the telepresence robot.
Keywords: Telehealth applications; Telepresence robot; Remotely controlled; Medical consultation
INTRODUCTION
Medical consultation enables a doctor to investigate the patient’s illness, identify possible causes, and monitor
the rate of its progress. It provides the patient more awareness about his/her health condition through effective
communication with the medical specialist. There are instances, however, wherein the medical specialist needed is
currently unavailable. The patient could either wait for the doctor to arrive or travel in order to seek medical
attention.
Telepresence provides a solution to these problems. It is defined as the use of virtual reality technology,
especially for remote control of machinery or for apparent participation in distant events [1], allowing a person to be
present in the same place as the people he is interacting with. Telepresence has numerous applications, which
include video conferencing, performing tasks in hazardous environments, and even in fields such as education and
business. Not all countries are using telepresence robots in hospitals. The Philippines, for example, is still
developing attempts to implement such technology for medical applications.
The main objective of the study is to develop a remotely operated telepresence robot that could display the image
of the medical personnel, provide access to patient’s medical records, printout doctor’s prescriptions and is capable
of two-way video conferencing with clear audio and visual data as real time as possible through the use of a private
network.
The proposed telepresence robot will be used in the PGH specifically in the surgery wards for actual consultation
procedures whenever needed, especially in post-surgery medical rounds. The study is intended to aid and increase
the availability of medical specialists that cannot be physically present for consultation. The system is not intended
to directly acquire medical data through diagnostic equipment; rather, it will only act as a communication medium
for the medical specialists to perform consultation with their patients, with the assistance of a nurse or resident
doctor.
Biomedical Engineering’s Recent Progress in Biomaterials, Drugs Development, and Medical Devices
AIP Conf. Proc. 1817, 040002-1–040002-14; doi: 10.1063/1.4976787
Published by AIP Publishing. 978-0-7354-1485-3/$30.00
040002-1
Wireless communication is made possible by connecting both robot and control station to an access point [2].
The study follows a simple framework as shown in the Figure 1.
DESIGN CONSIDERATIONS
The telepresence robot system is composed of a robot and a workstation connected via wireless network, which
act as two endpoints capable of two-way communication and information exchange. The robot acts a receiver which
executes the commands sent by the workstation.
The telepresence robot system can be broken down into three main aspects: mechanical, electronic and software.
Network set-up is also an important factor to enable wireless operation.
Mechanical
The mechanical aspect of the robot mainly concerns the assembly structure and robot motion. The assembly
consists of (a) IP camera controlled by the workstation, (b) LCD screen displaying the doctor, (c) microphone, (d)
speakers, (e) receipt printer for prescriptions, (f) scanner for sending document images to the workstation, (g)
proximity sensors for obstacle detection, (h) motorized base containing the robot’s processor, circuits, power supply
and wheeling mechanism, and (i) access point for wireless communication, systematically assembled in a fiberglass
enclosure of 5 feet six inches in height [4]. For the circuits and electric components, the bottom of the robot is
designed with PVC layers for organization. For its motion, a tank-type mechanism or differential steering system is
used. The assembly of parts is illustrated in the Figure 2.
FIGURE 1. Conceptual framework
040002-2
FIGURE 2. Image of the actual telepresence robot.
Assembly and structure are the mechanical concerns of the workstation. The workstation, as shown in Figure 3,
has a dimension of 590 mm x 382 mm x 360 mm and is enclosed in a galvanized iron sheet with wheels to support
mobility. This device is composed of: (a) Tactical II system acting as viewing screen to see the patient on the other
end, (b) control screen displaying robot status such as direction of motion and patient database for creating
diagnostic reports and prescriptions, (c) communication devices such as a camera, microphone and speakers, (d)
remote control for connection and robot camera controls, (e) accessories such as keyboard, mouse and joystick, and
(f) access point in-charge of wireless communication with the robot. The size and design of the workstation is based
on the actual fitting of all its components.
FIGURE 3. Image of the actual workstation.
ELECTRONICS
The robot’s operation is made possible by circuits for motor control, obstacle detection and battery indicator
working hand-in-hand with a Pentium 4 processor. The researchers utilized an 8-way relay parallel port motor
controller for the robot motion for simplicity purposes, eliminating complexity involved in manipulating motor
functions and programming requirements [3]. A linked system of infrared (IR) sensors is used for navigation safety.
Four sensors are linked together as one front detection unit and six are linked as back side detection. Signals from
the IR sensor are channeled to the main program of the robot where it will send signal to the motor controller to cut
off the power to the motors should there be an object detected around the sensing perimeter of the robot. The robot
uses five 12-volt valve-regulated lead acid batteries to enable usage of 1-1.5 hours. These batteries are connected to
a power inverter to operate components working on AC power. A battery indicator located at the lower backside of
the robot indicates when the battery is full, half or near empty. It uses Zener diodes instead of a voltage divider
circuit. A green light will activate, indicating a fully charged battery from 12.8 V – 12 V. The amber light will soon
follow as soon as it reaches 50% of the required operating voltage which 12V and finally a red light will light if it
has reached an operating voltage of 11.6 V, which will then warn the user of a 10 min of battery life still remaining.
040002-3
For the workstation, the computer unit and Tactical II system are the only components for electronics. Fewer
circuits are used for operation since the workstation is to be operated by AC power in a stationary position. The
Tactical II system enables video-conferencing with the robot while computer unit processes all information
exchange such as motion control, printer commands and opening the patient database.
PROGRAMMING
A Java and PHP application is installed in the computer unit of the robot. These programs immediately run upon
computer start-up but are not visible to the users. In order to establish communication between two parties, the
program will implement a server-client concept similar to a telephone call. The server program is running inside the
robot while the client program is inside the workstation. The input directions for movement are sent to the robot
through Java, which in turn sends the signals of sensors and battery indicator to the workstation. Printing is also
incorporated in the java program, executing a thread that immediately prints the specific file one selected by the
operator of the workstation. The PHP application contains the patient database, in which the diagnostic report made
by the examining medical consultant is stored, as well as the document images scanned from the robot’s scanner.
Similar to the robot, a Java and PHP application is installed in the computer unit. The Java application will
handle the communication of controls for motion and the sensor and battery status. The workstation displays a
graphical interface for the operator’s ease of use. Once the robot receives a signal from its circuits, a corresponding
notification is displayed for the user. The PHP application is used for the Patient Database, enabling the doctor to
create and update patient information and receive data such as scanned images. Creating the interface, the
researchers used Knowledgeroot, a web-based application with content management similar to Blogspot and
Wordpress for simplicity. PHPtriad was installed to the computers to enable hosting of the web-based interface.
Since the webpage already has set functions, special features such as the patient database, printing and scanning are
incorporated to the interface through adding a PHP code in the content. The Patient Database works in a separate
application through any web browser connecting to the IP address of the server (robot).
Network Set-up
The main components of the workstation and robot are manually configured with an IP address to facilitate the
wireless operation. The robot has its own Ethernet switch where the codec, computer unit and access point is
connected. The workstation’s Tactical II component, computer unit and access point is connected to the PGH
network. Sending information wirelessly is made possible through the use access points. For this purpose, the
researchers made use of two (2) Cisco Aironet 1130 access points, capable of WiFi 802.11a and 802.11g, one
connected to the robot’s Ethernet switch and another placed in the area of the surgery wards which is connected to
the PGH network along with the workstation. During operation, the access point of the workstation, for example,
relays the command given by the workstation’s computer unit to the access point of the robot, consequently sending
the command to the robot computer unit for execution. The same concept applies for the other components. Figure 4
illustrates how the components communicate wirelessly.
FIGURE 4. Network connection between the robot and workstation
040002-4
RESULT AND DISCUSSION
Several tests were done to examine the reliability of the telepresence robot system in terms of network, speed,
safety, additional features and battery life. The results are collected and analyzed to determine the effectiveness of
the system.
Average Speed
The chart in Figure 5 displays, the noted range for the robot speed was 0.22 to 0.26 m/s, which is within the
acceptable objective speed of below walking speed (1.39 m/s). From the experiment, the type of surface does not
greatly affect the speed of the robot. One probable reason behind this is the adhesive reaction of the material of the
wheel (polyurethane) with smooth finish surfaces.
FIGURE 5. Network connection between the robot and workstation
The researchers have also implemented Z-test computations for much of the experiment data, with a set critical
level of +/- 2.326 at 0.01 confidence level as referenced in the T-chart. The chart in Figure 6 also indicates that all
surfaces do not gravely affect the robot speed.
FIGURE 6. Network connection between the robot and workstation
Average Robot Start-up Time
From the data, presented in Figure 7, it can be observed that the time difference among the three distances (2, 21,
and 32 meters) is only between 0.10-0.21 minutes with an average of 3.69 minutes. This proves the claim of the
manufacturer (Cisco) that provided the access point is within their specified effective range distance, the start-up
time is relatively consistent.
040002-5
FIGURE 7. Network connection between the robot and workstation
The Z-test values shown in Figure 8 are all within the critical level. Therefore, there is no significant difference
among the three distances.
FIGURE 8. Z Test Results for Range vs Start-Up Time
Ping Time & Packet Loss for Robot Motion
The data in Figure 9 shows that aside from the 32-meter open area test, all other setups of ping time tests varied
only by 0.13-5.3 milliseconds while averaging a time of 3.526 milliseconds. Given that no visible trend can be
inferred from tests on an open space to a thickly walled place, this would show that the network system transmits
relatively stable.
FIGURE 9. Network connection between the robot and workstation
040002-6
As shown in Figure 10, about three instances exceeded the 500-millisecond ping time threshold, and they were
relatively equidistant from each other. As any network connection is susceptive to “speed burps”, there can be
instances where ping time will suddenly rise, but these occur rarely and on a certain repetition trend. Also, the spikes
may be caused by interferences in the environment (i.e. low frequency sounds, radio frequencies, etc.) that got
intercepted in the network being used. Nevertheless, the “burps” only happened by about 4% of the test and is within
the target ping time of less than 1 second.
milliseconds
PC Ping Time of Open Area at 32 m
1600
1400
1200
1000
800
600
400
200
0
1
6
11
16
21
26
31
36
41
46
51
56
Trials
FIGURE 10. Time spikes of robot motion ping time
Results for relation between PC packet loss percentage and wall thickness show that there is only an average of
1% packet loss at the 32-meter open space setup. It can be observed that regardless of any wall thickness or distance,
so long as it is within the manufacturer’s claimed effective range, packet loss is insignificantly minimal.
The Z-test of PC delay demonstrated in Figure 11, showed significant difference on comparing the open test with
the walled test at the 32-meter setup. It can be inferred from this result that at a significantly long range, an enclosed
or open environment can give some considerable effect to the PC data delay time. In particular, when distance is
quite far, signals may bounce on the walls of an enclosed space and contribute performance unlike in an open area
where the signals will diffuse due to the distance. Hence, a closed area may provide better performance.
FIGURE 11. Z- Effects of open vs walled at various ranges for pc delay
040002-7
Reliability of Robot Motion
All four directional tests, shown in Figure 12, signify that the workstation consistently controls the robot on any
kind of setup and wall thicknesses at any direction, provided that they are within the manufacturer’s specified
effective range distance. This consequently proves that the robot is able to perform the transmitted command every
time without fail.
FIGURE 12. Motion repeatability on various setups
Reliability of Printing Function
The chart, presented in Figure 13, shows the repeated printing test for each of the open-area, thin-walled, and
thick-walled setups. All tests performed consistently within the manufacturer’s specified effective range distance,
with each printout having complete data transfer from workstation website to robot printer.
FIGURE 13. Printer repeatability test on various setups
040002-8
Reliability of Scanning Function
The chart, presented in Figure 14, shows the repeated scanning test for each of the open-area, thin-walled, and
thick-walled setups. All tests performed consistently within the manufacturer’s specified effective range distance,
with each scanned image having completely sent to the workstation website from the robot scanner.
FIGURE 14. Scanner repeatability test on various setups
Ping Time & Packet Loss for Robot Video and Audio
The chart, presented in Figure 15, shows something quite similar to the ping time data for robot motion. The
setups excluding the 32-meter open area have insignificant performance differences, ranging from 0.16-3.69
milliseconds with an average of all setups to be 2.9 milliseconds.
Having a 6% occurrence throughout the ping trials and only averaging an in-the-target value of 145.5
milliseconds, the speed burps were infrequent and were having a pattern tendency and there might be minor
interferences too, similar to the previous analysis for the robot motion. Likewise, this higher collection of ping time
only occurred in the 32-meter open setup, giving the presumption that the setup relatively stretches the capability of
the network connection.
FIGURE 15. Comparison of average ping time on various setups
040002-9
The 32-meter open space setup also produced some packet loss among all other setups, much similar to the
outcome for the packet loss of robot motion. It can be observed that packet loss is truly kept to a minimal at any wall
thickness or distances so long as covered within the manufacturer’s claimed effective range.
In this Z-test results chart, shown in Figure 16, it is evident that operating in the close setup at 32 meters
can have significantly better performance than one at an open area. This time however, the range distance also
becomes a significant factor. But if we take a look on the time variation between the two ranges on either the closed
or open setup, the differences only ranges from 5 milliseconds and lower.
FIGURE 16. Effects of open vs walled at various ranges for video & audio delay
Average Connection Time
All set-ups can be summarized within 9.85-10.34 seconds with a total average of 10.10 seconds. No significant
trend of deviations is observable among the incrementing distances or wall thicknesses. Interestingly, the Z-test of
this experiment does not reflect congruence among other previous Z-test charts. Nonetheless, we must again take
into consideration the diminutive unit size used for the range of the graph, which is in milliseconds. So in actuality,
the difference of performance in an open-vs-walled setup at 21 meters is just 0.44 milliseconds.
Obstacle Detection
The chart in Figure 17 plots the average maximum and minimum detection distance of each sensor used in the
telepresence robot. Among the sensors, sensing distance can be as far as almost 28 centimeters. The reason behind
the variation of detectable distances is mostly due to the differences of the sensor boards, being only manually
soldered and assembled, not to mention the possible discrepancies of the components.
040002-10
Minimum and Maximum Sensing Distance of Sensors
30
Centimeters
25
Maximum
Sensing
Distance
Minimum
Sensing
Distance
20
15
10
5
0
1
2
3
4
5
6
7
8
9
10
Sensors
FIGURE 17. Comparison of sensing distances among all sensors
In observation, the maximum values are high enough for the sensors to be evaluated as sensitive, while the
minimum values are not too low that the sensors are able to be specific in determining true obstacles. Furthermore, it
can be observed from Fig. 17 that the farthest detection range was achieved on sensors 2 and 7 (backside), then 1
and 8 (frontside) respectively. These positions are the usual blind spots of the robot device, hence to have the
experiment data show that they have the best range among all the sensors gives the best setup for the researchers and
future device users.
Battery Life
When the robot device is stationary, the consumption time is much longer compared to one in motion by 3.855.935 minutes (231-356 seconds) or an average of close to 5 minutes. This can be equivalent to 10-20 minutes.
Figure 18 shows that the motor operated at the longest time given any voltage difference variation while the
Cisco and other equipment worked at the shortest time. It can then be intuitively expected that the operating time is
inversely proportional to the consumption of the components to the battery power.
Given that the voltage differences on the four voltage setups are varied (i.e. 12.7-12.0 is 0.7 while 12.7-11.6 is
1.1, etc.), their total VAs also becomes varied. However, it is gravely misleading to conjecture or preclude that the
VA rating of the device is inversely related on the voltage drop that the battery can allow (i.e. on the 12.7-12.0 drop,
total VA 490.72 is while on 12.7-11.6 is 312.27, etc.). The proper observation on this data is actually that the total
VA rating is constant regardless of whatever voltage drop setup was used. Fig. 19 provides the researchers the
justification that using a four-unit battery array and setting its operating voltage from 11.8 volts to its full voltage of
12.7 volts is sufficient to power the robot for more than an hour.
040002-11
FIGURE 18. Equipment Operational Battery Time at Various Voltage Difference Ranges (right), Equipment Power
Consumption at Various Voltage Difference Ranges (left).
Battery Indicator
Table 1 shows how the LEDs of the indicator would light up on voltages of every 0.1 V increment. The battery,
although theoretically designed to indicate battery capacity in three discrete levels, actually performed more of a
fuzzy logic, as it can be observed that two LEDs may light up at a certain voltage, but the light intensity will vary
(i.e. on 11.8 V, the yellow will be much brighter than on 11.5 V, etc.). This performance may actually be
advantageous to the user as it can provide a better judgment of battery level.
040002-12
TABLE 1. Battery Indicator Light Test on Various Voltage Readings
BATTERY NOTIFICATION
Voltage
Red
Yellow
11
x
11.1
x
x
11.2
x
x
11.3
x
x
11.4
x
x
11.5
x
x
11.6
x
x
11.7
x
x
11.8
x
x
Green
11.9
x
12.0
x
12.1
x
12.2
x
12.3
x
12.4
x
x
12.5
x
x
12.6
x
x
12.7
x
x
12.8
x
x
12.9
x
x
13.0
x
FIGURE 19. Comparison of sensing distances among all sensors
In the Z-test graph in Figure 19, it is visibly clear that the battery indicator significantly shows voltage difference
on two working scenarios (moving and non-moving). It can be inferred that the battery indicator can positively and
consistently display precise voltage of the battery.
040002-13
CONCLUSION
The application of telepresence technology is fast expanding worldwide. In fact, countries that pioneered and
have such technology are already using it in order to address the increasing number of people who cannot be
physically present to attend meetings or perform professional consultation.
This paper presented the development of a telepresence robot fit for medical consultation in a Philippine
hospital. The telepresence system that consists of two (2) parts – the robot and the workstation - had been made to
support the Surgery Ward of the Philippine General Hospital (PGH) with regards to professional consultation.
The workstation was designed to manage most of the system’s functions. It is used as the robot’s main controller
for motion, camera viewing and teleconferencing with the patient via wireless connection. It also contains the
interface of the system used for hospital patients’ database.
The robot, on the other hand, was designed to interact with the in-house patients, specifically in the Surgery
Ward of PGH. This battery-operated device is equipped with a monitor, a camera, a printer and a scanner. These
components help make medical consultations easier and more convenient for both the doctor and the patient.
Various experiments had been done in order to test the reliability of the system. From motion testing to
networking, from lag tests to battery consumption monitoring, each experiment yielded results, which were justified
by the group’s standards. Therefore, the telepresence robot is proven to be successful and ready for implementation
in the environment of the Philippine General Hospital.
FUTURE DIRECTIVES
In any project, new ideas for improvement must be taken into consideration. With regards to the telepresence
system made by the group, there are several aspects that need consideration for improvement.
Without the limitation of working with a closed system, smaller components could have been used for lighter
load and slimmer dimensions.
All mobile devices have a common limitation: battery life. For TeD (telepresence doctor) more power-efficient
devices could be used, as well as acquiring batteries with higher energy density. It is recommended to find a battery
charging system that can improve recharge time without compromising battery life, or perhaps have backup battery
set with efficient battery slot-on system that can incorporate the backup battery and facilitate its replacement.
Future developments for the sensor system may also be taken into consideration. The available resources of
sensors limited the researchers. Using infrared sensors are also effective but the location of sensors must be placed
in such a way that there are no blind spots in the critical areas of the robot (front and back) and the sensing distance
would have enough distance to prevent collision without being too close to the obstacle.
Mobility of the workstation could also be improved. Given these limitations of having a closed-system, future
researchers must be able to utilize more mobile equipment to increase the mobility of the workstation so as to
expand the services that TeD can provide. Instead of creating a workstation, its function could be integrated into one
application.
Furthermore, other researchers could help improve the medical capabilities of TeD such as integrating diagnostic
equipment for additional features since its main purpose is for medical consultations.
ACKNOWLEDGEMENTS
The researchers would like to acknowledge the following people for the completion of this paper: CISCO
Systems Inc., Dr. Serafin Hilvano and the PGH Surgery Department, CTIS, Mr. Jhonatan Tapay and Mr. Rodelio
Barcenas.
REFERENCES
1.
2.
3.
4.
5.
Oxford, “Compact oxford english dictionary – Telepresence,” [Online]. Available:
http://www.askoxford.com/concise_oed/telepresence?view=uk [Accessed: June 8, 2010].
B. Wilson. “Wireless Network” n.d. [Online]. Available: http://computer.howstuffworks.com/wirelessnetwork1.htm#, Apr. 30, 2001. [Accessed June 16, 2010]
“Relays” [Online]. Available: http://www.kpsec.freeuk.com/components/relay.htm [Accessed July 26, 2011]
“Fiberglass”. [Online] Available: http://www.shanquan.com/fiberglass27.html [Accessed July 25, 2011
040002-14
Документ
Категория
Без категории
Просмотров
0
Размер файла
2 049 Кб
Теги
4976787
1/--страниц
Пожаловаться на содержимое документа