Sensor Drone for Aerial Odor Mapping for Agriculture and Security Services Theerapat Pobkrut1, Tanthip Eamsa-ard2, Teerakiat Kerdcharoen1,3,* 1
Department of Physics, Faculty of science, Mahidol University, Bangkok 10400, Thailand (
[email protected]) 2 Materials Science and Engineering Programme, Faculty of Science, Mahidol University Bangkok, Thailand (
[email protected]) 3 NANOTEC Center of Excellence at Mahidol University, National Nanotechnology Center, Bangkok 10400, Thailand (
[email protected]) * Corresponding author Abstract— In this work, an electronics nose (E-nose) based on six polymers and functionalized single walled carbon nanotube (SWCNT) nanocomposite gas sensors was developed and installed on a small unmanned aerial vehicle (UAV or drone) platform for detection of volatile compounds in the air. The efficiency of each gas sensor was tested in a static gas measurement chamber with presence of volatiles. The gas sensors were observed to increase response with increasing concentration of ammonia and toluene. Polyvinyl pyrolidon (PVP)/SWCNTCOOH shows the highest sensor response to both ammonia and toluene. The E-nose drone has then been demonstrated under two situations, i.e., in a closed clean room with presence of ammonia evaporation, and in open air with low wind environment. It was found that the pattern of sensor data obtained from flying the E-nose drone under different situations can be clearly distinguished. It is hoped that the E-nose drone can be a very useful technology for military usage; such as to detect explosives, as well as for farmers; such as to map the malodor emission from their cattle farms or to search for ethylene for fruit ripeness detection, etc. Keywords— Electronic nose; flying E-nose drone; UAV
I.
of drone by alter the rotation rate of one rotor disc or more. The advantages of quadcopter compare to others drone are low cost , steady flying, accurate position to handle works and most importantly the ability to carry various instruments to operate in the air. Some previous works have shown installation of various types of gas sensors on quadcopter or hexacopter to detect odor diffusion pattern [2, 3, 4] and find the odor source in many situations [5], but there are quite few works that employ sufficient number of gas sensor in an array to work as a so-called E-nose system.
(a)
INTRODUCTION
Autonomous systems are becoming more and more important technology in many fields due to the lack of labor and risk reduction of any hazardous cases in working procedures. Unmanned Aerial Vehicle (UAV), an aircraft with no pilot onboard, is commonly known as a drone and it is in use in several fields such as military, aerial surveillance, aerial photography, cargo delivery, forest fire detection, oil/gas and mineral exploration and manufacturing, etc. Recently, using drone as a remote sensing apparatus draws interest from researchers in numerous fields; various sensors to be equipped on drone have been demonstrated, e.g., electromagnetic spectrum sensors and gamma ray sensors to observe Earth’s atmosphere, biological sensors to detect the various airborne microorganism presences, and chemical sensors to analyze the concentration of interesting elements. Quadcopter is one type of drone that is lifted and propelled by four rotors on fix-wings [1]. It use two set of identical pitches propeller; clockwise (CW) and counter-clockwise (CCW), to control the movement
978-1-4673-9749-0/16/$31.00 ©2016 IEEE
(b) Fig. 1. (a) Drones are flying above the farm land and working like swarm and (b) Sensor drone is detecting the hidden explosive material.
On the other hand, many types of drone or UAV commonly used in military or security services are very expansive, hard to maintenance and required to be operated by professional user. Survey drone is one type of military drones used for detecting the dangerous targets by installed sensors such as infrared camera [6], motion detector [7], etc. It means that the targets must be seen by the drone. In case of hidden or invisible targets, it will be quite difficult for a visual survey drone to detect them. Installation of an E-nose system on the fix-wings
drone to detect and classify the chemical volatiles or odors is very useful method for seeking the hidden hazardous target [8] such as bombs and chemical weapons because explosive part of the weapon usually leaks some chemical volatiles which are detectable [9]. This technology will increase the performance of survey robot and very helpful for security service to reduce the damage from the explosion as shown in “Fig.1 (b)”. In agriculture, E-noses were applied in many applications such as detection of awful smell from swine farm [10], hazardous elements from food industry [11,12], quality grade identification of green tea [13], etc. E-nose consists of an array of various chemical gas sensors [14] that present distinct gassensing behaviors. The different selectivity of each gas sensor is very useful for classification and analysis of the odor data via pattern recognition techniques such as principal component analysis (PCA) [15]. Because a typical quadcopter cannot load a heavy thing, so the weight of E-nose system and other necessary equipment must be light enough. Among the available types of sensing materials applicable for E-nose system on mobile robot, polymer/carbon nanotube nanocomposite gas sensors are the most popular due to their high sensitivity, high reversibility, ease of maintenance and low power consumption [16,17,18] that helps reduce weight of the battery. However, development of E-nose on drone for agriculture and industry has been rare, and most E-nose on drone applications are still involved with the use of only one or two gas sensors. For such reasons, it is very important to develop a drone platform that integrates both E-nose and robotic technology together for field deployment. In this work, the applications of E-nose on drone have been demonstrated in order to explore the opportunities in the field of precision agriculture, security and environmental monitoring. II.
EXPERIMENTAL METHODS
A. Drone setup and Electronic nose module. To install electronic nose system on the drone we have to modify the drone’s frame and arrange the gimbal and camera holder to attach the e-nose chamber. Auto-pilot software was used to control the flying course of the drone, as well as an inhouse Labview program was developed for collection of the enose data. Configuration of the drone will be optimized for electronic nose applications. Flight parameters, e.g., altitude, positioning data, wind speed and direction, etc., from the sensors attached on drones are very useful for analysis of the odor data. B. Design and test of the E-nose system
Fig. 2. Drone with E-nose chamber and navigation system
The flying robot, a QuadCopter from 3DRobotics, consists of two main systems; e-nose system and flying system, as shown in “Fig. 2”. Arduino mega 265 is used to control the enose system and for receiving the electrical response of the polymer gas sensors which are listed in Table 1. This type of gas sensors, polymer/functionalized-SWNT, is quite suitable to be used in drone because of its affordability, high sensitivity, high reversibility, low susceptibility to environmental disturbance such as temperature and humidity, and most importantly low power consumption resulting to increasing operation time. These gas sensors are enclosed in a chamber attached underneath the drone’s frame in order to avoid the turbulence generated by the drone’s propellers that may affect gas sensor response. The chamber is horizontally mounted at the gimbal part at the bottom of drone’s body as shown in Fig. 2. The odor data are sent in real time to a computer via Zigbee wireless communication for collecting, analyzing and visualizing the results. The gas sensing materials are nanocomposites between functionalized single-walled carbon nanotubes (SWNT) and various types of polymers. Thus, SWNT is working as conducting materials while polymer serves as the supporting matrix and adsorbent to the analyte gases. Table 1 lists the types of functionalized SWNT and polymers used to fabricate the nanocomposite gas sensors. To prepare the gas sensors, the SWNT (1-2 nm in diameter) was dispersed in the polymer solution, previously prepared by dissolving the polymer in the proper solvent (acetone, ethanol, and water). Put the SWNT/polymer solution bottle in an ultrasonic bath for 15 min to obtain uniform mixture. Then, the solution was deposited onto interdigitated electrodes by spin-coating and annealed at 150 ºC for 3 hours to remove the residual solvent and impurities. TABLE I.
THE FUNCTIONALIZED GROUPS OF SWNT AND THE POLYMERS THAT ARE USED FOR FABRICATING GAS SENSORS
Sensor
Functionalize d SWCNT
1
-COOH
Polyvinyl chloride (PVC)
2
-COOH
Poly(styrene-co-maleic acid) partial isobutyl/methyl mixed ester (PSE)
3
-COOH
Polyvinyl pyrolidon (PVP)
4
-OH
Polyvinyl chloride (PVC)
5
-OH
Polyvinyl alcohol (PVA)
-NH2
Poly(2-Acrylamido-2-Methyl-1Propanesulfonic acid-co Acrylonitrile 95%)
6
Polymer
The performance of gas sensors at room temperature were tested within a static gas measurement chamber with some common volatiles such as ammonia and toluene as shown in Fig. 3. The sample concentration was varied between 10 – 150 ppm. The initial baseline resistance of each sensor was recorded for 2 minutes prior to the injection of the sample into the chamber. Then, the sensing response was observed for another 5 min after injecting the sample. The percentage change of individual gas sensor resistance was determined as
sensor response related to the volatile concentration. The gas sensor response can be calculated using the following equation: S%change = (Vsample – Vref)/Vref x 100
(1)
Where S%change is percentage change of the sensor response. Vsample is voltage across gas sensor exposed to the sample odors. and Vref is voltage across gas sensor exposed to the reference air. Gas inject
DAQ USB Chamber
Fig. 3. Schematic diagram of the experiment setups for detecting sample in the static measurement
The change of each gas sensor’s electrical property is measured by a voltage divider circuit using 10 kΩ constant resistances where each signal wire is connected to 10 bit analog pin of Arduino Mega 256. Gas sensor response refers to the analog signal that is converted into a voltage across the gas sensor and it depends on sensor’s resistor which will change when the sensing material interacts with the odor molecules. Each gas sensor response was plotted versus time and will be demonstrated in the next section. The navigation system of the quadcopter is based on the APM Mission Planner, a freeware control interface that can access all of the flying data, e.g., moving speed and direction, GPS position, and Altitude, as shown in “Fig. 4.” Moreover, path planning can also be setup via Mission Planner combined with the GPS map from Google Earth software to specify the test area and flying route. An ultrasonic landing sensors, gyromagnetic, and accelerometer were used for autonomous flying with real time data communication with the ground station application using telemetry module via 2.4 GHz radio frequencies. C. Experimental setup The E-nose system equipped quadcopter was experimented when weather at the time was clear with low wind velocity. We have done the experiments at 3 different locations; (1) a closed clean room, (2) a closed room with a 30%wt ammonia solution open container, and (3) an open-air clean area. To detect an odor data from the source, we have to measure wind speed and direction around the odor source to choose the most suitable position and attitude of the drone related to that condition which are 1.5 m, 1.5 m, and 15 m above the target points, respectively. Quadcopter flied at the determined attitude and position for 5 minutes to record odor data and send to a laptop computer via Xbee wireless network every second for one cycle of each sensor’s resistance measurement in order to be plotted in real-time by an in-house Labview software.
III.
RESULTS AND DISCUSSION
In this section, E-nose system on quadcopter will be presented. The gas sensor array was designed to detect and measure various volatile organic compounds commonly released from agricultural and industrial processes. The autonomous drone was designed and configured to use E-nose system for surveying and detecting VOCs in the air. E-nose quadcopter can autonomously fly with a predetermined route facilitated by a ground control program. The results from this demonstration can be given as follows. Drone has been demonstrated that it can steadily fly in autonomous mode when equipped with E-nose module. To improve flying property, we have to balance the weight that load on the quadcopter and limit the weight of all stuffs which it carry not over the 500g maximum loading weight. After finishing the hardware configuration, we also have to program and configure some of the flying algorithms such as PID control of pitch, yaw, and row, manual control, etc. In principles, we have to keep the center of mass of all loads to be low and at the center comparing to the drone’s body. Nevertheless, we cannot put every tool in a balance location so we have to add some unimportant weight and make them balance as much as possible. Therefore, the flying e-nose drone now can steadily fly in the air and can perfectly perform autonomous flying via mission planner software. The autonomous flying drone equipped with e-nose module was used to measure the VOCs contained in the air at 3 different locations simulating the real situations. Fig.4 shows the route, altitude and velocity of the drone as preset by a ground station application. Odor data from E-nose module was obtained wirelessly via XBee wireless network and plotted using a Labview program.
Fig. 4. Route plan for autonomous flying using ground station application on a laptop computer (left) and real-time odor data from E-nose module (right).
Performance of each gas sensor was tested in a static gas measurement chamber and their average of percent change in the resistances of six gas sensors as exposed to ammonia and toluene are shown in Fig.5. It was demonstrated that the gas response is higher with increasing amine and toluene concentration. All sensors yield better response to amine than toluene and SWNT-COOH with polyvinyl pyrolidon have the highest response to both volatiles. The sensing mechanism of the composite sensors explained in terms of percolation theory. The analyte gases diffuse into the porous materias and interact weakly with the polymer and carbon nanotube through dipoledipole and van der Waals interactions. Thus, all composite
sensors employed in this study yield somewhat response to ammonia and toluene.
Fig. 7 show the sensor data from the E-nose drone flying in the open air with low wind environment.
Fig. 5. The average of percent change in resistance of each gas sensor when exposed to ammonia and toluene at the concentrations of 10-150 ppm. Sensors 1-6 are SWCNT-COOH with PVC, SWCNT-COOH with PSE, SWCNTCOOH with PVP, SWCNT-OH with PVC, SWCNT-OH with PVA, SWCNT-NH2 with Poly(2-Acrylamido-2-Methyl-1-Propanesulfonic acid-co Acrylonitrile 95%), respectively.
After the performance test of each gas sensor in a static gas chamber, we have to test the real-world applications by installing the gas sensor into the E-nose module on the drone. Identification of a baseline or reference signal of the gas sensors was done by flying the drone at 5 meters above the floor of clean closed room without ventilating air for 5 minutes and then put the 30%wt ammonia solution on the floor ground and record the odor data for another 5 minutes. Sensor response of each sensor was plotted as shown in Fig. 6. This first 5-minute graph of sensor response is quite linear because there is no volatile interacting with the sensors, but as soon as we introduced the ammonia solution into the room, it was clearly seen the change of sensor response. Thus, each sensor has different response to the ammonia depending on their characteristics.
Fig. 6. The odor data from the E-nose drone during the first 5 minutes in clean air and the next 5 minutes exposed to the atmosphere in presence of the 30%wt ammonia solution bottle.
Fig. 7. The odor data from the E-nose drone in reatime measurement of 5 minute autonomously fly above open-air with low wind velocity environment.
These graphs clearly demonstrate that the volatiles in the open air environment yield different pattern from closed clean room in the presence of ammonia gases. Thus, the E-nose drone is capable of discriminating clean air from the contaminated environment. IV.
CONCLUSION
The E-nose module based on six polymers/SWCNT gas sensors was installed and tested on quadcopter. This system was found to detect volatile compounds such as ammonia and toluene. In the static gas measurement, the sensors were observed to increase response with increasing concentration of ammonia and toluene. Polyvinyl pyrolidon (PVP)/SWCNTCOOH shows the highest sensor response to both ammonia and toluene. The E-nose drone has been demonstrated under two situations, i.e., in a closed clean room with presence of ammonia evaporation, and in open air with low wind environment. The sensor responses from all sensors under different situations are clearly distinguished. Therefore, if we collect enough characteristic odor data from various places into the database, we can identify irregular situation when new data pattern shows difference from the database. In the future, we can develop the new sensing materials for high performance gas sensors in order to detect amine, toluene and other volatile gases as present in the compounds of chemical warfare and explosive materials. In fact, toluene is an organic substance concentrated in many explosives, so we can use it as a trace chemical for detection of the explosive. Apart from military usage, the E-nose drone can be a very useful technology for farmers in many applications, such as to map the malodor emission from their cattle farms, to search for ethylene for fruit ripeness detection, etc. The gas sensor performance, efficiency, durability, selectivity to target gases of interest as well as
pattern searching algorithms will be further improved to a achieve high quality E-nose drone in the future. ACKNOWLEDGMENT The authors would like to thank Mahidol University, National Nanotechnology Center (NANOTEC) and Development and Promotion of Science and Technology Talents Project (DPST) for their support to this research project. Authors also would like to thank Chayanin Khunaruk and Satetha Siyang that collaborated to the construction of the e-nose drone.
[7]
[8] [9] [10]
[11]
REFERENCES [12] [1]
[2]
[3]
[4]
[5]
[6]
X. Zhang, X. Li, K. Wang, and Y. Lu, “A Survey of Modelling and Identification of Quadrotor Robot,” in Abstract and Applied Analysis, vol. 2014, 2014, pp. 16. P.P. Neumann, S. Asadi, A.J. Lilienthal, M. Bartholmai, J.H. Schiller, “Autonomous Gas-Sensitive Microdrone: Wind Vector Estimation and Gas Distribution Mapping,” in Robotics & Automation Magazine, vol.19, no.1, March 2012, pp.50-61. V.M.H. Bennetts, A.J. Lilienthal, A.A. Khaliq, V.P. Sese, and M. Trincavelli, “Towards real-world gas distribution mapping and leak localization using a mobile robot with 3d and remote gas sensing capabilities,” in Proceedings of IEEE International Conference on Robotics and Automation (ICRA), May 2013, pp. 2335-2340. V.M.H. Bennetts, A.J. Lilienthal, P.P. Neumann, and M. Trincavelli, “Mobile Robots for Localizing Gas Emission Sources on Landfill Sites Is Bio-Inspiration the Way to Go,” in Frontiers in Neuroengineering, vol. 4, 2011, pp. 1-12. M. Bartholmai and P. Neumann, “Micro-Drone for Gas Measurement in Hazardous Scenarios via Remote Sensing,” in Proceedings of the 6th WSEAS International conference on remote sensing (REMOTE '10) Selected topics in power systems and remote sensing (2010), WSEAS Press, Iwate Prefectural University, Japan, 2010, pp. 149-152. C.C.D. Lelong, P. Burger, G. Jubelin, B. Roux, S. Labbé, and F. Baret, “Assessment of Unmanned Aerial Vehicles Imagery for Quantitative Monitoring of Wheat Crop in Small Plots,” in Sensors, vol. 8, 2008, pp. 3557-3585.
[13]
[14]
[15]
[16]
[17]
[18]
N. Dijkshoorn, and A. Visser, “Integrating Sensor and Motion Models to Localize an Autonomous AR. Drone,” in International Journal of Micro Air Vehicles, vol. 3, 2011, pp. 183-200. L. Senesac, and T. G. Thundat, “Nanosensors for trace explosive detection,” in Materialstoday, vol. 11, 2008, pp. 28-36. J.S. Caygill, F. Davis, and S.P.J. Higson, “Current trends in explosive detection techniques,” in Talanta, vol. 88, 2012, pp. 14-29. P. Lorwongtragool, C. Wongchoosuk, and T. Kerdcharoen, “Portable artificial nose system for assessing air quality in swine buildings,” in Proceedings of the 7th Annual IEEE International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (IEEE ECTI-CON 2010), Chaing Mai, Thailand, May 2010. E. A. Baldwin, J. Bai , A. Plotto and S. Dea, “Electronic Noses and Tongues: Applications for the Food and Pharmaceutical Industries,” in Sensors, vol. 11, 2011, pp. 4744-4766. P. Lorwongtragool, A. Wisitsoraat and T. Kerdcharoen, “An Electronic Nose for Amine Detection Based on Polymer/SWNT-COOH Nanocomposite”, in Journal of Nanoscience and Nanotechnology 11, 2011, pp. 10454-10459. H. Yu, and J. Wang, “Discrimination of LongJing green-tea grade by electronic nose” in Sensors and Actuators B: Chemical, vol. 122, 2007, pp. 134-140. T. Seesaard, P. Lorwongtragool, and T. Kerdcharoen, “Development of Fabric-Based Chemical Gas Sensors for Use as Wearable Electronic Noses,” in Sensors, vol. 15, 2015, pp. 1885-1902. C. Wongchoosuk, M. Lutz, and T. Kerdcharoen, “Detection and Classification of Human Body Odor Using an Electronic Nose,” in Sensors, vol. 9, 2009, pp. 7234-7249. S. Siyang, T. Seesaard, P. Lorwongtragool, and T. Kerdcharoen, “E-nose based on metallo-tetraphenylporphyrin/ SWNT-COOH for alcohol detection,” in Proceedings of IEEE International Conference on Electron Devices and Solid-State Circuits (EDSSC), 2013, pp.1-5. C. Wongchoosuk, A. Wisitsoraat, D. Phokharatkul, M. Horprathum, A. Tuantranont and T. Kerdcharoen, “Carbon Doped Tungsten Oxide Nanorods NO2 Sensor Prepared by Glancing Angle RF Sputtering”, in Sensors and Actuators B, vol. 181, 2013, pp. 388-394. Y. Seekaew, S. Lokavee, D. Phokharatkul, A. Wisitsoraat, T. Kerdcharoen and C. Wongchoosuk, “Low-Cost and Flexible Printed Graphene-PEDOT:PSS Gas Sensor for Ammonia Detection”, in Organic Electronics, vol. 15, 2014, pp. 2971-2981.
