Development and application of electronic nose for agricultural robot Satetha Siyang1, Panida Lorwongtragool2, Atirach Noosidum3, Chatchawal Wongchoosuk4, Teerakiat Kerdcharoen* 1
2
Material Science and Engineering Programme, Faculty of Science, Mahidol University, Bangkok, Thailand Faculty of Science and Technology, Rajamangala University of Technology Suvarnabhumi, Nonthaburi,Thailand 3 Department of Entomology, Faculty of Agriculture, Kasetsart University, Bangkok, Thailand 4 Department of Physics, Faculty of Science, Kasetsart University, Bangkok, Thailand * NANOTEC Center of Excellence at Mahidol University, National Nanotechnology Center, Thailand 1
[email protected] [email protected] 3
[email protected] 4
[email protected] *
[email protected] 2
Abstract— A portable electronic nose (E-nose) based on metal oxide gas sensor was used for detection of a total volatile compound in soil headspace and greater wax moth (Galleria mellonella L.). Sensor elements were selected in order to possibly respond to a wide range of the volatiles including both of oxidizing and reducing gases. The aim of this work was to relate the soil volatile fingerprints at their depth levels and less number of greater wax moth that can be detected by electronic nose. Soil was sampled from the field of the vinery located near Khao Yai world heritage, Thailand at a depth level of 10 cm, 30 cm and 50 cm. Male and female greater wax moths were treated from egg to senile adult in plastic box and measured with electronic nose by varying number of moth. The results could identify the difference of the volatile fingerprints by employing principal component analysis (PCA) as the signal pattern recognition technique. The volatile profiles from eight TGS elements relating chemical emissions will be discussed in more details. Keywords— Soil volatile fingerprint; Greater wax moth; Electronic nose; TGS sensor; PCA
I. INTRODUCTION Nowadays, there has been an increased interest in smart farm technology since an increase of the world’s population leading to an increment of food demand. From the report of UN population Division, the world’s population is projected to increase from 5.6 billion in 2009 to 9.8 billion in 2050 [1]. A team from the University of Minnesota predicted that food demand will increase by 100% in 2050 [2]. To prevent food shortages in the future, smart farm technology needs to be developed. There are many technologies that were used in agriculture such as sensor technology, GPS, GIS and a computer [3]. One of important technology in the future is the smart robot for agriculture. By combined with technologies, a smart robot can work more efficiency for agriculture over people. Autonomous swarm robots may become popular future trends in precision agriculture as show in Fig. 1. Electronic nose (E-nose) is an interesting technology that has widely used in many fields including biomedical, cosmetics, environmental, pharmaceutical, military, food, manufacturing, and agricultural research fields [4]. The most 978-1-4799-0545-4/13/$31.00 ©2013 IEEE
common application of electronic nose in agricultures is to detect diseases, to identify insect infestation and soil volatile fingerprints, and to monitor food quality. For example, the E-nose was used for sampling the headspace volatiles of three different soils. The result showed a clear discrimination between soil types [5]. In another example, Y. B. Lan et al. [6] reported the identification of insect by using E-nose. The results showed significant potential for identifying stink bugs and the E-nose can classify the species and gender of stink bug samples. However, the development of E-nose for using in robot is stills lack and most E-nose for agriculture applications can only work in laboratory. Therefore, development of E-nose for robots with agricultural research fields is still necessary. In this paper, we have developed an E-nose for smart robot in agriculture. This E-nose was tested for detecting different soil volatile fingerprints in each depth level and insect pests. It should be noted that the quality of the soil and pest control play important role for increasing agricultural products. If the E-nose can detect different soil and insect pest in real time and open air condition, it can be the sensing part of smart robot leading to be an emerging technology for precision agriculture.
Fig. 1 The future trend in agriculture “Smart Farm”
II. MATERIAL AND METHOD A. Collection of Soil Soil samples were collected from the field of the vinery located near Khao Yai world heritage, Thailand down to the depth of 10 cm, 30 cm and 50 cm. The 50 g of soil without any treatment was put in the closed glass bottle before placed in the sample holder of the E-nose. B. Treatment of Greater Wax Moth Male and female greater wax moth (Galleria mellonella L.) were treated from egg to senile adult in plastic box that has honey comb as show in Fig. 2. Note that honey comb is the main food and residence for the larval stage of the moths. Adult moths (two weeks after larval stage) were used to test the efficiency of E-nose for insect pest detection. In this study, alive greater wax moths were measured in dynamic flow without any injury.
The condition of flow rate, reference time, and sample time of E-nose on experiments are summarized in table II. In this experiment, we used open clean air as the reference gas. The percent change in resistance of an individual sensor was determined as sensor response related to the total volatile concentration. The sensor response of gas sensor was calculated by the following equation; Where and
Si = ( R0 - Rs ) / R0
(1)
Si is sensor response of gas sensor R0 is resistance of sensor in open clean air Rs is resistance of sensor in sample odors
Principal component analysis was employed to identify the soil volatile patterns which may reflect the difference of gases and their quantities due to the depth of soil. TABLE II CONDITION OF EXPERIMENT
Soil
Flow rate (l/min) 0.5
Reference time (min) 4
Sample time (min) 1
Greater wax moth
0.7
5
1
Sample
III. RESULTS AND DISCUSSION
Fig. 2 The treatment of greater wax moth in plastic box
C. Electronic Nose The soil and greater wax moth odor in the containers were flowed to the gas sensor chamber containing different eight sensor elements. Details of the E-nose system were given in the literature [7]. The eight gas sensors based on sensing element of metal oxide semiconductor from Figaro USA, Inc. with difference in the response to wide range of the target gases are shown in Table I. TABLE I TARGET GAS OF TGS SENSOR USED IN THIS WORK
Sensor
Target gas
TGS 821
Hydrogen gas
TGS 822
Organic solvent vapors and other VOCs
TGS-825
Hydrogen sulfide
TGS 826
Ammonia and other VOCs
TGS 2600
Air contaminants (ethanol, iso-butane, hydrogen)
TGS 2602
VOCs and odorous gases
TGS 2610
Liquefied petroleum (LP) gas and its component
TGS 2620
alcohol and organic solvent vapors
A. Detection and classification of Soil The portable E-nose has been demonstrated that it could be used to study in the soil volatile fingerprints. Currently, a few researches have focused on detection of VOCs by using an E-nose [8]-[10]. Soil was reported that it may act as both source and sink of various volatile organic compounds (VOCs) resulting from metabolic activity of microorganisms and atmospheric gases [10]. With the working principle of the E-nose system in this work having a switching of reference and sample parts, the output signal could be reduced the effect of atmospheric volatiles which possibly diffuse into the soil samples during investigation. This trouble was mentioned by Hernández, et al. [11] while the E-nose could overcome the trouble concerning about the effects of atmospheric gases and the soil permeability due to the types of soil. Soil sample preparation can be performed easily without any extraction step. However we still go along with other researchers who concern with quantitative analysis that the soil should be carefully stored in the evacuated container and considering about the effect of soil permeability. Based on qualitative analysis using the portable E-nose, the soil volatile profiles due to the different depth levels are shown in Fig. 3. Positive and negative signals are referred to the effect of oxidizing and reducing gases to the sensing elements of metal oxide semiconductor gas sensor. In this work, the sensing elements of all TGS sensors are intrinsically n-type semiconductors when the reducible gases such as H2, CH4, CO, C2H5, or H2S are adsorbed on the surface resulting in the electrical conductivity to increase due to increase of the carriers. In the opposite way, the oxidizing gases such as O2, NOx, CO2 or the noble gases tend to remove free electron
from the surface, consequently the electrical conductivity is decreased. Means of the E-nose is detection of total volatile including both reducing and oxidizing gases. Note that the percent response indicated in the Fig. 3 was determined from a change in resistance of an individual sensor therefore its negative signal represents existing reducing gases in the soil samples. In the case of the sensors such as TGS821, TGS2600 and TGS2610 with highly response to hydrogen gas, methane and LP gases (propane and butane), they have demonstrated that down to the vertical depth the soil could be indicated about increasing of these gases. Methane available in soil which is a natural and potential greenhouse gas is the end product of the anaerobic decomposition of organic matter [12]. In 2002, the investigated results by Sheppard and Lloyd [13] could observe a significant increase of CH4 concentration according to the depth profile. The indicated hydrogen gas possibly relates to the land around the vinery which was the area of old volcanic rock. The soil sample was found that it contains with many small pebbles (10% - 60%) therefore we have presumed that available hydrogen gas might be the product of water reaction [11]-[14].
reducing gases. Consequently their responses could potentially reflect to the efficiency of classification. The sensor responses were used as the important feature to process pattern analysis. Fig. 4 shows the three dimensional plot of principal components based on PCA. One point on the PCA plot represents a soil volatile fingerprint from the eight sensors. The output signals responding to the soil volatile could be clearly classified into three groups. Six points in each cluster come from repeating measurements. The variances in the 1st to 3rd principal components for all soil volatile fingerprints due to various soil depths were accounted as 77.6%, 17.6% and 3.5%, respectively.
Fig. 4 Three dimensional plot of PCA of soil volatile fingerprints due to the various depths
B. Detection of Greater wax moth To study the sensor that has highest response for detecting greater wax moth, the average percent sensor responses were calculated and plotted against with each Figaro sensor as show in Fig. 5. The result shows that TGS 826 has highest sensor response while an increasing of insect sample can increase Fig. 3 Precents sensor responses of eight elements used in the lab-made percent sensor response (based on TGS 826). From the data sheet of Figaro sensor shows that this sensor can detect E-nose system to soil volatiles at the different depths ammonia and alcohol gases. From the early study about According to the results in the sensors of TGS822, TGS825, chemical releasing from greater wax moth, male greater wax TGS826, TGS2602 and TGS2620, they could indicate an moth will emit plenty amount of pheromone, undecanal and cisincrease of oxidizing gases rather than reducing gases at 11-octadecenal.These chemicals stimulate female behaviour certain depths. These sensors were noted that they could over long distance [16]. Such chemicals are alcohol. This is respond to a wide range of volatile compounds such as the reason of high increment of sensor response. hydrogen sulphide, ammonia, organic solvent vapours and In nature, greater wax moth mainly flies at night. During day other VOCs which compose both of reducing and oxidizing light, they rest in dark places. Sometime they rest together but gases. Among these volatiles, carbon dioxide (CO2) which is sometime they rest only one [17]. From this information, if we one of oxidizing gases has been demonstrated in the literature want to find them by E-nose. First of all, we need to know the that it strongly significant increase in concentration when lowest number of greater wax moth that can detect by E-nose. observed down to vertical depth [8]. In Fig. 3, The TGS825 The E-nose was trained to differentiate between the presence could highly response to some oxidizing gases however its of greater wax moth by varying a number of presence moth datasheet does not inform about the response to any gas from absence of moth2, 4, and 6. The PCA result shows the two except hydrogen sulfide [15]. main grouping. The PCA points of empty bottle, 2 and 4 insects Based on the soil volatile measurement, we would recommend occur at the same region while PCA points of 6 insects clearly that the sensing unit should be selected to respond to a wide range separate from each other. It can imply that our E-nose can well of volatile compounds and their concentrations. In this work, we detect greater wax moths when numbers of moths are 6. used the TGS sensors which could response both oxidizing and Although this number is too high for real world used, the Enose can show a possibility of detection of insect pests by odor.
REFERENCES [1] [2] [3] [4] [5] [6] [7] Fig 5. Percent sensor responses of eigth elements used in the E-nose system to greater wax moth odor.
[8]
[9] [10] [11] [12] [13] [14] [15] Fig. 6 Two dimension plot of PCA of insect sample due to the various numbers
IV. CONCLUSION The E-nose has successfully detected the different soil volatile fingerprints in each depth level and insect pests. The commercial TGS gas sensors have been used as the sensing section and they have demonstrated that down to the vertical depth the soil could be indicated about increasing of hydrogen gas, methane gas and LP gases (propane and butane). Results show that an increase of oxidizing gases is rather than that of reducing gases at certain depths. Moreover, the E-nose can classify the soil volatile fingerprints at any depths level. By varying the number of insect samples, the number of moth is an important factor for detection and classification of the greater wax moth. The minimum number of insect that E-nose can be well-detected is around 6. This E-nose will be installed in robot in near future.
[16] [17] [18]
[19] [20] [21] [22]
[23]
ACKNOWLEDGMENT This research was supported by Mahidol University and [24] National Nanotechnology Center (grant no. P-12-01157). C.W. gratefully acknowledges the funding supports from Kasetsart [25] University Research and Development Institute (Ref. Number 5610186000/2556).
(2009) The United Nations website. [Online]. Available: http://www.un.org/esa/population/publications/wpp2008/pressrelease.pdf D. Tilman, C. Balzer, J. Hill, and B. L.Befort, “Global food demand and the sustainable intensification of agriculture,” PNAS, Oct. 12, 2011 Agri Technica special for smart farming, 2011 A. D. Wilson, and M. Baietto, “Applications and Advances in Electronic-Nose Technologies,” Sensor 2009, vol. 9, pp. 5099-5148, June 2009. A.C. Bastos, N. Magan, Soil volatile fingerprints: Use for discrimination between soil types under different environmental conditions, Sensors and Actuators B 125 (2007), pp. 556–562. Y. B. Lan, X. Z. Zheng, J. K. Westbrook, J. Lopez, R. Lacey, and W. C. Hoffmann, “Identification of Stink Bugs Using an Electronic Nose,” Journal of Bionic Engineering Suppl., pp 172-180, 2008 C. Wongchoosuk, M. Lutz, T. Kerdcharoen, “Detection and Classification of Human Body Odor Using an Electronic Nose”, Sensors, 9, 2009, p.7234-7249. C. Arnold, et al., Sub-surface probe module equipped with the Karlsruhe Micronose KAMINA using a hierarchical LDA for the recognition of volatile soil pollutants, Sensors and Actuators B 116 (2006), pp. 90–94. F. De Cesare, et al., Use of electronic nose technology to measure soil microbial activity through biogenic volatile organic compounds and gases release, Soil Biology & Biochemistry 43 (2011), pp.2094-2107. H. Insam and M. S. A. Seewald, Volatile organic compounds (VOCs) in soils, Biol Fertil Soils 46 (2010), pp.199–213. P. Hernández, et al., Soil gas CO2, CH4, and H2 distribution in and around Las Cañadas caldera, Tenerife, Canary Islands, Spain, Journal of Volcanology and Geothermal Research 103 (2000), pp. 425-438. E. Topp and E. Pattey, Soils as sources and sinks for atmospheric methane, Can. J. Soil. Sci. 77, pp. 167-178. S.K. Sheppard, D. Lloyd, Effects of soil amendment on gas depth profiles in soil monoliths using direct mass spectrometric measurement, Bioresource Technology 84(2002), pp. 39–47. R. Sugisaki, Geochemical features of gases and rocks along active faults, Geochemical Journal 14 (1980), pp. 101-112. Figaro product information TGS 825, Figaro USA Inc. website. [Online]. Available: http://www.figarosensor.com/products/825pdf.pdf [Accessed on Nov 17, 2012] H. G. Spangler, M. D. Greenfield, and A. Takessian, “Ultrasonic mate calling in the lesser wax moth,” Physiological Entomology 1984, vol. 9, pp. 87-95. D. A. Cushman, “Life Cycle of the Lesser Wax Moth (Achroia grisella),” 2012 W. G. Henderson, A. Khalilian, Y. J. Han, J. K. Greene, D. C. Degenhardt, “Detecting stink bugs/damage in cotton utilizing a portable electronic nose,” Computers and Electronics in Agriculture, vol. 70, pp. 157-162, Sept. 30, 2009 G. C. Rains, J. K. Tomberlin, and D. Kulasiri, “Using insect sniffing devices for detection,” Trends in Biotechnology, vol.26, Mar. 28, 2008. R. M. Negri, D.L. Bernik, “Tracking the sex pheromone of codling moth against a background of host volatiles with an electronic nose,” Crop Protection, vol. 27, pp. 1295-1302, Feb. 29, 2008. B. Zhou, and J. Wang, “Detection of insect infestations in paddy field using an electronic nose,” Int. J. Agric. Biol., vol. 13, pp. 707-712, 2011. D. C. Degenhardt, J. K. Greene, and A. Khalilian, “Temporal dynamics and electronic nose detection of stink Bug-induced volatile emission from cotton bolls,” Hindawi Publishing Corparation Psyche, Jan. 12, 2012. W. A. Snedden, C. R. Tosh, and M. G. Ritchie, “The ultrasonic mating signal of the male lesser wax moth,” Physiological Entomology, vol. 19, pp. 367-372, 1994. A. Perera, T. pardo, T. Sundiae, R. Gutierrez-Osuna, S. Marco, and J. Nicolas, “Electronic nose for remote bad odour monitoring system in landfill sites,” H. Zhang, and J. Wang, “Detection of age insect damage incurred by wheat with an electronic nose,” Journal of Stored Products Research, vol. 43, pp. 489-495, Jan. 8, 2007.
