Machine Olfaction

The goal of this work was to enable a mobile robot to navigate autonomously towards a stationary odour source with the help of a sense of smell. Two electronic noses, each containing a set of gas sensors, mounted on top of a Koala mobile robot were used for detection of the odour. The sensing strategy used for data collection was investigated in order to reduce the influence of air turbulences on the sample handling process. Then a multi-layer artificial neural network was used to learn both the direction to the source and the required turning speed of the robot. An experimental validation was carried out to evaluate the performance of the complete system.

The responses generated by a gas sensor array are difficult to classify due to their inherent imprecision, uncertainty and the procedures of computational intelligence are appropriate to deal with such imperfect knowledge. In recent years, rough set theory has attracted more attention of many researchers even though it was proposed in the early 1980's by Z. Pawlak. The rough set based analysis makes it very convenient for classification of data especially with huge volume of information, as the method is very efficient to find the optimal subset of attributes. In this paper, the rough set based algorithm has been applied to generate representative rules using the datasets obtained from a gas sensor array in an electronic nose instrument, capable of sensing aroma of black tea samples and these rules are used to classify the black tea quality.

Out of all the components of a mobile robot, its sensorial system is undoubtedly among the most critical ones when operating in real environments. Until now, these sensorial systems mostly relied on range sensors (laser scanner, sonar, active triangulation) and cameras. While electronic noses have barely been employed, they can provide a complementary sensory information, vital for some applications, as with humans. This chapter analyzes the motivation of providing a robot with gas-sensing capabilities and also reviews some of the hurdles that are preventing smell from achieving the importance of other sensing modalities in robotics. The achievements made so far are reviewed to illustrate the current status on the three main fields within robotics olfaction: the classification of volatile substances, the spatial estimation of the gas dispersion from sparse measurements, and the localization of the gas source within a known environment.

Saffron (Crocus sativus L.) in either unprocessed or distillate form is valuable in terms of its color, flavor and medicinal properties. Due to its high economical value and demanding production, quality assessment of saffron as well as detection of adulterated products is of interest to the food industry and important to the consumers. A plethora of instruments and analytical methodologies have been already developed to determine not only the quality and quantity of the main ingredients but also the type and level of adulterations in saffron. Apart from traditional techniques, biomimetic-based techniques have shown reliable candidates for saffron quality assurance and control. After briefly reviewing the principles of these methodologies and some of the more recent applications for saffron quality characterization, the paper discusses the advantages and limitations of the methods being employed. Also, this paper provides an overview of the more recent and possible innovative systems, as well as, their future perspectives.

Diagnosis of adulteration in cow ghee is one of the key concerns of recent years. In this study, the aroma fingerprints of cow ghee were detected. For this purpose, an electronic nose system was developed and its ability for detection of different amounts of margarine mixed with pure cow ghee (10, 20, 30, 40, and 50% levels) was investigated. The system was equipped with eight sensors (MOS type), that each of them reacts to specific volatile compounds in the samples. The features of the signals were considered for data analysis. In this research, principal components analysis (PCA) and artificial neural networks (ANN) methods were used for data analyzing. According to the results, the PCA analysis explained 98% of the variance in the data set. Also, ANN analysis identified 85.6 and 97.2% of pure cow ghee from the adulteration ones when samples were classified in 7 and 2 classes, respectively. Practical applications Diagnosis and estimation of adulteration in cow ghee is one of the key concerns for consumers. Electronic nose is a new method that can be used for detection of products quality. We show that electronic nose system is a valuable tool for detection of margarine as a one of common adulteration in cow ghee.

The electronic nose (e-nose) would simulate the human sense (smell) to identify and realize the complex aromas by employing a chemical sensors array. One of the most common sensors used in electronic nose systems are metal oxide semiconductor (MOS) sensors. In this research, a low cost e-nose system based on six metal oxide semi-conductor (MOS) sensors as a non-destructive instrument for recognition pomegranate varieties is investigated. Principal component analysis (PCA) and linear discriminant analysis (LDA) techniques are used for this purpose. The proposed e-nose has a capability of demonstrating a clear difference in aroma fingerprint of pomegranate by PCA and LDA analysis. Using LDA analysis, it is possible to identify and to categorize the difference between pomegranate varieties, and based on the results, the classification accuracy of 95.2% was obtained. Sensor array capabilities for classification of pomegranate varieties using loading analysis were investigated too. Resul...

The responses generated by a gas sensor array are difficult to classify due to their inherent imprecision, uncertainty and the procedures of computational intelligence are appropriate to deal with such imperfect knowledge. In recent years, rough set theory has attracted more attention of many researchers even though it was proposed in the early 1980's by Z. Pawlak. The rough set based analysis makes it very convenient for classification of data especially with huge volume of information, as the method is very efficient to find the optimal subset of attributes. In this paper, the rough set based algorithm has been applied to generate representative rules using the datasets obtained from a gas sensor array in an electronic nose instrument, capable of sensing aroma of black tea samples and these rules are used to classify the black tea quality.

Gas sensors are critical components when adhering to health safety and environmental policies in various manufacturing industries, such as the petroleum and oil industry; scent and makeup production; food and beverage manufacturing; chemical engineering; pollution monitoring. In recent times, gas sensors have been introduced to medical diagnostics, bioprocesses, and plant disease diagnosis processes. There could be an adverse impact on human health due to the mixture of various gases (e.g., acetone (A), ethanol (E), propane (P)) that vent out from industrial areas. Therefore, it is important to accurately detect and differentiate such gases. Towards this goal, this paper presents a novel electronic nose (e-nose) detection method to classify various explosive gases. To detect explosive gases, metal oxide semiconductor (MOS) sensors are used as reliable tools to detect such volatile gases. The data received from MOS sensors are processed through a multivariate analysis technique to cl...

In this work, an array of low-cost cross-sensitive sensors is used for discriminating the best candidate within a set of volatile organic compounds (VOCs). The challenge of our experimental setting is to deal with the problems of low selectivity, especially in normal operating conditions, so that ambiguous sensor responses (i.e. referable to more than one VOC) can be given, at least, a qualitative interpretation. In order to carry out the signal disambiguation task, a computational technique employing simple classifying rules and fuzzy descriptions has been engineered. The basic idea is that, if the same gas is actually measured by two or more sensors, then the estimated concentrations will show a low variance, with an accuracy related to the number of concordant sensors. Experiments show that, despite the cheapness of the setup and the coarse-grained nature of the provided response, encouraging results can be obtained and prospective work can follow.

To determine whether an electronic nose can differentiate cultured nonmalignant and malignant prostatic cells from each other and whether the smell print is secreted to the surrounding medium. Materials & methods: Prostatic nonmalignant (EP-156T and controls) and malignant (LNCaP) cell lines, as well as conditioned and unconditioned media, were collected. The smell prints of the samples were analyzed by a ChemPro ® 100 electronic nose device. The data were normalized and dimension reduction was conducted. The samples were classified and misclassification rates were calculated. Results: The electronic nose differentiated the nonmalignant and malignant cell lines from each other, achieving misclassification rates of 2.9-3.6%. Cells did not differ from the cond itioned med i u m but d if fered f rom the u ncond itioned med ium (misclassification rates: 0.0-25.6%). Conclusion: Malignant and nonmalignant prostatic cell lines have distinct smell prints. Prostatic cancer cells seem to modify the smell print of their medium. Keywords n cell lines n electronic nose n machine olfaction n olfactory detection n prostate cancer n volatile organic compounds

Aim: To determine whether an electronic nose can differentiate cultured nonmalignant and malignant prostatic cells from each other and whether the smell print is secreted to the surrounding medium. Materials & methods: Prostatic nonmalignant (EP-156T and controls) and malignant (LNCaP) cell lines, as well as conditioned and unconditioned media, were collected. The smell prints of the samples were analyzed by a ChemPro® 100 electronic nose device. The data were normalized and dimension reduction was conducted. The samples were classified and misclassification rates were calculated. Results: The electronic nose differentiated the nonmalignant and malignant cell lines from each other, achieving misclassification rates of 2.9–3.6%. Cells did not differ from the conditioned medium but differed from the unconditioned medium (misclassification rates: 0.0–25.6%). Conclusion: Malignant and nonmalignant prostatic cell lines have distinct smell prints. Prostatic cancer cells seem to modify the...

A methodology to calculate analytical figures of merit is not well established for detection systems that are based on sensor arrays with low sensor selectivity. In this work, we present a practical approach to estimate the Resolving Power of a sensory system, considering non-linear sensors and heteroscedastic sensor noise. We use the definition introduced by Shannon in the field of communication theory to quantify the number of symbols in a noisy environment, and its version adapted by Gardner and Barlett for chemical sensor systems. Our method combines dimensionality reduction and the use of algorithms to compute the convex hull of the empirical data to estimate the data volume in the sensor response space. We validate our methodology with synthetic data and with actual data captured with temperature-modulated MOX gas sensors. Unlike other methodologies, our method does not require the intrinsic dimensionality of the sensor response to be smaller than the dimensionality of the inp...

In this article, wavelet energy feature (WEF) has been extracted from the responses of e-nose and e-tongue for the classification of different grades of Indian black tea. For WEF, different decomposition levels of wavelet packet transform have been tested for both the systems and performance is evaluated with K-Nearest Neighbors classifier. Energy features of the best-suited decomposition level for e-nose and e-tongue have been calculated and fused to get a combined sensor response. Results confirm that the clustering nature (PCA plot) and classification accuracy (10-fold cross-validated based on KNN) have improved (accuracy 99.75%) with the applied method on the combined data. Moreover, this method is compared with some benchmark classification methods namely PLS-DA and Sammon's projection method which exhibits the superiority of the WEF extraction method combining the responses of multi sensory system.

Robots can be equipped with a range of senses to allow them to perceive and interact with the world in a more natural and intuitive way. These senses can include vision, hearing, touch, smell, and taste. Vision allows the robot to see and recognize objects and navigate its environment. Hearing enables the robot to recognize sounds and respond to vocal commands. Touch allows the robot to perceive information about the texture, shape, and temperature of objects through the sense of touch. Smell enables the robot to recognize and classify different odors. Taste enables the robot to identify the chemical composition of materials. The specific senses used in a robot will depend on the needs of the application, and many robots use a combination of different senses to perceive and interact with the environment. This paper reviews the five senses used in robots, their types, how they work, and other related information, while also discussing the possibility of a Sixth Sense.

Robots can be equipped with a range of senses to allow them to perceive and interact with the world in a more natural and intuitive way. These senses can include vision, hearing, touch, smell, and taste. Vision allows the robot to see and recognize objects and navigate its environment. Hearing enables the robot to recognize sounds and respond to vocal commands. Touch allows the robot to perceive information about the texture, shape, and temperature of objects through the sense of touch. Smell enables the robot to recognize and classify different odors. Taste enables the robot to identify the chemical composition of materials. The specific senses used in a robot will depend on the needs of the application, and many robots use a combination of different senses to perceive and interact with the environment. This paper reviews the five senses used in robots, their types, how they work, and other related information, while also discussing the possibility of a Sixth Sense.

Sniffing team campaigns were conducted in Flanders (northern Belgium) to evaluate the applicability of field olfactometry to improve the plume method. Field measurements were carried out around five different industrial sites, under neutral or slightly unstable weather conditions. Preliminary results show some limitations and possible applications of in-field odour-measuring devices to enhance the method by supplementing bare nose sniff testing observations.

This work deals with the development and evaluation of an integrated system based on computer vision system (CVS) and electronic nose (e-nose) for saffron adulteration detection. Ten saffron samples adulterated with two common illegal constituents, namely, Artificially Colored Safflower (ACS) and Artificially Colored Yellow Styles of Saffron (ACYSS) at levels ranging from 10 to 50% (w/w) were characterized in this work. First, the developed CVS and e-nose system were integrated to form a unit system. This set up was utilized to extract color and aroma characteristic variables of each sample. The extracted variables were processed using Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), and Support Vectors Machines (SVMs) to demonstrate the discrimination capability of the developed system. Two multilayer artificial neural network (ANN-MLP) models were also employed for saffron color and aroma strength prediction based on ISO standards. PCA and HCA results of the color and aroma datasets revealed that the adulterated samples have different color and aroma strength compared to authentic saffron and they can clearly be distinguished. SVMs classifier showed good agreement with the PCA results and reached 89% and 100% success rate in the recognition of the different saffron samples based on their color and aroma datasets, respectively. Results of the two ANN-MLP models proved that the developed system is capable of differentiating the authentic and adulterated saffron samples based on their color and aroma strength (R 2 Color analysis P 0:95 and R 2 Aroma analysis P 0:97).

Quality assurance has always been an important issue in the production and utilization of medicinal and aromatic plants. Most medicinal plants have aromatic characteristics due to the presence of Volatile Organic Compounds (VOCs) that can be analyzed using gas chromatography and mass spectroscopy. More modern techniques use sensors along with signal processing and advanced data processing algorithms. Electronic noses (e-noses), are instruments designed to mimic the sense of smell found in biological systems. E-noses contain numerous types of gas sensors that interact with odor molecules to produce electronic signals. Then, the sensor responses are sent to a computer system that uses multivariate data analysis methods to distinguish differences in the data based on the VOCs in the sample headspace. This paper provides a review of the applications of e-nose technologies for assessing the quality and safety of commercially-available medicinal plant products. Advantages and limitations of these systems are also covered.

Control of air pollutant emissions is a mean to protect air quality. In most cases, this is realized by the imposition of specific emission standards, which usually are based on national and international experience. However, in areas with intense and dense polluting activities (e.g. industrial sites, trafficked roads) these horizontal measures will most probably fail to sustain local air quality. This is mainly because they do not consider the local transport mechanisms of pollutants from the source to the receptor. High local concentration levels might be the result of specific meteorological conditions, amplification owe to neighbour sources and deficient design of the air pollution control (APC) systems. Present work describes a conceptual and holistic approach based on the downwind optimisation, which takes into consideration also the transfer mechanisms that induce the ground level concentration of pollutants. This method can be used to design efficiently and dynamically contr...

This dissertation is organized as four stand-alone papers. Paper No. 1 describes the development of the Mask Scentometer and reports dilution ratios measured during use by twelve different people. Dilution ratios at the Mask Scentometer's five dilution-to-threshold (D/T) settings were found to be 0.35, 1, 2, 4.5 and 18. In Paper No.'s 2 and 4, ambient odor assessment methods were compared in both controlled laboratory conditions and in the field. Laboratory analysis of ambient air samples using dynamic triangular forced-choice olfactometry (DTFCO) did not correlate well with any of the ambient odor assessment methods. Average intensitypredicted D/T was roughly five times higher than D/T measured using a Nasal Ranger ® , and D/T obtained using a Nasal Ranger ® was roughly two to five times higher than corresponding D/T from a Mask Scentometer. In Paper 3, odor intensity ratings and Mask Scentometer readings were used to calibrate the AERMOD dispersion model for predicting odor concentrations downwind of area sources. Dispersion of odor from a swine waste treatment lagoon and two cattle feedlots was modeled with AERMOD and the predictions were compared to the observations using a statistical approach to develop scaling factors. These were found to be 12 for odor intensity and 0.15 for the Mask Scentometer (although a scaling factor between 0.5 and 0.7 is also justified). Random effects and autocorrelation were found to be significant in ambient odor assessment data. In Paper 4, the dispersion of odors from a swine production building complex was studied. CALPUFF and AERMOD were used to predict short-time-step (one minute) odor concentrations. Source emission measurements and meteorological data were collected to coincide with ambient odor measurements obtained using the Nasal Ranger ® , Mask Scentometer, field odor intensity ratings, and DTFCO. In general, odor concentrations predicted using CALPUFF were found to be about twice those predicted by AERMOD. Model predictions agreed best with the readings from the Nasal Ranger ® and Mask Scentometer; and both of these ambient odor assessment methods are suited for ground truthing AERMOD and CALPUFF, although some model scaling factor adjustment is needed.

A collaborative research effort by several institutions is investigating odor emissions from swine production facilities, and the impacts of those emissions on farm neighbours. Trained human receptors were used to measure the downwind odor concentrations from four tunnel ventilated swine barns near Story City, Iowa. Twenty-six measurement events were conducted between June and November 2004 and modeled using a specially coded short time-step version of CALPUFF to predict short time step durations. Source emission measurements and extensive meteorological data were collected along with ambient olfactometry analysis using the Nasal Ranger™ device (St. Croix Sensory, St. Paul MN). Approximately 64% of measured odor generally falls within the range of modeled values. Analysis of measured odor concentration and corresponding meteorology indicate that maximum ambient odor impacts occur with lower ambient temperature during non-turbulent conditions. Analysis of the data set did not yield a strong relationship directly (R 2 =0.33), but a regression analysis indicated that the modified CALPUFF model yielded a slope or scaling factor of 0.99, indicating overall a good relationship between model and observed. However, when the data is tested with the Spearman's rank correlation coefficient an r s of 0.17 was calculated, indicating a poor rank correlation and was not significant (p≤0.05). Statistical analysis is inconclusive as to whether the results have bias, but indicate large error in the results. Given that there were no scaling or peak to mean ratio adjustments to the model predictions, the results are very promising for predicting odors using CALPUFF.

Among the different cases of emerging food fraud during the post-harvest processing, the adulteration in grape syrup is one. Typically, the grape syrup is adulterated with some illegitimate foreign materials such as grape paste (sauce), date syrup and even adding sugar-water solution to the pure grape syrup. The present study deals with assessing an electronic nose (e-nose) consisting of eight different metal oxide semiconductor (MOS) sensors for prompt detection of adulteration in the grape syrup. Three different adulterants i.e. grape paste, date syrup and sugar-water solution, each at three levels of 50, 60 and 75%, were tested. The collected data from MOS were normalised and visualised with the help of standard normal variate (SNV) and principal component analysis (PCA), respectively. Moreover, the scores obtained from PCA were used to perform hierarchal cluster analysis (HCA) to identify the similarities between different adulterated mixtures and pure grape syrup. Three different classification cases were considered to (i) address the presence of adulteration, (ii) detect the different adulterants and (iii) classify the amount of each adulteration. Linear discriminant analysis (LDA) and multi-class support vector machine (SVM) were used for classification analyses. Results showed that PCA identified provided separate clusters for the MOS data corresponding to different adulterants and their levels. The HCA showed a hierarchal of similarities between pure grape syrup and different levels of adulterations. LDA and SVM resulted in a successful classification modelling. However, the performance of SVM was considerably better than LDA with classification accuracies of 98.6 ± 0.10%, 98.9 ± 1.16% and 95.1 ± 1.39% for detecting adulteration, different adulterants and different concentrations of adulterants, respectively. MOS sensors coupled with chemometrics could provide a useful instrument and fast procedure for detection of adulteration in grape syrup.

Deploying chemosensor arrays in close proximity to stationary phases imposes stimulus-dependent spatio-temporal dynamics on their response and leads to improvements in complex odour discrimination. These spatio-temporal dynamics need to be taken into account explicitly when considering the detection performance of this new odour sensing technology, termed an artificial olfactory mucosa. For this purpose, we develop here a new measure of spatio-temporal information that combined with an analytical model of the artificial mucosa, chemosensor and noise dynamics completely characterizes the discrimination capability of the system. This spatio-temporal information measure allows us to quantify the contribution of both space and time to discrimination performance and may be used as part of optimization studies or calculated directly from an artificial mucosa output. Our formal analysis shows that exploiting both space and time in the mucosa response always outperforms the use of space alo...

The development of analytical methods aimed at tracing agri-food products and assessing their authenticity is essential to protect food commercial value and human health. An NMR-based non-targeted method is applied here to establish the authenticity of saffron samples. Specifically, 40 authentic saffron samples were compared with 18 samples intentionally adulterated by using turmeric and safflower at three different concentration levels, i.e., 5, 10, and 20 wt%. Statistical processing of NMR data furnished useful information about the main biomarkers contained in aqueous and dimethyl sulfoxide extracts, which are indicative of the presence of adulterants within the analyzed matrix. Furthermore, a discrimination model was developed capable of revealing the type of agronomic practice adopted during the production of this precious spice, distinguishing between organic and conventional cultivation. The main objective of this work was to provide the scientific community involved in the q...

An autonomous robot using symbolic reasoning, sensing and acting in a real environment needs the ability to create and maintain the connection between symbols representing objects in the world and the corresponding perceptual representations given by its sensors. This connection has been named perceptual anchoring. In complex environments, anchoring is not always easy to establish: the situation may often be ambiguous as to which percept actually corresponds to a given symbol. In this paper, we extend perceptual anchoring to deal robustly with ambiguous situations by providing general methods for detecting them and recovering from them. We consider different kinds of ambiguous situations. We also present methods to recover from these situations based on automatically formulating them as conditional planning problems that then are solved by a planner. We illustrate our approach by showing experiments involving a mobile robot equipped with a color camera and an electronic nose.

In 2006 the WG27 of CEN TC264 started with the normalisation of the different methods for determination of odour in ambient air. Existing VDI standards for the grid method and the stationary plume method were used as a starting point in this standardization committee. In Belgium however, there is since the early nineties a lot of experience with a slightly different method. This method, called the dynamic plume method, is since then widely used for determination of odour downwind of a source. The method itself is very easy to execute, doesn't require special instruments and gives a very comprehensible output. The results of these observations are used in modelling exercises to determine the emission strength of a source and hence the odour impact on the surroundings. Emission is expressed as a number of sniffing units per second (su/s). The Flemish odour legislation is based on this methodology. This paper gives an overview of the advantages and drawbacks of this method and indi...

h i g h l i g h t s < We compare methods to assess odor pollution from an anaerobic digestion plant for MSW. < We applied the atmospheric dispersion model CALPUFF and the Odor Field Inspection. < They were compared quantitatively and qualitatively over the same area for 6 months. < The methods assessed similar spatial extents of nuisance in terms of odor frequency. < The two methods provide complementary information about odor nuisance.

Nowadays olfactory discomfort is one of the most important factors of well-being. Different methods exist to calculate a discomfort indicator, but these indicators are based on one year human observations. However, only few methods are able to give information on concentration and acceptability of odours. These values are rarely combined to build an annoyance potential for odorous sources. In this paper, a new method is proposed for the evaluation of odours acceptability. In this method the odour concentration determination (according to the EN 13725) is followed by odour acceptability determination. The measurement of acceptability is realised at different concentrations just above the threshold at which all panel members detect the odour. An extrapolation is done to obtain the acceptability of the sample at the real concentration. A potential of annoyance is calculated by multiplying odour concentration and acceptability. In the last part of this paper, three examples are given to...

Gas sensors are critical components when adhering to health safety and environmental policies in various manufacturing industries, such as the petroleum and oil industry; scent and makeup production; food and beverage manufacturing; chemical engineering; pollution monitoring. In recent times, gas sensors have been introduced to medical diagnostics, bioprocesses, and plant disease diagnosis processes. There could be an adverse impact on human health due to the mixture of various gases (e.g., acetone (A), ethanol (E), propane (P)) that vent out from industrial areas. Therefore, it is important to accurately detect and differentiate such gases. Towards this goal, this paper presents a novel electronic nose (e-nose) detection method to classify various explosive gases. To detect explosive gases, metal oxide semiconductor (MOS) sensors are used as reliable tools to detect such volatile gases. The data received from MOS sensors are processed through a multivariate analysis technique to cl...

The recent development of MAU-9 electronic sensory methods, based on artificial olfaction detection of volatile emissions using an experimental metal oxide semiconductor (MOS)-type electronic-nose (e-nose) device, have provided novel means for the effective discovery of adulterated and counterfeit essential oil-based plant products sold in worldwide commercial markets. These new methods have the potential of facilitating enforcement of regulatory quality assurance (QA) for authentication of plant product genuineness and quality through rapid evaluation by volatile (aroma) emissions. The MAU-9 e-nose system was further evaluated using performance-analysis methods to determine ways for improving on overall system operation and effectiveness in discriminating and classifying volatile essential oils derived from fruit and herbal edible plants. Individual MOS-sensor components in the e-nose sensor array were performance tested for their effectiveness in contributing to discriminations of...

Odour emissions from livestock operations in Ontario have raised significant public concerns. In this study, downwind odour measurements were conducted during a 7-week period in July and August in 2004. The study involved taking measurements at two poultry farms, six dairy farms, and six pig farms in southern Ontario using both human assessor observed Nasal Ranger field olfactometers and an electronic nose developed in our research laboratory. The factors considered include animal species, distance to the odour source, and weather conditions (i.e. wind speed, temperature, cloud cover, and atmospheric stability). The collected data were studied and the relationships between individual factors and downwind odour intensity were analysed. In addition, the significance of the factors was investigated and ranked. The results show that downwind livestock farm odour should be considered as a complex system and it is suggested that multi-factor analysis be performed. The ranking of the significance of factors to odour intensity can improve the understanding of downwind odour systems and enhance the efficiency of odour evaluation and reduction techniques. Responses of the gas sensors in the electronic nose were compared with the perceived odour intensities, and good agreement was reported.

h i g h l i g h t s < We compare methods to assess odor pollution from an anaerobic digestion plant for MSW. < We applied the atmospheric dispersion model CALPUFF and the Odor Field Inspection. < They were compared quantitatively and qualitatively over the same area for 6 months. < The methods assessed similar spatial extents of nuisance in terms of odor frequency. < The two methods provide complementary information about odor nuisance.

While the inclusion of visual, aural, and tactile senses into virtual reality systems is widespread, the sense of smell has been largely ignored. We have developed a chemical vapor sensing system for the automated identification of chemical vapors (smells). Our prototype chemical vapor sensing system is composed of an array of tin-oxide vapor sensors coupled to an artificial neural network. The artificial neural network is used in the recognition of different smells and is constructed as a standard multilayer feed-forward network trained with the backpropagation algorithm. When a chemical sensor array is combined with an automated pattern identifier, it is often referred to as an electronic or artificial nose. Applications of electronic noses include monitoring food and beverage odors, automated flavor control, analyzing fuel mixtures, and quantifying individual components in gas mixtures. Our prototype electronic nose has been used to identify odors from common household chemicals. An electronic nose will potentially be a key component in an olfactory input to a telepresent virtual reality system. The identified odor would be electronically transmitted from the electronic nose at one site to an odor generation system at another site. This combination would function as a mechanism for transmitting olfactory information for telepresence. This would have direct applicability in the area of telemedicine since the sense of smell is an important sense to the physician and surgeon. In this paper, our chemical sensing system (electronic nose) is presented along with a proposed method for regenerating the transmitted olfactory information.

Gas sensors are critical components when adhering to health safety and environmental policies in various manufacturing industries, such as the petroleum and oil industry; scent and makeup production; food and beverage manufacturing; chemical engineering; pollution monitoring. In recent times, gas sensors have been introduced to medical diagnostics, bioprocesses, and plant disease diagnosis processes. There could be an adverse impact on human health due to the mixture of various gases (e.g., acetone (A), ethanol (E), propane (P)) that vent out from industrial areas. Therefore, it is important to accurately detect and differentiate such gases. Towards this goal, this paper presents a novel electronic nose (e-nose) detection method to classify various explosive gases. To detect explosive gases, metal oxide semiconductor (MOS) sensors are used as reliable tools to detect such volatile gases. The data received from MOS sensors are processed through a multivariate analysis technique to cl...

In this paper we describe an implemented framework that integrates knowledge representation and reasoning in a symbiotic system. In such systems a number of heterogeneous sensors pervasively embedded in the environment, mobile robots and humans co-exist and communicate. In this work, the integration is mediated through perceptual anchoring, which creates and maintains the correspondences between the symbol system and the perceptual data that refer to the same physical object. The overall framework is evaluated using ResearchCyc as the knowledge representation and reasoning system, within the context of a physical testbed, which consists of a small apartment-like home.

Autonomous search of odor sources is a problem that once solved can bring large improvements to tasks like humanitarian demining, airport patrolling, and hazardous chemical leaks detection. The main obstacles to the research in this area are the properties of current gas sensors (slow, weakly selective and poorly sensitive) and the difficulty to estimate the physical characteristics of natural odor plumes due to the "disturbances" caused by turbulence, namely meandering and intermittence. This present work describes on the design and development of a mobile robot equipped with semiconductor gas sensors which has been designed to locate the odor sources. A search algorithm is proposed to give movements to the robot in the direction of odor source.

This study investigated the capacity of a deep neural network to distinguish tea types based on their aromas. The data set of aromas from tea leaves, which contained sensor responses measured with a gas–sensing system using a mass– sensitive chemical sensors namelyMQ3, MQ5, MQ7, was used to evaluate the recognition accuracy. To define the input vectors of the deep neural network in aroma recognition experiments, frequency analysis using a continuous wavelet transform, with the Morlet function as the mother wavelet, was used to extract features from the sensor signals of the data set. The deep neural network achieved a recognition accuracy of 100% for the three tea types: oolong, jasmine and pu’erh, and the base gas of dehumidified indoor air. Comparing the recognition accuracy of the deep neural network to that obtained from other pattern recognition methods, such as naive Bayes and random forests, the experimental results demonstrated the effectiveness of applying a deep neural network to this task.

This study investigated the capacity of a deep neural network to distinguish tea sources based on their aromas. The data set of aromas from tea sources, which contained sensor responses measured with a gas–sensing system using a mass– sensitive chemical sensors namelyMQ3, MQ5, MQ7, was used to evaluate the recognition accuracy. To de?ne the input vectors of the deep neural network in aroma recognition experiments, frequency analysis using a continuous wavelet transform, with the Morlet function as the mother wavelet, was used to extract features from the sensor signals of the data set. The deep neural network achieved a recognition accuracy of 100% for the tea sources and the base gas of de humidified indoor air. Comparing the recognition accuracy of the deep neural network to that obtained from other pattern recognition methods, such as naive Bayes and random forests, the experimental results demonstrated the effectiveness of applying a deep neural network to this task.

Windbreaks are believed to help disperse odours emitted by livestock facilities. The objective of the project was to measure the effect of windbreaks on the size and intensity of odour dispersion plumes developed in the field when subjected to a point odour source. Comparisons were made for odour plumes observed with and without windbreaks, and with windbreaks exposed to different conditions. Besides a control site without windbreak, four windbreak sites were selected, two of which had one row of deciduous trees while the other two had one row of coniferous trees. Odour dispersion plumes were measured 6 times on the control site and 33 times on the windbreak sites. Each time, an odour generator was used to produce a controllable level of odour emission. Three groups of four trained panellists measured the size and intensity of the odour plume developing in the field downwind from the odour generator. Using a forced choice dynamic olfactometer, all 12 panellists were calibrated every test day and the group's field odour intensity perception was correlated to odour concentrations. Windbreaks were found to have an effect on odour dispersion. This effect was more pronounced when the windbreak was dense (lower optical porosity) and consisted of coniferous trees. Moreover, odour dispersion was improved when the source was located 15 m upwind from the windbreak, rather than 60 m. When temperatures were above 15 8C, odours were dispersed over a shorter distance, likely because of added convective effects. Wind speed was found to have a limited effect on the size and intensity of the odour plume while wind direction perpendicular to the windbreak reduced the size of the odour plume but not the trapping of odours on the leeward side of the windbreak. In general, windbreaks can improve odour dispersion, but a better study of their performance is required through modeling. #

Olfaction is a challenging new sensing modality for intelligent systems. With the emergence of electronic noses it is now possible to detect and recognise a range of different odours for a variety of applications. An existing application is to use electronic olfaction on mobile robots for the purpose of odour based navigation. In this work, we introduce a new application where electronic olfaction is used in cooperation with other types of sensors on a mobile robot in order to acquire the odour property of objects. The mobility of the robot facilitates the execution of specific perceptual actions, such as moving closer to objects to acquire odour properties. Additional sensing modalities provides the spatial detection of objects and electronic olfaction then acquires the odour property which can be used for discrimination and recognition of the object being considered. We examine the problem of deciding when, how and where the e-nose should be activated by planning for active perception. We investigate the use of symbolic reasoning techniques in this context and consider the problem of integrating the information provided by the e-nose with both prior information and information from other sensors (e.g., vision). Finally, experiments are performed on a mobile robot equipped with an e-nose together with a variety of sensors that can perform decision making tasks in realistic environments.

Olfaction is a challenging new sensing modality for intelligent systems. With the emergence of electronic noses (enoses) it is now possible to train a system to detect and recognise a range of different odours. In this work, we integrate the electronic nose on a multi-sensing mobile robotic platform. We plan for perceptual actions and examine when, how and where the e-nose should be activated. Finally, experiments are performed on a mobile robot equipped with an e-nose together with a variety of sensors and used for object detection.

An important requirement for autonomous systems is the ability to detect and recover from exceptional situations such as failures in observations. In this paper we demonstrate how techniques for planning with sensing under uncertainty can play a major role in solving the problem of recovering from such situations. In this £rst step we concentrate on failures in perceptual anchoring, that is how to connect a symbol representing an object to the percepts of that object. We provide a classi£cation of failures and present planning-based methods for recovering from them. We illustrate our approach by showing tests run on a mobile robot equipped with a color camera.

We present a new approach for odour detection and recognition based on a so-called PEIS-Ecology: a network of gas sensors and a mobile robot are integrated in an intelligent environment. The environment can provide information regarding the location of potential odour sources, which is then relayed to a mobile robot equipped with an electronic nose. The robot can then perform a more thorough analysis of the odour character. This is a novel approach which alleviates some the challenges in mobile olfaction techniques by single and embedded mobile robots. The environment also provides contextual information which can be used to constrain the learning of odours, which is shown to improve classification performance.

Olfaction is a challenging new sensing modality for intelligent systems. With the emergence of electronic noses, it is now possible to detect and recognize a range of different odours for a variety of applications. In this work, we introduce a new application where electronic olfaction is used in cooperation with other types of sensors on a mobile robot in order to acquire the odour property of objects. We examine the problem of deciding when, how and where the electronic nose (e-nose) should be activated by planning for active perception and we consider the problem of integrating the information provided by the e-nose with both prior information and information from other sensors (e.g., vision). Experiments performed on a mobile robot equipped with an e-nose are presented.

An autonomous robot using symbolic reasoning, sensing and acting in a real environment needs the ability to create and maintain the connection between symbols representing ob- jects in the world and the corresponding perceptual repre- sentations given by its sensors. This ...

This study focuses on the development and evaluation of a portable electronic nose (e-nose) system for identification of different types of saffron, stigma of Crocus sativus L. (Iridaceae), based on their Volatile Organic Compounds (VOCs). The system utilizes metal oxide semiconductor gas sensors and direct head space sampling. Real-time data acquisition system, microcontroller devices and a laptop computer along with multivariate computational tools were used for development of an expert system. Eleven saffron samples from different regions were prepared for the experiments. Principal Component Analysis (PCA) and Hierarchical Clustering Analysis (HCA) as unsupervised models and Multilayer Perceptron (MLP) neural networks and Partial Least Squares (PLS) as supervised models were utilized to develop the e-nose discrimination capability. Based on the results, PCA of volatile compounds fingerprints revealed eleven distinct groups corresponding to the eleven different saffron samples. This was further confirmed by HCA which classified the groups into five distinct Quality Classes (QCs) (excellent, very good, good, medium, and poor quality) which were used as the MLP and PLS classification goals. Results of analysis showed that performance of the MLP model for prediction of saffron samples QC was better than the PLS model, with 100% success rate and high correlation coefficients of cross validation (R 2 = 0.989 and relatively low RMSE value of 0.0141). These results show that the developed system is capable of discriminating saffron samples based on their aroma and can be utilized as an aroma quality control system.

An important requirement for autonomous systems is the ability to detect and recover from exceptional situations such as failures in observations. In this paper we demonstrate how techniques for planning with sensing under uncertainty can play a major role in solving the problem of recovering from such situations. In this £rst step we concentrate on failures in perceptual anchoring, that is how to connect a symbol representing an object to the percepts of that object. We provide a classi£cation of failures and present planning-based methods for recovering from them. We illustrate our approach by showing tests run on a mobile robot equipped with a color camera.