Odor Source Localization Research Articles

A Deep Learning-Based Indoor Odor Compass

Mobile robot-based odor source localization (OSL) has broad applications in various industrial and daily-life scenarios. To this end, a deep learning-based odor compass is designed in this work. Functionally, the designed odor compass is divided into three primary modules, which are the sensing module (i.e., a sensor array composed of four metal-oxide-semiconductor (MOS) gas sensors), the communication module, and the remote data processing module (i.e., a deep learning-based algorithm). In particular, a deep learning-based odor attention (DL-OA) model is proposed to realize an end-to-end odor source direction estimation (OSDE) based on the responses of gas sensor array. Moreover, the proposed DL-OA model adopts a separated spatial-temporal attention-based encoder-decoder structure. Furthermore, the average validation error in estimating the OSD in an indoor environment is 4.98°, essentially demonstrating the effectiveness of designed odor compass.

Odor source localization of multi-robots with swarm intelligence algorithms: A review.

The use of robot swarms for odor source localization (OSL) can better adapt to the reality of unstable turbulence and find chemical contamination or hazard sources faster. Inspired by the collective behavior in nature, swarm intelligence (SI) is recognized as an appropriate algorithm framework for multi-robot system due to its parallelism, scalability and robustness. Applications of SI-based multi-robots for OSL problems have attracted great interest over the last two decades. In this review, we firstly summarize the trending issues in general robot OSL field through comparing some basic counterpart concepts, and then provide a detailed survey of various representative SI algorithms in multi-robot system for odor source localization. The research field originates from the first introduction of the standard particle swarm optimization (PSO) and flourishes in applying ever-increasing quantity of its variants as modified PSOs and hybrid PSOs. Moreover, other nature-inspired SI algorithms have also demonstrated the diversity and exploration of this field. The computer simulations and real-world applications reported in the literatures show that those algorithms could well solve the main problems of odor source localization but still retain the potential for further development. Lastly, we provide an outlook on possible future research directions.

A reinforcement learning fuzzy system for continuous control in robotic odor plume tracking

AbstractIn dynamic outdoor environments characterized by turbulent airflow and intermittent odor plumes, robotic odor plume tracking remains challenging, because existing algorithms heavily rely on manually tuning or learning from expert experience, which are hard to implement in an unknown environment. In this paper, a multi-continuous-output Takagi–Sugeno–Kang fuzzy system was designed and tuned with reinforcement learning to solve the robotic odor source localization problem in dynamic odor plumes. Based on the Lévy Taxis plume tracking controller, the proposed fuzzy system determined the parameters of the controller based on the robot’s observation and guided the robot to turn and move towards the odor source at each searching step. The trained fuzzy system was tested in simulated filament-based odor plumes dispersed by a changing wind field. The results showed that the performance of the proposed fuzzy system-based controller trained with reinforcement learning can achieve a similar success rate and higher efficiency compared with a manually tuned and well-designed fuzzy system-based controller. The fuzzy system-based plume tracking controller was also validated through real robotic experiments.

Rapid location technology of odor sources by multi‐UAV

AbstractA multi‐unmanned aerial vehicle (UAV) offers the advantages of flexibility, safety, and efficiency with respect to odor source localization. When the traditional particle swarm optimization (PSO) algorithm is used to locate the odor source, the search efficiency of the UAVs is reduced owing to unplanned search routes, and local extreme points and internal collisions may occur, making it impossible to quickly locate the odor source. In response to these problems, this paper proposes an improved particle swarm optimization (IPSO) algorithm, which combines the PSO with the Zigzag algorithm to preplan the search routes and improve the efficiency of the UAVs to search the plume. In addition, the distance between the UAVs is used as an adjustment factor to adaptively adjust the inertia weights of the particles to prevent the UAVs from falling into a local optimum during the plume‐tracking stage and further improve the positioning efficiency. Finally, the PSO combined with the artificial potential field algorithm is used to avoid the internal collision of the UAVs when tracking the plume. The algorithm is verified through simulations and physical experiments. The results show that the IPSO significantly improves the efficiency and can achieve rapid and effective location of the odor source, which verifies the effectiveness and applicability of the algorithm.

Robotic odor source localization via adaptive bio-inspired navigation using fuzzy inference methods

Robotic odor source localization (OSL) has been viewed as a challenging task due to the turbulent nature of airflows and the resulting odor plume characteristics. The key to solving an OSL problem is designing an effective olfactory-based navigation algorithm, which guides a plume-tracing robot to find the odor source via tracing emitted plumes. Inspired by the mate-seeking behaviors of male moths, this article presents a behavior-based navigation algorithm for using on a mobile robot to locate an odor source in an unknown environment. Unlike traditional bio-inspired algorithms, which use fixed parameters to formulate robot search trajectories, we design a fuzzy controller to perceive the environment and adjust trajectory parameters based on the current search situation. Therefore, the robot can automatically adapt the scale of search trajectories to fit environmental changes and balance the exploration and exploitation of the search. Simulation and on-vehicle results show that compared to two classical olfactory-based navigation algorithms, the proposed algorithm is more efficient and outperforms them in terms of the averaged search time and success rate.

Electroantennography-based Bio-hybrid Odor-detecting Drone using Silkmoth Antennae for Odor Source Localization

Small drones with chemical or biosensor devices that can detect airborne odorant molecules have attracted considerable attention owing to their applicability in environmental and security monitoring and search-and-rescue operations. Small drones with commercial metal-oxide-semiconductor (MOX) gas sensors have been developed for odor source localization; however, their real-time-odor-detection performance has proven inadequate. However, biosensing technologies based on insect olfactory systems exhibit relatively high sensitivity, selectivity, and real-time response with respect to odorant molecules compared to commercial MOX gas sensors. In such devices, excised insect antennae function as portable odorant biosensor elements and have been found to deliver excellent sensing performance. This study presents experimental protocols for odorant-molecule detection in the air using a small autonomous bio-hybrid drone based on a mountable electroantennography (EAG) device incorporating silkmoth antennae. We developed a mountable EAG device including sensing/processing parts with a Wi-Fi module. The device was equipped with a simple sensor enclosure to enhance the sensor directivity. Thus, odor source localization was conducted using the spiral-surge algorithm, which does not assume an upwind direction. The experimental bio-hybrid odor-detecting drone identified real-time odorant-concentration differences in a pseudo-open environment (outside a wind tunnel) and localized the source. The developed drone and associated system can serve as an efficient odorant molecule-detection tool and a suitable flight platform for developing odor source localization algorithms owing to its high programmability.

Electroantennography-based Bio-hybrid Odor-detecting Drone using Silkmoth Antennae for Odor Source Localization

Small drones with chemical or biosensor devices that can detect airborne odorant molecules have attracted considerable attention owing to their applicability in environmental and security monitoring and search-and-rescue operations. Small drones with commercial metal-oxide-semiconductor (MOX) gas sensors have been developed for odor source localization; however, their real-time-odor-detection performance has proven inadequate. However, biosensing technologies based on insect olfactory systems exhibit relatively high sensitivity, selectivity, and real-time response with respect to odorant molecules compared to commercial MOX gas sensors. In such devices, excised insect antennae function as portable odorant biosensor elements and have been found to deliver excellent sensing performance. This study presents experimental protocols for odorant-molecule detection in the air using a small autonomous bio-hybrid drone based on a mountable electroantennography (EAG) device incorporating silkmoth antennae. We developed a mountable EAG device including sensing/processing parts with a Wi-Fi module. The device was equipped with a simple sensor enclosure to enhance the sensor directivity. Thus, odor source localization was conducted using the spiral-surge algorithm, which does not assume an upwind direction. The experimental bio-hybrid odor-detecting drone identified real-time odorant-concentration differences in a pseudo-open environment (outside a wind tunnel) and localized the source. The developed drone and associated system can serve as an efficient odorant molecule-detection tool and a suitable flight platform for developing odor source localization algorithms owing to its high programmability.

A Deep Q-Network for robotic odor/gas source localization: Modeling, measurement and comparative study

Robotic odor/gas source localization is a widely studied field, but most of the existing works are about rule-based algorithms. In this paper, the Deep Q-Network algorithm is applied to solve the odor source localization problem. An odor hits distribution model is proposed to model the odor concentration distribution in indoor environments, taking the dispersion by airflow, the odor molecular random walk, and the obstacles into account. The Deep Q-Network takes the stacked historic measurement data as the input and outputs the expected cumulative future reward of actions of the robots. The network is trained through 35,000 repeated episodes of random odor source localization tasks. The Deep Q-Network method is evaluated under four different environment settings in a simplified simulator and compared with two widely used odor source localization algorithms. The evaluation results demonstrate the advantages of the proposed algorithm. The algorithm is also validated in more complex indoor environments.

Recent Progress and Trend of Robot Odor Source Localization

Accurate and prompt localization of hazardous release sources ensures a timely emergency response to potential damages. Robot odor source localization is the study of locating the release source based on mobile robots in real‐time, which has been experiencing rapid growth in recent years. This paper presents a review of this field, focusing on recent progress and development trends. Research works are grouped according to method principles, leading to four categories: reactive methods, heuristic searching methods, probabilistic inference methods, and learning methods. The odor‐source direction and distance prediction viain situsensing and developments of simulators are also listed separately for discussion. Several outlooks are highlighted at the end. Along with the developments of related fields and practical needs, we believe this field is bound to keep on flourishing in the future. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

A novel plume tracking method in partial 3D diffusive environments using multi-sensor fusion

In recent years, dangerous gas leakage has led to serious consequences in social security. Pervious strategies for locating the odor sources were carried out in 2D environments or the gas sensors were installed at a fixed height, ignoring the characteristics of plume distribution in 3D environments. In some cases, strategies developed in 2D environments may be completely ineffective when be applied to 3D environments. To enhance the success rate and rapidity, a novel method of odor source location in partial 3D diffusive environments based on multi-sensor fusion is proposed. Considering the suspected source existed and weak wind environments, machine olfactory and vision information are both employed to track the plumes and identify its source. To enhance the speed of plume searching and the reliability of plume tracking in the early stages, an autonomous mobile robot (AMR) simulates the social mechanism and hunting behaviors of the gray wolf population and tracks the plumes. A technical strategy is also designed to enable the AMR to complete the task of gas source location successfully and the subsumption architecture is adopted to define and arbitrate behavior priorities to coordinate different behaviors in this paper. The on-site test results show that the average positioning error is 0.13 m, the average running time is 147.7 s and the average distance traveled is 21.35 m. The results show that the proposed method is competent to accomplish the task of leakage gas source localization in partial 3D diffusive environments.

Real-time odor concentration and direction recognition for efficient odor source localization using a small bio-hybrid drone

• Autonomous small drone with an insect antennae-based sensor device was developed. • Drone recognized real-time odorant concentration differences in air. • Sensor enclosure that enhanced the sensor directivity was developed. • Drone with an enclosure could distinguish front–back directions in an odor plume. • Drone localized the odor source in a pseudo-open environment using a spiral-surge algorithm. Small drones with biosensor devices have great potential for detecting odorant molecules in air and can be applied to environmental and security monitoring. To realize these applications, two important factors are considered: first, development of highly sensitive, selective, and real-time odorant sensor devices, and second, construction of a highly maneuverable platform with efficient odor source localization. Previously, small drones with commercial gas sensors or biosensors based on insect antennae have been developed. However, the performance of gas sensors proved to be inadequate for real-time sensing; the flight performance of a bio-hybrid drone was limited because the yaw turn was not considered and flight tests were conducted in a wind tunnel. In this study, we developed a fully autonomous small drone with a portable electroantennogram (EAG) based on silkmoth antennae mounted on it. The EAG device was also equipped with a sensor enclosure to enhance sensor directivity. The bio-hybrid drone can recognize real-time odorant concentration differences in a pseudo-open environment (outside the wind tunnel). We also developed an enclosure for enhancing the sensor directivity of EAG. Owing to the enclosure, the drone could distinguish the front-back direction in the odor plume without a wind direction sensor. Based on these results, the drone recognized the maximum value of odorant concentration during an over 360° yaw turn and localized the odor source using the spiral-surge algorithm without any assumption of the upwind direction. This study proposes an efficient flight platform for detecting odorant molecules in air and localizing their sources.

Plume Source Localization on Enceladus by Sequential Monte Carlo Method

Water vapor plumes emanating from the geyser vents at Enceladus’s south pole area invite the possibility of direct access to the subsurface liquid reservoir to acquire pristine biological material if it exists. Any descending lander adapted for plume localization is required to not only explore the icy plume environment during its descent, but it must also infer the location of the landing target—the plume source—autonomously. Compared with existing scenarios of terrestrial plume source localization methods, the source likelihood map (SLIM) method for an Enceladus mission offers a more extensive search area, a higher maneuver velocity, and a shorter search time. This paper investigates a particle-based odor source localization (pOSL) approach that offers the prospect of targeting one of the plume sources by autonomously measuring the concentration field. Reasons for the negative likelihood and overfitting issues associated with Bayesian SLIM are analyzed to build a novel probabilistic model. By implementing this model, the proposed pOSL algorithm evaluates the observation likelihood via posterior maximization method and estimates the source location via the sequential Monte Carlo method. The pOSL algorithm resolves difficulties associated with other methods while reducing the time complexity from to . The numerical simulations illustrate that the proposed approach is feasible and permits accurate targeting of Enceladus’s geyser vents.

Particle Source Localization With a Low-Cost Robotic Sensor System: Algorithmic Design and Performance Evaluation

In this paper we tackle the problem of finding the source of particulate matter with a mobile robot equipped with a low-cost multi-channel optical particle counting sensor. The proposed method is based on the Infotaxis odor source localization algorithm and makes multiple modifications to adapt it to particle plumes. In particular, we propose three simple but efficient ways to fuse multiple probability maps associated with various particle sizes and use the resulted integrated map to guide the robot’s movements. A refined measurement data collection is conducted in a wind tunnel to fit the particle plume model. The method with three proposed integration strategies is evaluated in simulation and in the wind tunnel emulating realistic environmental conditions in a repeatable fashion. In particular, we have investigated the impact of two environmental parameters - the wind speed and source release rate on the algorithm performance. The proposed algorithm with the weighted multi-modality map integration strategy outperforms the original Infotaxis and the other two variants. In high wind speed, the proposed algorithm is able on average to estimate the source location with less than 1 meter error in the $80~{m}^{2}$ wind tunnel arena.

A novel odor source localization system based on particle filtering and information entropy

So far, gas leakage caused by natural or human factors has led to serious consequences in terms of social security. Previous strategies for locating the odor sources appear to be either defective or incomplete. For enhancing the success rate and rapidity, this paper aims to present a novel and complete strategy in search of lurking gas sources. Particle filtering and information entropy are both employed to track the plume information. To improve the tracking efficiency in this process, a novel objective function is designed by considering the entropy gains of the suspected targets as well as the repeated exploration scores. Considering the pseudo sourced caused by obstacles, a statistics-based source determine algorithm is proposed to confirm the source’s authenticity, while the artificial potential field method is subsequently applied to eliminate the distractions introduced by the pseudo sources. Simulations and on-site tests are both carried out while results showed that the proposed scheme is competent to complete sources localization task in the scene that contains randomly distributed obstacles and pseudo source.

Olfactory-Based Navigation via Model-Based Reinforcement Learning and Fuzzy Inference Methods

This article presents an olfactory-based navigation algorithm for using a mobile robot to locate an odor source in a turbulent flow environment. We analogize the odor source localization as a reinforcement learning problem. During the odor plume tracing process, the belief state in a partially observable Markov decision process model is adapted to generate a source probability map that estimates possible odor source locations, and a hidden Markov model is employed to produce a plume distribution map that premises plume propagation areas. Both source and plume estimations are fed to the robot, and a decision-making approach based on fuzzy inference is designed to dynamically fuse information from two maps and to balance the exploitation and exploration of the search. After assigning the fused information to reward functions, a value iteration based path planning algorithm is presented to solve for the optimal action policy. Comparing to other commonly used olfactory-based navigation algorithms, such as moth-inspired and Bayesian inference methods, simulation results show that the proposed method is more intelligent and efficient.

Leakage source location based on Gaussian plume diffusion model using a near-infrared sensor

Developing sensitive sensors to measure trace gas concentration levels is important for environmental monitoring. Leakage source location based on measured concentration levels to prevent potential danger is of significance. A near-infrared methane (CH4) sensor system with a modulation frequency of 4 kHz targeting the absorption line at 6046.95 cm−1 and an Allan deviation of 850 parts-per-billion by volume (ppbv) was employed for leakage source location. A static odor source localization algorithm based on the Gauss plume diffusion model and a least square method is proposed. Formulations and equations of leakage location were deduced. An iterative search program was used to obtain the optimal solution. Simulation result showed the synthetic location error of 7.91%, which was close to the experimental result of 8.93%. Feasibility of the location algorithm was verified through the comparison.

Insect olfactory communication in a complex and changing world.

Insect intraspecific olfactory communication occurs in a complex sensory environment. Here we present recent results on how the olfactory system extracts specific information from a sensory background, and integrates it with complementary information to improve odor source localization. Recent advances on mechanisms of olfactory mixture processing, multi-modal integration, as well as plasticity of sensory processing are reviewed. Significant progress in the understanding of neural coding and molecular bases of olfaction reinforce our perception of the tremendous adaptability of insects to a changing environment. However several reports demonstrate that anthropogenic environmental perturbations interfere with insect olfactory communication and might as a consequence significantly alter the functioning of ecosystems and agroecosystems.

Veco-Taxis as a Novel Engineered Algorithm for Odor Source Localization

Algorithms with limited intelligence are unable to localize an odor source in an indoor environment with weak or no airflow. Stage wise solutions to odor source localization has been provided with a novel engineered algorithm called veco-taxis for plume traversal. It finds turn angles by calculating concentration gradients using vector algebra-based search algorithms. Levy walk is used in the plume finding phase. The concept of last chemical detection points (LCDPs) has been adopted for source declaration. The success rate of implemented algorithms is quantified using minimum and maximum move lengths—a key parameter—during source localization. A unified success and performance index (SPI) of the search algorithm is presented for the first time. SPI uncovers implicit parameters accountable for success in locating source and considers a qualitative performance. Higher SPIs are observed when the move length in plume finding is minimum and kept smaller than the plume traversal move length by some factor. It has been also demonstrated through simulations that veco-taxis is superior to the E. coli algorithm.

Single and multiple odor source localization using hybrid nature-inspired algorithm

In this paper, optimization-based approach has been adopted to localize the odor source in an unknown environment. Two scenarios taken into consideration, first single odor source (SOS) with a point source emission at a constant rate and four multiple odor sources (MOS) with point source emissions and different release rates constant in time. In context to SOS, four environments that have distinct dimensional layout have been generated with slight variation in wind velocity and diffusion constant. In case of MOS, there are five environments with same layout but different contributing factors such as wind velocity, placement of odor sources and emission rates which are considered to demonstrate its impact on success rate of algorithms. A recent optimization technique called hybrid teaching learning particle swarm optimization (HTLPSO) has been adopted and implemented in all the arenas, namely SOS and MOS, where mobile robots AKA virtual agents (VAs) are working in collaboration. There are group of VAs deployed in this operation ranging from {3–15}. To investigate the effectiveness of the algorithm, results of HTLPSO are compared with classical particle swarm optimization (PSO) and teaching learning-based optimization (TLBO). It is observed that HTLPSO outperforms TLBO and PSO in arenas with larger dimensions while utilizing few iterations in comparison with other algorithms in case of SOS. HTLPSO also performs best in case of MOS, surviving the effect of wind velocity and change in emission rates. Only when odor sources are placed differently and scattered, TLBO gives the best result. Another highlight of HTLPSO is convergence with high accuracy even with less number of VAs.

Consensus-Based Odor Source Localization by Multiagent Systems Under Resource Constraints.

With advancements in mobile robot olfaction, networked multiagent systems (MASs) are used in odor source localization (OSL). These MASs are often equipped with small microprocessors that have limited computing capabilities, and they usually operate in a bandwidth and energy-constrained environment. The exigent need for a faster localizing algorithm under communication and computational resource constraints invites many design challenges. In this paper, we have designed a two-level hierarchical cooperative control strategy for heterogeneous nonlinear MASs for OSL. The agents are forced toward consensus expeditiously once the information on the whereabouts of the source is attained. The synthesis of the controller occurs in a hierarchical manner-obtaining a group decision, followed by resource-efficient robust control. Odor concentration and wind information have been used in a group decision-making layer to predict a probable location of the source as a tracking reference. This reference is then fed to the control layer that is synthesized using event-triggered sliding-mode control (SMC). The advantage of using event-triggered control scheduling in conjunction with the SMC is rooted in retaining the robustness of the SMC while lowering the resource utilization burden. Numerical simulations confirm the efficiency of the scheme put forth.