Filtern
Dokumenttyp
- Beitrag zu einem Tagungsband (11)
- Vortrag (10)
- Posterpräsentation (3)
- Zeitschriftenartikel (1)
Sprache
- Englisch (23)
- Deutsch (1)
- Mehrsprachig (1)
Schlagworte
- Mobile Robot Olfaction (8)
- Sensor network (8)
- Environmental monitoring (7)
- Low-cost (7)
- Gas Distribution Mapping (4)
- Spatial Interpolation (4)
- Wireless Sensor Network (4)
- Dust sensor (3)
- LoRaWAN (3)
- Occupational health (3)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (10)
Gas distribution mapping is important to have an accurate understanding of gas concentration levels in hazardous environments. A major problem is that in-situ gas sensors are only able to measure concentrations at their specific location. The gas distribution in-between the sampling locations must therefore be modeled. In this research, we interpret the task of spatial interpolation between sparsely distributed sensors as a task of enhancing an image's resolution, namely super-resolution. Because autoencoders are proven to perform well for this super-resolution task, we trained a convolutional encoder-decoder neural network to map the gas distribution over a spatially sparse sensor network. Due to the difficulty to collect real-world gas distribution data and missing ground truth, we used synthetic data generated with a gas distribution simulator for training and evaluation of the model. Our results show that the neural network was able to learn the behavior of gas plumes and outperforms simpler interpolation techniques.
Gas distribution mapping is important to have an accurate understanding of gas concentration levels in hazardous environments. A major problem is that in-situ gas sensors are only able to measure concentrations at their specific location. The gas distribution in-between the sampling locations must therefore be modeled. In this research, we interpret the task of spatial interpolation between sparsely distributed sensors as a task of enhancing an image's resolution, namely super-resolution. Because autoencoders are proven to perform well for this super-resolution task, we trained a convolutional encoder-decoder neural network to map the gas distribution over a spatially sparse sensor network. Due to the difficulty to collect real-world gas distribution data and missing ground truth, we used synthetic data generated with a gas distribution simulator for training and evaluation of the model. Our results show that the neural network was able to learn the behavior of gas plumes and outperforms simpler interpolation techniques.
Remote gas sensors mounted on mobile robots enable the mapping of gas distributions in large or poorly accessible areas. A challenging task however, is the generation of three-dimensional distribution maps from these spatially sparse gas measurements. To obtain high-quality reconstructions, the choice of optimal measuring poses is of great importance. Remote gas sensors, that are commonly used in Robot Assisted Gas Tomography (RAGT), require reflecting surfaces within the sensor’s range, limiting the possible sensing geometries, regardless of whether the robots are ground-based or airborne. By combining ground and aerial robots into a heterogeneous swarm whose agents are equipped with reflectors and remote gas sensors, remote inter-robot gas measurements become available, taking RAGT to the next dimension – releasing those constraints. In this paper, we demonstrate the feasibility of drone-to-drone measurements under realistic conditions and highlight the resulting opportunities.
In many cases, science falls back on self-developed prototype systems, which are used and developed for the measurement and execution of the tasks. About 80 % of the development is based on the same hardware design, which is used in only one application scenario and then discarded. For the most part, there are also uncalibrated sensors, since it is costly to calibrate a complicated sensor measurement system or the entire measurement chain since access to the sensor systems is not always available. This paper describes a conceptual design to implement a versatile sensor system with the motivation to fuse the data recording and data reception, which can cover large areas with the help of LoRaWAN (Long Range Wide Area Network) technology. To overcome disadvantages of LoRaWAN, namely the slow data rates, the proposed sensor system can also cover smaller areas with the widespread WiFi technology. An enormous advantage over individual complete systems in the form of a prototype, is the rapid expansion, uncomplicated calibration of the individual sensors and the ecological relief.
A modular design is used, where individual stacks with sensors and peripherals can be added separately. The stacks are standalone low-power systems and can be calibrated, maintained, and replaced separately and do not require the entire measurement chain. The measured and sent values are stored locally on the main stack and sent to the data collector (gateway) and evaluated by means of the automated selection between WiFi and LoRaWAN.
To better understand the dynamics in hazardous environments, gas distribution mapping aims to map the gas concentration levels of a specified area precisely. Sampling is typically carried out in a spatially sparse manner, either with a mobile robot or a sensor network and concentration values between known data points have to be interpolated. In this paper, we investigate sequential deep learning models that are able to map the gas distribution based on a multiple time step input from a sensor network. We propose a novel hybrid convolutional LSTM - transpose convolutional structure that we train with synthetic gas distribution data. Our results show that learning the spatial and temporal correlation of gas plume patterns outperforms a non-sequential neural network model.
To better understand the dynamics in hazardous environments, gas distribution mapping aims to map the gas concentration levels
of a specified area precisely. Sampling is typically carried out in a spatially sparse manner, either with a mobile robot or a sensor network and concentration values between known data points have to be interpolated. In this paper, we investigate sequential deep learning models that are able to map the gas distribution based on a multiple time step input from a sensor network. We propose a novel hybrid convolutional LSTM - transpose convolutional structure that we train with synthetic gas distribution data. Our results show that learning the spatial and temporal correlation of gas plume patterns outperforms a non-sequential neural network model.
Wireless sensor networks provide occupational health experts with valuable information about the distribution of air pollutants in an environment. However, especially low-cost sensors may produce faulty measurements or fail completely. Consequently, not only spatial coverage but also redundancy should be a design criterion for the deployment of a sensor network. For a sensor network deployed in a steel factory, we analyze the correlations between sensors and build machine learning forecasting models, to investigate how well the sensor network can compensate for the outage of sensors. While our results show promising prediction quality of the models, they also indicate the presence of spatially very limited events. We, therefore, conclude that initial measurements with, e.g., mobile units, could help to identify important locations to design redundant sensor networks.
Sensors can fail. Redundancy should therefore be a design driver of wireless sensor networks. For a sensor network deployed in a steel factory, we analyze the correlations between sensors and build machine learning forecasting models, to investigate how well the network can compensate for the outage of sensors.