Resilient Environmental Monitoring Utilizing a Machine Learning Approach
A wide range of regulations is established to protect citizens health from the noxious consequences of aerosols, e.g. particulate matter (PM10). To ensure a public information and the compliance to given regulations, a resilient environmental sensor network is necessary. This paper presents a machine learning approach which utilizes low-cost platforms to build a resilient sensor network. In particular, malfunctions are compensated by learning virtual models of various particulate matter sensors. Such virtualized sensors are already utilized in the field of proprioceptive robotics  and are comparable to a digital twins definition. Several experiments show the proposed method yields PM10 estimates and forecasts similar to high-performance sensors.
KeywordsEnvironmental monitoring Virtual sensor Machine learning Volunteered geographic information
- 1.Berger, E.: Behavior-specific proprioception models for robotic force estimation: a machine learning approach, Ph.D. thesis (2018)Google Scholar
- 5.McCrory, G., Veeckman, C., Claeys, L.: Citizen science is in the air – engagement mechanisms from technology-mediated citizen science projects addressing air pollution. In: Kompatsiaris, I., et al. (eds.) INSCI 2017. LNCS, vol. 10673, pp. 28–38. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-70284-1_3CrossRefGoogle Scholar
- 6.Blon, M.: Research regarding PM measurements. The citizen science project luftdaten.info, M.Sc. thesis, University of Esslingen (2018). (In German)
- 12.Berger, E., Vogt, D., Grehl, S., Jung, B., Amor, H.B.: Estimating perturbations from experience using neural networks and information transfer. In: IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Daejeon, pp. 176–181 (2016)Google Scholar
- 15.Srivastava, R.K., Greff, K., Schmidhuber, J.: Training very deep networks. In: Advances in Neural Information Processing Systems, pp. 2377–2385 (2015)Google Scholar