L’Institut de Recherche Technologique (IRT) Saint Exupéry est un accélérateur de science, de recherche technologique et de transfert vers les industries de l’aéronautique et du spatial pour le développement de solutions innovantes sûres, robustes, certifiables et durables.
Nous proposons sur nos sites de Toulouse, Bordeaux, Montpellier, Sophia Antipolis et Montréal un
environnement collaboratif intégré composé d’ingénieurs, chercheurs, experts et doctorants issus des milieux industriels et académiques pour des projets de recherche et des prestations de R&T adossés à des plateformes technologiques autour de 4 axes : les technologies de fabrication avancées, les technologies plus vertes, les méthodes & outils pour le développement des systèmes complexes et les technologies intelligentes.
Nos technologies développées répondent aux besoins de l'industrie, en intégrant les résultats de
la recherche académique.
Note : Le genre masculin est utilisé dans le seul but d’alléger le texte.
The IRT is a collaborative and integrated technological research center bridging the public research to the industrial one. Technologies that are developed answer to industrial needs, benefiting of the academic researches.
IRT Saint Exupéry is a private research foundation supported by the French State funding projects in proportion to industrial contribution and defining the regulatory framework of the foundation.
The IRT Saint Exupery is an essential player in French and international aerospace industry through its multidisciplinary expertise, human knowledge and skills.
We anticipate the innovation needs of our members and clients, and develop technologies in line with the foremost environmental issues.
That gives meaning to innovation’ projects, sustain its employees passion and make IRT Saint Exupery appealing to new talents.
The selected candidate will participate to the SB project led by IRT St Exupéry. This project aims to apply integrate hybrid Artificial Intelligence based on physicals and Machine learning for monitoring the local nearshore environment, especially bathymetric evolution in shallow waters.
The work will be done in collaboration with the LabCom KOSTARISK. The LabCom KOSTA RISK (Kosta = Coast in Basque) is a project for a joint cross-border laboratory for applied research in observation and modelling to support coastal risk management.
It associates the Wave Interaction and Structure team of the SIAME laboratory, the Spanish technological center AZTI and the monitoring and forecasting center Rivages Pro Tech (RPT) of the SUEZ group, relying on a strong complementarity of their respective expertise, built on the basis of a collaboration of more than 15 years.
Nearshore processes are strongly influenced by bathymetry features. Thus, the accurate determination of the bathymetry is usually a prerequisite to the application of any nearshore modelling application. Uncertainty related to bathymetry is especially a current limitation of most early warning systems designed for coastal risk management such as, for instance, the prediction of surfzone currents for swimmer safety or the computation of nearshore wave field for coastal flooding risk.
Traditional bathymetry survey, usually based on echo sounder mounted on a floating vessel, provides accurate measurements of bed morphology. However, this method is costly and requires restrictive climatic conditions that limit their deployment.
Alternative bathymetric estimation methods based on remote sensing techniques have received increasing interest in the last decades. Among those techniques, video-based bathymetry estimation systems afford numerous advantages. They rely on video monitoring station that can be remotely controlled and can provide data whatever the climatic conditions. Two types of approach are usually applied to derive a bathymetry estimation in the nearshore zone using coastal imagery: (i) physics-based inversion methods or (ii) data assimilation technique based on Machine Learning (ML) techniques. Physics-based approaches like e.g. cBathy system (Holman et al., 2013) has proven to provide accurate estimate of nearshore bathymetry based on the computation of wave phase celerity from hourly time stack images collection and the application of linear wave theory.
However, its performance can vary as a function of the tidal range (Bergsma et al., 2016) and of the incident wave energy (Broadie et al., 2018). More recent alternative ML approaches using e.g. convolutional neural network models were proposed and tested on synthetic datasets with promising results in an extended range of wave conditions, possibly overcoming several of the limitations related to physics-based approaches (Collins et al., 2020).
In this context, this Post-Doctoral project aims to investigate the potential of new inversion methods based on ML techniques in order to estimate nearshore bathymetry for complex beach environments. Studied environments may feature either fast morphology changes, nearshore circulation controlled by embayment and geological heritage, or pronounced tidally modulated water levels.
The work proposed will first consist in testing a ML approach on an embayed beach located in a meso-tidal environment, taking benefit of the available historical data. Second, the method will be replicated on an open beach located in a micro-tidal environment with large expected sediment transport during storms.
More precisely, the following actions will be carried out:
Bergsma, E. W. J., Conley, D. C., Davidson, M. A., & O'Hare, T. J. (2016). Video-based nearshore bathymetry estimation in macro-tidal environments. Marine Geology, 374, 31-41.
Brodie, K.L.; Palmsten, M.L.; Hesser, T.J.; Dickhudt, P.J.; Raubenheimer, B.; Ladner, H.; Elgar, S.C.E. Evaluation of video-based linear depth inversion performance and applications using altimeters and hydrographic surveys in a wide range of environmental conditions. Coast. Eng. 2018, 136, 147–160.
Collins, A. M., Brodie, K. L., Spicer, B. A., Hesser, T. J., Farthing, M. W., Lee, J., & Long, J. W. (2020). Bathymetric Inversion and Uncertainty Estimation from Synthetic Surf-Zone Imagery with Machine Learning. Remote Sensing, 12(20), 3364.
Holman, R., Plant, N., & Holland, T. (2013). cBathy: A robust algorithm for estimating nearshore bathymetry. Journal of Geophysical Research: Oceans, 118(5), 2595-2609.
PhD in Machine/Deep Learning, computer vision
Experience: Research & Development in artificial intelligence