PIMEO AI targets the development and operational use in multiple representative environments of an artificial intelligence (AI) powered unmanned surface vehicle (USV) that is capable to perform complete suites of water quality measurements in all types of sensitive aquatic ecosystems. The resulting PIMEO AI USV will be a next-generation advanced analysis tool for studying sensitive ecosystems, identifying pollution sources, and mapping their environmental impact. It will fill an important market need for comprehensive water quality USVs, the market today being highly limited and aimed primarily at hydrology research.
PIMEO AI will use an embedded AI system to provide augmented (and eventually automated) piloting making use of machine-learning algorithms and will be connected to a secure cloud data platform to provide archival, centralization and analytics capabilities. Blockchain technology will be used to provide trust and traceability, such as securely managing the sensor data information as well as the identity of the stakeholders. Security and privacy compliance with GDPR will be ensured by implementing reliable, secure data transport and access.
The project will be tested operationally in highly-sensitive and fragile ecosystems that are subject to increasing anthropic pressures. Three distinct ecosystems will be studied:
- France: a lake in a dense urban environment (Créteil lake), used for recreational activities but subject to regular pollution events;
- Romania: the Danube delta – one of the largest and the best-preserved European river deltas, subject to different types of pollution and eutrophication
- Romania: coastal waters in the Constanta area, affected by frequent microbiological pollution episodes.
PIMEO AI has strong support of several key stakeholders: GrandParis Sud-Est Avenir (France), Cluster EMS (France), Constanta National Company Maritime Port Administration (Romania) Cluster IND-AGRO-POL (Romania) and the Romanian Water National Authority (letters of intent).
Project code: ERANET-MARTERA-PIMEO-AI-1
Project title: Pollution Identification, Mapping, and Ecosystem Observation with AI-powered water quality USV
Acronym: PIMEO-AI
The start date of the contract: 09/03/2020
The end date of the contract: 28/02/2023
The total value of the project: 627.000 lei / 132.000 Euro
Duration: 36 months
Project financed by UEFISCDI through the European and International Cooperation Program
Consortium
FLUIDION – France
LOCEAN – France
BEIA Consult International –Romania
Constanta Maritime University, Electronics and telecommunications –Romania
Institute pf Geography, Romanian Academy, ENVIRONMENT AND GIS – Romania
Project’s Phases and Activities
| Phase 1 | Identify requirements and develop usage scenarios |
| A1.1 | Establishing user and cloud platform requirements |
| A1.2 | Establishing scenarios for case studies |
| A1.3 | Establish requirements for data sets used to drive AI algorithms |
| A1.4 | Communication and dissemination of results from stage I |
| 2nd Phase | Design and development of the PIMEO AI platform |
| A2.1 | PIMEO AI platform design (platform architecture) |
| A2.2 | Preliminary tests in case studies |
| A2.3 | Generating data sets to drive ML algorithms |
| A2.4 | Development of the PIMEO AI platform and testing of its components |
| A2.5 | Communication and dissemination of the results of the second stage |
| 3rd Phase | Integration and testing of the PIMEO AI platform |
| A3.1 | Integration of the PIMEO AI platform |
| A3.2 | Testing and validation of the PIMEO AI platform in real conditions (in case studies) |
| A3.3 | Communication and dissemination of the results of the third stage |
| 4th Phase | Evaluation and exploitation of results |
| A4.1 | Evaluation of test results, development of guides and recommendations |
| A4.2 | Exploitation plan of results |
| A1.4 | Communication and dissemination of the results from stage IV |
Phase I – Identification of requirements and development of use cases – November 2020
This phase reviewed the current state of the art on the use of unmanned surface vehicles (USV) in environmental water quality monitoring in sensitive estuarine, urban and coastal environments. The requirements of the users and the platform to be developed were identified, as well as the requirements for the datasets used in training the machine learning algorithms. The report is complemented by the presentation of scenarios and related data flows.
A1.1 Establishing user and cloud platform requirements
A1.1 presents a review of the current state of knowledge in the field of water quality monitoring and the use of unmanned surface vehicles. The analysis is complemented by setting out the requirements for users and the cloud platform to be developed as part of the project.
PIMEO AI meets the new administrative requirements and expands on those already mentioned by being able to monitor water quality over a wide area and provides a pollution source identification system.
A1.2 Establishing scenarios for case studies
A1.2 presents the case studies, scenarios and data flow for each scenario. PIMEO AI will be operationally tested in highly sensitive and fragile ecosystems that are subject to increasing anthropogenic pressure. The data collected will help assess the health of the ecosystems studied, quantify anthropogenic impacts and provide a reliable in-situ reference for modelling. The cloud platform will allow large-scale access to data in specific formats for all categories of potential users (scientific community, stakeholders, population).
The main water pollution issues for case study #1 (Creteil Lake) are the high nutrient content of the water which increases the number of aglets and can lead to eutrophication. The full description of this case study will be carried out by the French partners (FLUIDION and Sorbonne University).
Case study #2 is the Danube Delta. In this area water quality parameters will be studied, relevant parameters for the verification of the criteria are bathing, water stratification in narrow channels, etc. The detailed presentation of this area is made by the Institute of Geography of the Romanian Academy in their report.
Case study #3 is the port area of Constanta and will focus on water quality in terms of hydrocarbon pollution from ship traffic, but also on bathing water quality. The detailed presentation of this case study is made by the Maritime University of Constanta.
The purposes for the case studies addressed are: conservation of flora and fauna, reduction of water pollution and protection of human health.
A1.3 Establish requirements for data sets used to drive AI algorithms
A1.3 reviewed the requirements for datasets used in training machine learning algorithms. It also presented and benchmarked existing algorithms as well as other applications in which they have been used. A comprehensive state of the art of AI and ML algorithms has been approached in this activity for the purpose of finding the most suitable algorithms for the PIMEO AI project.
Data validation was also reviewed in this phase because when discussing Artificial Intelligence (or Machine Learning) algorithms, data errors can propagate and affect the final result of the algorithm.
A1.4 Communication and dissemination of results from stage I
The communication and dissemination activities of the project results are presented in A1.4. These included participation in events, publication of scientific articles, publication of posts on Twitter and on the website, etc.).
Results
The unmanned aquatic vehicle to be used in this project has features that allow it to perform complex water quality analyses (e.g. microbiological – E.coli, nutrient monitoring, water depth profiling) in sensitive estuarine, urban and coastal environments.
The USV PIMEO AI will be complemented by a cloud platform that will allow storage, processing, analysis, and visualization of the collected data. The report outlines the main functionalities to be met, the criteria for evaluating the performance of the system and the specifications of the sensors to be used. For the development of the artificial intelligence component in the report, the requirements for the datasets used in the training of machine learning algorithms were analyzed. Existing algorithms and other applications where they have been used are also presented and benchmarked.
PIMEO AI will be operationally tested in highly sensitive and fragile ecosystems that are subject to increased anthropogenic pressures. The scenarios and data flow for each scenario have been analyzed in a first step and will be detailed in the second phase.
Thus, the results of this report are to be used in the next phase.
Posters and publications:
Iordache G., Bălănescu M., Suciu G., Bîrdici A., Pasat A., Zătreanu I., 2021, The PIMEO AI project – a Cloud based platform for water quality monitoring, in „The 23rd Conference on Control Systems and Computer Science” – CSCS23, 2021, pp. 529-535, doi:10.1109/CSCS52396.2021.00092. https://ieeexplore.ieee.org/document/9481025
A second flyer of the project was designed, shared on social media and printed for offline events to enhance the visibility of the project.
Phase II – Design and development of the PIMEO AI platform – November 2021
In this phase, the requirements identified in Phase I of the project regarding the architecture of the PIMEO AI platform were implemented. At the same time, the methodologies for implementing and testing the components underlying the PIMEO AI architecture are presented.
Activity 2.1 – PIMEO AI platform design (platform architecture)
In A2.1, the PIMEO AI architecture was designed and it presents the PIMEO AI components, describing the level of data acquisition from sensors (USV type device – Unmanned Surface Vehicle or Unmanned Surface Vehicle), communication / data transmission components for centralization in Fluidion Cloud, as well as PIMEO AI specific components that allow interfacing through various protocols (MQTT / REST) with Fluidion Cloud to allow storage and presentation of measurements made in PIMEO AI Cloud.
Activity 2.2 – Preliminary tests in case studies
In this activity, the PIMEO AI platform has been preliminarily tested operationally in highly sensitive and fragile ecosystems that are subject to increased anthropogenic pressure. The data collected will help to assess the health of the studied ecosystems, to quantify the anthropogenic impact and provide a reliable in-situ reference for modeling. The cloud platform allows large-scale access to data in specific formats for all categories of potential users (scientific community, stakeholders, population).
In order to carry out the preliminary tests, areas were selected to cover as diverse requirements as possible. Thus, tests were performed in:
(i) surface waters (Danube river, Rosu lake, Rosulet lake)
(ii) beach areas located on the Black Sea (Sulina, Vadu, Corbu)
(iii) the port of Constanța.
The tests took place between August 26 and September 1, 2021 with the participation of the following partners: Fluidion (France), Beia Consult International, Institute of Geography of the Romanian Academy, Maritime University of Constanța
Activity 2.3 – Generating data sets to drive ML algorithms
In A2.3 the process of generating the data sets for the training of machine learning algorithms was carried out only in France on Lake Creteil due to delays generated by the COVID-19 pandemic in the delivery of specific electronic components and chemicals.
Activity 2.4 – Development of the PIMEO AI platform and testing of its components
A2.4 presents the implementation methodologies and testing of PIMEO AI components. The first part of the chapter describes the hardware requirements implemented for running the software components described in A2.1, then discussing the method of implementing the MQTT broker, the Connector developed based on AlertLab API, database implementation and presentation component (view of data), respectively, the development and testing of the blockchain component.
A2.5 – Communication and dissemination of the results of the second phase
The communication and dissemination activities of the project results are presented in A2.5. These included participation in events, publication of scientific articles, publication of posts on Twitter and on the website, etc.).
The first conference with potential beneficiaries of the project took place on 20.07.2021 in order to identify their interests in monitoring water quality in the Danube Delta and Black Sea area. The National Research Institute of the Danube Delta – INCDDD, the National Research and Development Institute for Marine Geology and Geoecology GeoEcoMar and the National Research and Development Institute for Biological Sciences participated.
During the preliminary tests, discussions took place with several entities potentially interested in the results of the PIMEO AI project: the Inspectorate for Emergency Situations, Sulina City Hall, Vadu City Hall, Corbu City Hall, National Environmental Guard, Environmental Manager and Operational Director of Constanța Port.
During the current phase of the project, the scientific results were also disseminated through the drafting and presentation of a scientific article:
Iordache G., Bălănescu M., Suciu G., Bîrdici A., Pasat A., Zătreanu I., 2021, The PIMEO AI project – a Cloud based platform for water quality monitoring, in „The 23rd Conference on Control Systems and Computer Science” – CSCS23, 2021, pp. 529-535, doi:10.1109/CSCS52396.2021.00092. https://ieeexplore.ieee.org/document/9481025
Results:
In the current phase of the project, the implementation methodologies and testing of PIMEO AI components was performed. The hardware requirements implemented for running the software components and the method of implementing the MQTT broker, the Connector developed based on AlertLab API, the implementation of the database and the presentation component (data visualization), respectively, the development and testing of the blockchain component have been described.
The components of the PIMEO AI platform have been preliminarily tested in selected areas to cover as diverse requirements as possible. Thus, tests were performed in: (i) surface waters (Danube river, Red lake, Red lake), (ii) beach areas located on the Black Sea (Sulina, Vadu, Corbu) and (iii) Constanța port.
The PIMEO AI USV is completed with a cloud platform that will allow the storage, processing, analysis and visualization of the collected data. The report presents the architecture of PIMEO AI components, describing the level of data acquisition from sensors (USV type device – Unmanned Surface Vehicle or Unmanned Surface Vehicle), communication / data transmission components for centralization in Fluidion Cloud, as well as components specific PIMEO AI that allow interfacing through various protocols (MQTT / REST) with Fluidion Cloud to allow storage and presentation of measurements made in PIMEO AI Cloud.
The results of the project were disseminated by participating in national and international events, seminars and workshops. The identity of the project was also established by creating a logo, a web page and an account on a social network. The dissemination of scientific results was achieved by publishing an article in the volume of an international conference.
In the next stage, the integration and testing of the PIMEO AI platform will be performed, including in real conditions in the case studies.
Phase 3 – Integration and testing of the PIMEO AI platform
In this phase, we achieved the integration and validation of the PIMEO AI platform components. The integration methodology followed through the platform architecture and components data flow.
A3.1 Integration of the PIMEO AI platform
Activity A3.1 managed the integration methodology for the PIMEO AI components. The main achievements for this deliverable regards the implementation of a third party water quality monitoring solution.
A3.2 Testing and validation of the PIMEO AI platform in real conditions (in case studies)
Activity A3.2 presented the validation results of the preliminary case studies performed last year. In this phase, we managed the integration of the data collected in the preliminary case studies in the PIMEO AI data visualization components. Also, since the new geopolitical conditions, the microbiological samples were collected from another site location (Constanta, Beach area).
A3.3 Communication and dissemination of the results of the third stage
The communication and dissemination activities of the project results are presented in A3.3. These included participation in events, publication of scientific articles, publication of posts on Twitter and on the website, etc.). The participation at GoTechWorld 2022 edition was a success for disseminating the PIMEO AI project results. Also, during the current phase of the project, the scientific results were also disseminated through the following papers:
M. Paun, S. Bucuci, R. D. Tamas, G. Suciu and M. Balanescu, “Wireless channel propagation characterization for USV communication using UWB signals,” 2021 IEEE Conference on Antenna Measurements & Applications (CAMA), 2021, pp. 153-156, doi: 10.1109/CAMA49227.2021.9703535. https://ieeexplore.ieee.org/abstract/document/9703535
“A novel electronic switch for VHF/UHF low-cost radars”, Andreea Furtuna, M. Pastorcici, R. D. Tamas, Constanta Maritime University, M. Bălănescu, V. Suciu, G. Suciu, BEIA Consult International, Poster presentation.
Phase 4 – Evaluation and Exploitation of Results
The objectives of the current stage, entitled ‘Evaluation and Exploitation of Results’ and with a deadline for submission of February 28, 2023, are as follows:
● Evaluation of test results and creation of guidelines and recommendations.
● Results exploitation plan.
● Communication and dissemination of results from Phase IV.
Thus, the main objectives of Phase IV of the PIMEO AI project were achieved through the development of a document containing guidelines and recommendations for the users of the PIMEO AI platform, the completion of the exploitation plan, and the fulfilment of the dissemination activities, as presented in a summary report of the results obtained throughout the project period.
Communication and dissemination of the results of the fourth stage:
The second conference with potential beneficiaries of the project took place on February 23, 2023, with the aim of identifying their interests regarding water quality monitoring in the Danube Delta and the Black Sea area. The National Institute for Research and Development of the Danube Delta – INCDDD, the National Institute for Research and Development in Geology and Marine Geoecology GeoEcoMar, the National Institute for Research and Development for Biological Sciences, as well as representatives of the Natural Heritage Administration – DELTA DUNĂRII BIOSPHERE RESERVE ADMINISTRATION and the National Administration of Romanian Waters, participated in the conference.
Posters and publications
Balanescu, M., Suciu, G., Badicu A., Birdici A., Pasat A., Poenaru, C., Zatreanu I., 2020. Study on Unmanned Surface Vehicles used for Environmental Monitoring in Fragile Ecosystems. 2020 IEEE 26th International Symposium for Design and Technology in Electronic Packaging (SIITME),. https://ieeexplore.ieee.org/document/9292219Iordache G., Bălănescu M., Suciu G., Bîrdici A., Pasat A., Zătreanu I., 2021. The PIMEO AI project – a Cloud based platform for water quality monitoring, in „The 23rd Conference on Control Systems and Computer Science” – CSCS23, 2021, pp. 529-535, doi:10.1109/CSCS52396.2021.00092. https://ieeexplore.ieee.org/document/9481025
M. Paun, S. Bucuci, R. D. Tamas, G. Suciu and M. Balanescu, 2021. Wireless channel propagation characterization for USV communication using UWB signals, 2021 IEEE Conference on Antenna Measurements & Applications (CAMA), 2021, pp. 153-156, doi: 10.1109/CAMA49227.2021.9703535. https://ieeexplore.ieee.org/abstract/document/9703535
“A novel electronic switch for VHF/UHF low-cost radars”, Andreea Furtuna, M. Pastorcici, R. D. Tamas, Constanta Maritime University, M. Bălănescu, V. Suciu, G. Suciu, BEIA Consult International, Poster presentation.
Website
Check it here.
Social media
Check it here.
Acknowledgement
This work was supported by a grant of the Romanian National Authority for Scientific Research and Innovation, CCCDI – UEFISCDI, project number ERANET-MARTERA-PIMEO-AI-1, within PNCDI III