The objective of the AILERON project is to develop a methodology and algorithms to help maritime decision making, when several decision-makers are involved, as well as to propose a software platform, called DESEASION, to support the decision aiding process and to facilitate the collaboration on marine assessment or decision making issues. In that context, Maritime Spatial Planning (MSP) is a possible application, which could benefit from this general methodology and tool. The software platform is developed so that it can bring together the different actors of the decision problem around the same tool, either locally or remotely. The platform is currently under development.
Questions this practice may help answer
- How can stakeholders preferences and decision-makers’ constraints be brought together with the support of a software platform?
- How can compromises between different maritime uses be identified during the decision-making process with the support of a software platform?
The DESEASION platform is being developed in the context of a scientific collaboration called AILERON, between the French National Hydrographic Service (SHOM), IMT Atlantique, and the Lab-STICC CNRS laboratory.
Aspects / Objectives
The objective of the project is to develop a methodology and algorithms to help maritime decision making when several decision-makers are involved, as well as to propose a software platform called DESEASION.
DESEASION addresses the decision problem via several (possibly iterative) steps:
- Hierarchical structuring of the decision problem, where sub-problems are identified, actors and decision makers are determined and the necessary data is collected and uploaded.
- Evaluation of the area, which consists of applying a selection of evaluation models (expert rules, aggregated impacts, multi-criteria decision aiding), based on the decision-makers’ expertise, to obtain the overall assessment of the various involved geographical zones.
- Area recommendations for maritime activities, based on constraints expressed by the decision-makers concerned.
- Validation of the final recommendation through a guided explanation of the results and a negotiation phase (which can result in returning to previous steps with updated information).
During the evaluation step (2), three possible evaluation models can be implemented.
- Aggregation models from the field of Multi criteria decision aiding, which integrate precise preference models of the decision maker with the data to generate evaluations respecting the decision maker’s priorities. The preferences are determined through supervised learning algorithms.
- Expert rules are inferred from the decision maker’s expertise.
- Or from aggregation operators, as weighted sums.
All the manipulated data are in vector format, but it is also possible to add raster data to the platform through a conversion tool. The attributes can be either numerical or textual. The maps thus generated are exportable in shape format.
Once the assessment step (2) is finalized (corresponding to the final decision map that has been generated), it is possible to use an algorithm that generates several recommendations for locating an activity (step 3). The algorithm behind this functionality is a genetic algorithm that proposes a set of solutions, which facilitate the discussion necessary to reach a consensus, for example for MSP.
The architecture of the software platform is based on a web solution in which the data are stored on a server and calculations are performed remotely. Consequently, the users do not have to install anything on their computers; heavy calculations are transferred to servers making it easier for users to collaborate on the same project (in particular if they are not located at the same place, or in an asynchronous way).
Main Outputs / Results
- Hierarchical structuring of the decision problem
- Accurate modelling of the multiple objectives involved in the decision
- Integration of the perspectives of the multiple stakeholders
- Distant or local collaboration around a unique tool
- Qualitative or quantitative evaluation of the geographical zones, based on the multiple objectives of the decision problem
- Integration of the human perspective in this evaluation
- Taking into account the preferences of the multiple stakeholders
- Use of multiple evaluation models
- Taking into account complex constraints of the stakeholders
- Automatic generation of recommendations of zones
- Comparative study of multiple zones
- Human readable explanations of recommendations
- Traceability and justification of the decision recommendation
- Search for a compromise satisfying the actors of the decision problem
Next to MSP, other possible use cases are: impact assessment on ecological ecosystems, installation of renewable energy sources, controlled ship stranding, development of maritime areas or the coast.
French Hydrographic and Oceanographic Service (Shom), IMT Atlantique, CNRS laboratory Lab-STICC.
IMT Atlantique Bretagne Pays de la Loire