[agents] Fully-funded PhD proposal on multiagent planning

[Laurent Vercouter]

A PhD position in computer science is open at INSA Rouen, LITIS laboratory starting on the 1st September 2019 at the latest. This PhD position is funded as part of the international project IoT.H2O (https://www.mv.uni-kl.de/IoTDotH2O/en/home-2/).

Candidates are invited to contact me by email (laurent.vercouter at insa-rouen.fr) to get more information or to submit an application file including at least a CV, a cover letter and the latest transcripts.

Thesis title: Multi-agent planning for the control of a water distribution network

Context: the IoT.H2O project

This PhD thesis proposal is part of the IoT.H2O project led by an international consortium of research teams from Germany (Kaiserslautern, Aachen), Belgium (Liège), Brazil (Belo Horizonte) and France (Rouen). The objective of the project is to supervise, control and optimize the operation of large water distribution networks in a decentralized manner using an Internet of Things. The deployment of connected devices will make it possible to capture data at several locations on the network and adapt local operation based on data analysis. This analysis may lead to detect faults, to adapt the networks functioning or to optimize local decision-making.

PhD thesis objective: The thesis participates to the Workpackage 4 of the project entitled "Multiagent technology for decentralized AI" carried out by the LITIS laboratory (Rouen, France). The objective is to propose a decentralized coordination model allowing a network subsystem to supervise its operation based on sensor data, optimize its operation and detect and react to possible failures (sensor or actuator failures, leaks, etc.). Existing control systems are currently centralized and provide a global representation of the network's operation that is too complex and unreactive. The expected thesis work consists of three tasks:
- the definition of a model describing the functioning of a water distribution network, including both information related to its condition and the various actions that can be carried out. This task will be carried out in collaboration with the project partners who are experts in the field of water distribution networks
- the proposal of a multi-agent planning algorithm to define the sequences of actions to be carried out in a subsystem, taking into account possible dependencies with other subsystems
- the use of reinforcement learning to optimize planning through simulations.
The theoretical contributions of the thesis will be implemented by simulating a water distribution network.


PhD supervisor: Laurent Vercouter, LITIS/INSA Rouen Normandie

Start of the thesis: September 1, 2019 at the latest

Duration: 3 years

Location: INSA Rouen Normandie, Saint-Etienne du Rouvray, France

Net monthly salary: 1390 € (excluding teaching assignment)

—
*Laurent Vercouter*
Co-directeur du LITIS EA 4108
Coordinateur du Pôle Normand Stratégique de Recherche et de Formation Sciences du Numérique
Professeur des Universités en Informatique
Fédération CNRS NormaSTIC FR 3638
LITIS, INSA Rouen Normandie, France
Ph: (+33) 2 32 95 99 86

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cs.umbc.edu/pipermail/agents/attachments/20190404/10427db9/attachment.html>