Title: Temporal Planning and Plan Execution in a Multi-Agents Context

Subject:

Planning and plan execution control is a critical functionality of embedded autonomous multi-agents systems. Planning is required to allow any system to consider how it can reach a desired state in the future and to produce plans which lead to this goal. Plan execution control has to confront this plan to reality, which often requires the system to adjust, to repair or even to rebuild the whole plan.    Nowadays, there are numerous planning systems and approaches that can produce reasonable plans. However, many of them do not take into account the “execution” phase.
Another aspect, which is critical for embedded multi agents systems, is the management of time. Actions have different duration, they must be executed in parallel, agents must cope with rendezvous, and synchronization. As a result, the planning approach deployed must provide means to explicitly handle time and intrinsically parallel plans.
Last, in a multi-agent environment, the systems will have to handle coordination and cooperation at planning time but also at execution time.
Our recent experiences in this domain lead to the deployment of Temporal Partial Order Causal Link planner using Constraint Based approach (e.g. IxTeT). For various reasons, we are also considering approaches such as Europa/APSI/T-ReX which rely on similar concepts.
The objective of this post doc will be to study, develop and extend one of these approaches to deal with the execution aspect as well as the multi-agent aspect. The resulting system will be integrated in the overall architecture on our robots.

This study is conducted in the framework of the ROSACE project (Robots et Systèmes Auto-adaptatifs Communicants Embarqués – Robots and Embedded Auto-adaptive Communicating Systems) funded by the French Network for Advanced Research dedicated to Science and Technology for Aerospace Systems.  The project is jointly conducted by IRIT, LAAS and ONERA. The project is focused on a multi-robots platform composed of unmanned aerial/ground vehicles. This platform will be provided and maintains by LAAS-CNRS and ONERA as an experimental system that will be used to implement, illustrate and validate the outputs of the project.

Requirements: PhD in a related field (temporal planning, plan execution control, multi-agent system, constraint programming, control architecture, model-based reasoning, distributed systems).


Starting date: Spring 2009
Duration: 1 to 2 years

Applicants should send:
- a resume/bio
- pointers to their most important publications and/or their PhD (only French or English documents)
- recommendation letter(s)

Contact:
Félix Ingrand (felix@laas.fr) Phone: +33 5 61 33 78 04

Laboratory:
LAAS-CNRS, RIS group
7 avenue du Colonel Roche - 31077 Toulouse Cedex 4, France