This postdoc topic concerns the perception of natural environment by heterogeneous robots (unmanned aerial and ground vehicles). The postdoctoral  researcher will participate to the ROSACE project  funded by the RTRA STAE (in french, Fondation de coopération scientifique, Sciences et Technologies pour l'Aéronautique et l'Espace) : ROSACE studies how a set of mobile autonomous communicating and cooperating robots could be deployed  with well-established properties particularly in terms of safety, self-healability, ability to achieve a set of missions and self-adaptation in a dynamic environment. Concerning robotics, the objective consists in studying and developing distributed  robotics  functions to be integrated on  aerial and terrestrial robots.

ROSACE partners have defined a common scenario about rescue robotics, on which every research team must propose functions, multi-robot strategies and, when possible, experiments either in simulation or with true robots :

• rescue robotics : an alert occurs about an accident or a possible disaster on a large area  (e.g. a forest fire) ! An emergency service deploys robots, both with human-controlled vehicles.
• An a priori map is known on the area, but, this map must be considered as uncertain (new buildings, new roads …) … moreover due to the disaster, some roads are probably cut off by obstacles.
• Robots must cooperate in order to detect some situations (e.g. fire, humans or vehicles  in dangerous area…) and to  help people (e.g. guide humans towards safe areas)
• A communication network is deployed in the environment, but with a limited bandwidth and with partial connectivity between nodes (humans and robots).
• GPS is available, but navigation cannot be completely based on GPS-based localization (not accurate enough, possible occultations …)

Such a scenario gives only an applicative context in order to study scientific topics in robotics, multi-agents systems, AI, Ad hoc network, Fault tolerant systems ….

This postdoc topics is focused on mapping from several heterogeneous robots, using a priori information on a large area to be explored. The postdoc will be integrated in the Robotics and AI department of LAAS-CNRS in Toulouse  (http://www.laas.fr), but will have to cooperate with the DCSD department at ONERA also in Toulouse.

Some keywords only to give a general context : environment modelling, 3D heterogeneous  representations (DEM, landmarks, specific objects to be identified  like roads, rivers. buildings, fire…), SLAM, multiple and heterogeneous maps, large environment, vision, multi-robots cooperation, network connectivity…

A recent PhD is required. Applicants should possess knowledge and experience in SLAM, 3D vision, data association, estimation… Appointment will be initially for a 12 months duration, eventually  extended for 12 other months. The position is available now : the postdoc may start as soon as possible. Annual gross salary will be around 34000€ (annual net salary: ~ 27 000€).

    Letters of application, a C.V. with complete list of publications, a short statement of research interests and the name of two referees should be sent before 2009 December 20th by e-mail to Michel Devy:

Michel Devy
LAAS-CNRS, RAP group
E-mail: michel@laas.fr
Phone: +33 561 336 331