EXPRES: Automated exploration techniques for rescue aplications
The main project objective is the research in exploration strategies: a set of perception-action techniques that allow to obtain environment information, to plan motions for refining and completing this information (active perception), and to perform safe robot motions in non-structured scenarios. In recent years, these techniques have been greatly improved and have been applied in indoor environments with very good results. The goal of this project is to further develop these techniques to apply them to novel problems and more difficult scenarios, like rescue operations.
The research team will develop new Simultaneous Localization And Mapping (SLAM) techniques using partial information gathered from sensors such as vision, and multi-vehicle map building techniques for big environments. Also, dynamic environment modelling and object tracking techniques will be developed. To cope with non-linear problems and non-Gaussian distributions of error arising in complex and big scenarios, new Bayesian estimation methods for SLAM will be investigated. For computer vision, robust wide-baseline matching techniques will be developed. Furthermore, the team will develop new navigation techniques that take into account the robot and environment dynamics, and multi-robot navigation strategies.
The research project pursues two novel applications of the exploration strategies: (1) Search and rescue robots for confined or hazardous environments such as roadway tunnels after an accident or big outdoor and underground spaces like parkings; (2) An ARVA (Appareil de Recherche de Victimes d'Avalanches) system for rescue teams allowing fast and precise localization of multiple victims in snow avalanches, using radio-location and optimal estimation techniques.
