Placement interactif de capteurs mobiles dans des environnements tridimensionnels non convexes
|Authors:||De Rainville, François-Michel|
|Advisor:||Gagné, Christian; Laurendeau, Denis|
|Abstract:||This Thesis proposes a novel mobile sensor placement system working in initially unknown three dimensional environment. The mobile sensors are fix sensors placed on autonomous robots, which are ground and aerial vehicles equipped with computing units. The sensor placement is based on a user-defined view, named the virtual view. This view is manipulated through a virtual sensor intrinsic and extrinsic parameters, such as its position, orientation, field of view, resolution, etc. The virtual sensor is not subject to any physical constraint, for example it can be place where no sensor could be or it possess an arbitrary large field of view and resolution. The mobile (real) sensors have to acquire the entire information contained in this virtual view. It is only by combining the sensory capacity of an unknown number of sensors that they can acquire the necessary information. First, this Thesis addresses the sensor placement problem by defining a visibility function to qualify a group of sensor configurations in the environment. This function is applicable to three dimensional environments and is based on direct line of sight principle, where we compute the sensor sampling density in its visibility region. Then, this Thesis proposes the use of an incrementally built model of the environment containing all the information needed by the objective function. Next, a cooperative optimization algorithm is put forward to simultaneously find the number of sensors and their respective position required to capture all the information in the virtual view. Finally, the proposed system is experimentally shown to use less sensor to acquire the scene of interest at a higher resolution than state of the art methods in initially known two dimensional environments. It is also shown in simulation and practice that the performance of the system can be transposed to initially unknown non-convex three dimensional environments.|
|Document Type:||Thèse de doctorat|
|Open Access Date:||23 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.