ODIN (Ocellus DEM/3D aided Inertial Navigation) is an inertial navigation system aided by a multidirectional optical flow, sensor called Ocellus and a 3Dmodel of the environment. The Ocellus is constituted by an assembly of N elementary optical flow sensors. Each elementary sensor contains a mini-camera and a processor for computing, at high rate, the optical flow at the center of the focal plane (eg component used in optical mice). The 3D model of the environment to be used depends on the mission. For the medium-altitude (~ 200 m) flight of an aircraft (cruise missile) above a natural landscape, the environment is represented by a Digital Elevation Model (DEM) , which is a regular grid of elevations in geographic coordinates. For the very low altitude (~ 20 m) flight of an aircraft (mini-UAV) above an urban area, the environment is represented by a polyhedral model with a level of detail comparable to models produced by IGN from the Pleiades satellite tristéréo images. The ODIN navigation system allows the absolute location of an aircraft in the absence of GPS. ODIN is different from the usual odometry-based navigation systems, which do not use any knowledge of the environment. In these systems, the position error grows without limit because of the integration of the velocity errors. In the ODIN concept, the distortion of the optical flow pattern induced by a non-flat terrain allows for a quick non-ambiguous location of the aircraft. The feasibility of absolute navigation was demonstrated by IC3i in a study with MBDA, for a medium altitude flight (200 m) above a natural landscape modeled by a DEM at 1 arc second resolution. The proposed study aims to validate the ODIN concept using a real instrumented flight at medium altitude and to demonstrate, by simulation, the feasibility of a very low altitude absolute navigation over an urban area. It has two parts, and ODIN ODIN-MNE-3D. In the first part, corresponding to tasks 1 and 2, we propose to acquire experimental data by an instrumented flight (with a LIDAR system) above the Pyrenees, following the path that was used to validate the Scalp Naval. The Ocellus will be simulated by a wide-field, high speed video camera. The exact path will be restituted from a DGPS and high grade inertial components. The true terrain will be acquired by fusing LIDAR and interferometric SAR data. The track-mode filter will be implemented as an Extended Kalman filter (EKF). Filter performance will be communicated to MBDA and the DGA for comparison with the current performance of the Scalp terrain referenced naviagtion using a radar altimeter. In the second part, corresponding to tasks 3 and 4, we propose to work in pure simulation. We will achieve this using two 3D models of the same urban environment. The first high-definition model, will provide the simulation of optical flow measurements from several simulated trajectories. The second low resolution model will be used to predict the optical flow from the inertial kinematic state. The use of an EKF is inapplicable due to the non-differentiability of the surface representing the urban environment. The study will analyze various nonlinear filtering techniques, including particle filters, to estimate the errors of the inertial system (position, velocity and attitude) augmented with the gyroscopes and accelerometers biases from the comparison between the Ocellus simulated measurements and the predicted ones from the on-board model of the environment. The study will conclude with a design of the "optimal" Ocellus configuration to support both operational applications.

A good knowledge of the biomass location, its characteristics (quantity and quality) and its mobilization conditions (exploitability, service roads, mobilization costs) is essential to the development of the forest biomass industry. This knowledge is currently insufficient to provide at reasonable costs, the required guarantees on the wood supply and on its sustainability. The demand is however increasing due to a large number of new projects requiring increasingly large biomass volumes. Indeed, the only consistent data available today are those of the National Forest Inventory (NFI), which are nationwide statistics that do not lend themselves to mapping the resource at sub-regional levels such as a supply basin. Up to now, the use of remote sensing data did not meet the precision requirements to take into account stand structure, which is an essential parameter for the quantification and qualification of the forest resource. Recent developments in LiDAR technology (new sensors, GPS and inertial central), combined to other data sources already available (high resolution satellite images, aerial photographs), are now allowing a precise and fine forest description, in terms of both resource characterisation and mobilization conditions. Integrating this technology will provide an innovative response to the challenges of wood mobilization, including that of wood energy. The Foresee project aims to provide new tools for assessing the characteristics and dynamics of the forest resource biomass and the conditions of its mobilization at the supply basins level.