HNG: a framework for robot control using efficient data fusion
R. Jaulmes (Direction Générale de l'Armement, DET/CEP/EORD) and E. Moliné (Direction Générale de l'Armement, DET/CEP/EORD)
Filtering and data fusion methods such as Kalman or particle filtering are very important components of intelligent autonomous vehicles. With these tools, it becomes possible to build in real-time precise representations (maps) of the environment and use them to improve localization. However, each of these methods use sensors that sometimes fail, and the dynamic models they use may not be adapted to real-world conditions. The method we present, which is based on our Hybrid Network-based Generic framework, uses simultaneously different filtering methods and switches from one to another to improve the overall fault-tolerance of the system.