Making difficult decisions autonomously: the impact of integrated mapping and navigation

Abstract

The role of navigation is changing. The forces of increased autonomy, less prior knowledge, and larger missions are extending the navigation problem from the requirement of absolute localization to the larger question of context determination. Current technologies are inadequate in the face of such circumstances. The key to an evolved navigation technology begins with the ability to reason, in an integrated way, about the models used to determine vehicle context: physical models, dynamic models, sensor models, and behavior models. The integrated mapping and localization (IMAN) algorithm provides a hybrid estimation scheme to integrate decision-making about navigation events with navigation and mapping. An overview of IMAN is presented, along with an initial analysis of its performance. While IMAN is sensitive to the complexity of ambiguous situations, the algorithm demonstrates superior performance when complexity does not lead to failure. These results are used to examine the emerging set of technological needs for advanced navigation and mapping applications, including map representation, multiscale modeling, map fusion, and cross-model correlation.