Bayesian Incremental Inference Update by Re-using Calculations from\n Belief Space Planning: A New Paradigm

Inference and decision making under uncertainty are key processes in every\nautonomous system and numerous robotic problems. In recent years, the\nsimilarities between inference and decision making triggered much work, from\ndeveloping unified computational frameworks to pondering about the duality\nbetween the two. In spite of these efforts, inference and control, as well as\ninference and belief space planning (BSP) are still treated as two separate\nprocesses. In this paper we propose a paradigm shift, a novel approach which\ndeviates from conventional Bayesian inference and utilizes the similarities\nbetween inference and BSP. We make the key observation that inference can be\nefficiently updated using predictions made during the decision making stage,\neven in light of inconsistent data association between the two. We developed a\ntwo staged process that implements our novel approach and updates inference\nusing calculations from the precursory planning phase. Using autonomous\nnavigation in an unknown environment along with iSAM2 efficient methodologies\nas a test case, we benchmarked our novel approach against standard Bayesian\ninference, both with synthetic and real-world data (KITTI dataset). Results\nindicate that not only our approach improves running time by at least a factor\nof two while providing the same estimation accuracy, but it also alleviates the\ncomputational burden of state dimensionality and loop closures.\n