A novel method for computation of importance weights in Monte Carlo localization on line segment-based maps

Monte Carlo localization is a powerful and popular approach in mobile robot localization. Line segment-based maps provide a compact and scalable representation of indoor environments for mobile robot navigation. But Monte Carlo localization has seldom been studied in the context of line segment-based maps. A key step of the approach-and one that can endow it with or rob it of the attributes of accuracy, robustness and efficiency-is the computation of the so called importance weight associated with each particle. In this paper, we propose a new method for the computation of importance weights on maps represented with line segments, and extensively study its performance in pose tracking. We also compare our method with three other methods reported in the literature and present the results and insights thus gathered. The comparative study, conducted using both simulated and real data, on maps built from real data available in the public domain clearly establish that the proposed method is more accurate, robust and efficient than the other methods. Monte Carlo localization has rarely been studied on line segment-based maps.Importance weights play a key role in the performance of Monte Carlo localization.A heuristic-driven method for weight computation on segment-based maps is proposed.The proposed method is compared with three other weight computation methods.Results corroborate that the proposed method is more accurate, robust and efficient.