Monte Carlo Localization and registration to prior data for outdoor navigation

Abstract

Abstract-GPS has become the de facto standard for obtaining a global position estimate during outdoor autonomous navigation. However, GPS can become degraded due to occlusion or interference, to the detriment of autonomous performance. In addition, GPS positions must be aligned with prior data, a tedious and continual process. This work presents a solution to these two problems based on learning generic observation models in the presence of GPS to use in its absence. The models are non-parametric and compared to traditional approaches require few assumptions about either the prior data available or a robot's onboard sensors. Along with allowing for localization to prior data under GPS-denied conditions, this learning approach can be coupled with an EM procedure to automatically register GPS and prior data positions. Experimental results are presented based on data from more than 15 km of autonomous navigation through challenging outdoor terrain.

MCL with sensor fusion based on a weighting mechanism versus a particle generation approach

“…A common localization algorithm which integrates the information provided by different sensors is the Monte Carlo Localization (MCL) [7] method. It is based on particle filters (PF), whose samples (or particles) are weighted according to their likelihood computed from each available device [8]. Fusion of wheel odometry and GPS using MCL has been studied before [9], including omnidirectional vision [10], LIDAR Daniel Perea (corresponding author), Javier Hernández-Aceituno, Antonio Morell, Jonay Toledo, Alberto Hamilton and Leopoldo Acosta are with Departamento Ingeniería de Sistemas y Automática y Arquitectura y Tecnología de Computadores Universidad de La Laguna, 38203 La Laguna, Tenerife, Spain.…”

Observable 2D SLAM and Evidential Occupancy Grids

“…Currently, there are multiple ways of approaching the SLAM problem, which can be classified into one of the following categories-state estimation algorithms that encompass most variants of Kalman Filtering algorithms, probabilistic methods as in [16] and [17], and vision based SLAM that utilize neural network based solutions as in [18]. The classification of approaches are a result of focusing on different areas of the issue for improvement -speed, computational efficiency, accuracy and robustness.…”

“…Algorithms like Monte Carlo Localization (MCL) [16], [17], Particle Filters and Fast SLAM are techniques that are predominantly based on a family of probabilistic methods-Markov Localization and sampling particles representing a distribution for Particle Filter based approaches. The advantage with techniques like these lie in the fact that the distributions are not restricted to being solely Gaussian, or unimodal.…”