Resetting Method Using GNSS in LIDAR-Based Probabilistic Self-Localization

摘要

We propose an easy-to-use method that enables outdoor robots to use information from GNSS for solving kidnapped robot problems. When an outdoor robot uses a LIDAR for its localization, information from GNSS sometimes becomes a noise. The proposed method only uses GNSS information with a resetting method when trouble is detected on a LIDAR-based Monte Carlo Localization (MCL). We have investigated the character of this method with our outdoor mobile robot in an environment where the multipath interference frequently occurs. In this environment, the proposed method stabilized MCL under occurrences of multipath inference, and enabled it to recover from kidnappings until the absence of the multipath.