Detecting and solving the kidnapped robot problem using laser range finder and wifi signal

摘要

This paper presents an approach to detect and solve the kidnapped robot problem using range data from a laser range finder and wifi signals. Localization based on range finders has high accuracy, but fails to detect the kidnapped robot problem, a situation where a well-localized robot is moved to a random location without itself noticing about it. On the other hand, localization based on wifi signals has very high reliability and can be used to detect the occurrence of the kidnapped robot problem; but lacks accuracy. In our approach, a probability density function is constructed using particles sampled from the wifi signal models using kernel density estimation, then the likelihood of every laser range finder particle with respect to the constructed probability density function is calculated. The mobile robot reset its localization process if these probabilities are too low.