A new sensor fusion framework to deal with false detections for low-cost service robot localization

Abstract

The popularization of service robots requires that robots are not expensive and can work very well in human daily living environment. Due to lighting condition and/or cluttered background, there are false detections occasionally for landmark-based robot localization with low-cost color camera or infrared sensors. However, traditional frequently used methods, such as Extended Kalman Filter (EKF) and Particle Filter (PF), can't cope with the problem of false detection. A novel sensor fusion framework is proposed in this paper for robot localization, which is capable of dealing with the problem of false detection. It has been tested in a real receptionist robot equipped with an infrared camera and wheel encoders. Experiments illustrate that the proposed method has a better performance than EKF and PF when false detections occur, while maintaining almost same performance during other times.