Improving ultrasonic-based seamless navigation for indoor mobile robots utilizing EKF and LS-SVM

The ultrasonic positioning system is able to provide centimeter-level location information. However, the signal of the system is easy to be disturbed and the outages of the positioning system appear. Inertial measuring units (IMUs) is a self-contained device and can provide long-term navigation information independently, but it has the drawback of error drift. In order to obtain accurate and continuous location information indoors for indoor mobile robots, this work proposed a seamless integrated navigation utilizing extended Kalman filter (EKF) and Least Squares Support Vector Machine (LS-SVM). In this mode, the EKF estimates the position and the velocity of the robot while the signals of ultrasonic positioning system are available. Meanwhile, the compensation model is trained by LS-SVM with corresponding filter states. Once the signals of ultrasonic positioning system are outages, the model is able to correct inertial navigation system (INS) solution as filter does. A prototype of the system has been worked in a real scenario. The results show that the performance of EKF is robust, and the prediction of LS-SVM is able to work as EKF does during the outages.