An algorithm to detect zero-velocity in automobiles using accelerometer signals

Abstract

The vehicle is often operated in environments where GPS signal is lost. In such cases, the navigation has to be done alone by an inertial navigation system (INS), called free INS navigation. The positioning errors of INS grow exponentially with time, thus the duration of the free navigation should be monitored in order to give warning of possible loss of positioning accuracy, and if necessary, supported by the static INS calibration in Zero Velocity Update Mode (ZUPT). One of the main components of INS is the inertial sensor whose error is unfortunately, increased unboundedly with time. One way to bound this error is to use the absolute zero velocity to calibrate the inertial sensor from time to time. This requires to detect the zero velocity condition accurately and instantly. This thesis presents a novel way to detect the zero velocity condition using a method in pattern recognition. One feature is given and compared with other possible features. A simulation test has been done and performance has been evaluated based on the collected samples. /Kir10