A method for mitigating the heading error growth inpersonal dead reckoning systems using inertialmeasurement units is described. From data collectedduring laboratory tests on human subjects to characterizethe biomechanics of normal walking, four kinematicparameters that identify (or are signatures of) thebeginning and ending of turning motion are identified--trunk roll, trunk yaw, foot yaw and stride width.Observers for estimating these parameters are described.These parameters are used as flags in the dead reckoningalgorithms to identify actual heading changes fromapparent heading changes caused by inertial sensor outputerrors. Therefore, during straight line walking (whenheading changes are equal to zero) the algorithmestimates inertial sensor errors. When heading changesare identified, the sensor bias estimation stops and theinertial sensors are integrated in an open loop fashion togenerate an estimate of heading. Experimental resultsshow that this approach can result in a heading errorreduction between 6% and 60% compared to simple openloop integration of inertial sensor outputs.
展开▼