Location and tracking of pedestrians based on inertial navigation

Abstract

The solution addressed here is based on inertial navigation: the output (linear acceleration and angular velocity) of a small-size inertial measurement unit, weared by the operator to be tracked, is processed to estimate the present location and the walked path. This information can be transmitted outside the intervention area, e.g. to the manager of the fire squad, to coordinate operators and possibly to start recovery actions of operators in difficulty. This paper presents the derivation of the matrix differential equations of the navigation and of the errors, followed by their discretization and estimation of the errors by a Complementary Extended Kalman Filter. Then the detection of the Zero Velocity Update is introduced to estimate the intervals when the foot is approximately still, at each step during walking, as a method to decrease the errors. Eventually, navigation error estimates are subtracted from the navigation estimate, followed by an experimental validation of the method.