An improved pedestrian inertial navigation system for indoor environments

Belt mounted IMU with enhanced distance estimation for pedestrian indoor positioning

“…It is worth noting that those two sensors do not provide measurement corresponding to the effective walking behavior, but contain on one hand additive white Gaussian noises imputable, for example, to the electronics and thermal effects, and on the other hand biases that affect the outputs. Those last are distinguished from noises because their impact is similar to that of an offset, or a constant, perturbing the measurement, as discussed in [9] or in the model from [7].…”

“…The MEMS-based inertial solutions still suffer from trajectory drifts due to the integration, or summation, of errors. In this case, biased and noisy measurements are mainly addressed with Kalman filtering in the state of the art, possibly assisted with zero velocity and angular updates [7]. Then, the IMU placement is forced by the validity of assumptions on velocity and angular values, which often leads to foot-mounted solutions.…”

Basic Techniques for Filtering Noise Out of Accelerometer Data

“…The most common filtration techniques for the filtering the noise out of sensor data are an Extended Kalman filter (EKF) [1], an Uscented Kalman filter (UKF) [2], an Inderict Kalman filter (IKF) [3], a Particle filter (PF) [4] and an interacting multiple filter [9]. On the other hand, these filters are difficult to implement.…”

Localization and activity detection Based on the fusion of environment And inertial sensors = Localización y detección de actividad Basadas en integración de sensores y dispositivos inerciales

“…The use of a different filter is suggested in (Lamy-Perbal et al, 2011). The chosen filter is an Indirect Kalman Filter (IKF) in which zero velocity updates and a proposed approach for angular updating are introduced to reduce the errors.…”

“…Perbal et al(Lamy-Perbal et al, 2011) proposes a solution based on fuzzy logic in which the membership function of the accelerometer signal is defined by m f a = 1 − min( of the accelerometer measurements and s a is the associated threshold. Following the same expression a membership function is defined for the gyroscope.…”