This paper presents the localization problem of outdoor vehicles using Interacting Multiple Model (IMM) and Extended Kalman Filter (EKF), in their predictive step without exteroceptive sensors data. Usually, hybridization operates between exteroceptive sensors (e.g. GNSS1) and proprioceptive sensors (e.g. Odometer, Inertial Measurement Unit etc.) through a merging algorithm. Common experiments use the GPS receiver PPS time for stamping the odometric, gyrometric and IMU measurements, after what all these sensors are in the same UTC reference time. Now it is well known that the low cost GNSS devices have a very low frequency compared to proprioceptive sensors, combined to a low accuracy. Therefore in order to assess the vehicle positioning at higher frequency for safety applications, the sensors measurements are generally synchronized before being exploited in the merging algorithm. In our approach, the sensors remain in their original frequencies. The objective is to design a reliable and robust system that exploits asynchronous data. In order to reach this goal it is important to guarantee accuracy and integrity of filters even during the predictive steps, when exteroceptive GNSS data are not available: that is proprioceptive-sensors based positioning. We introduce in this paper, a study on the influence of the road bank angle assessment on the output. This parameter is used to correct the gyrometric and inertial unit measurements leading to an improvement of both IMM and EKF predictive output positioning. Tests performed with real data proved the suitability of introducing this parameter in the system.
Improvement of the Proprioceptive-Sensors based EKF and IMM Localization
2008-10-01
296864 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
External force estimation for teleoperation based on proprioceptive sensors
BASE | 2014
|Self-Supervised Learning of Terrain Traversability From Proprioceptive Sensors
Online Contents | 2009
Localizing external contact using proprioceptive sensors : the contact particle filter
BASE | 2018
|