Personal Navigation System using Inertial Navigation Sensors and GPS
Mayank Kumar and Sankalp Dayal
B.Tech Thesis, Indian Institute of Technology, Department of Electrical Engineering
Submitted, May, 2010
Abstract
Personal Navigation is one of the fastest developing fields due to ever increasing demand for Location Based Services. The basic challenge is to locate a person/vehicle both in indoors (shopping malls, convention center) and outdoors. GPS has conventionally been integrated in mobile phones and car infotainment systems to provide navigation information. But, GPS based positioning system fails when there is no clear line of sight (LOS) visibility of satellites. Urban high rise building, tunnels, indoors and dense vegetation may restrict the visibility of navigation satellite and thus renders it non-functional. One way to solve this problem is by using MEMS inertial sensors that do not require any outside signals.to perform localization.
Inertial sensors have the drawback that they accumulate error over time. GPS cannot work in indoors or in urban areas and is prone to jamming.These two technique of positioning could be integrated to develop a reliable and accurate navigation system for pedestrian and vehicle.
Our thesis deals with the development and implementation of a personal navigation system using MEMS inertial Sensors. New methods and algorithm for MEMS sensor noise reduction and sensor calibration have been developed. Sensor data fusion using Kalman filters has been done for attitude and heading determination leading to an optimal positioning system. Further, human walking model have been extensively studied to model the varying stride length among individuals for increasing navigation accuracy. The developed algorithms are tested using an online module housing inertial sensors and a GPS systems. These algorithms are aptly suited to work in embedded environment (mobile-phones) and the work has been taken forward by industry partners for commercialization.
Links
1. Presentation Award (PPT, 1.8 MB) – We received :”Best B.Tech Project Award in EE, IITD” for this work.
Error Analysis and Stochastic Modeling of MEMS INS Sensors
Mayank Kumar
Research Associate, Center for Applied Research in Electronics (CARE, IITD)
Submitted: Nov, 2010
Abstract
This work presents the error analysis and stochastic modeling of commercial low cost MEMS Accelerometer, Gyroscope and Compass and how these error elements and their characteristics get manifested in various measurements like attitude, velocity and trajectory calculated. Although Micro Electro Mechanical Systems (MEMS) based sensors have been utilized for the development of low-cost integrated navigation systems on the benefits of low inherent cost, small size, low power consumption, and solid reliability, it is significantly important to characterize the error behaviors of MEMS-based sensors and to construct more sophisticated mathematical modeling methods for better accuracy in measurements. The errors of MEMS-based sensors like accelerometer, gyroscope and magnetic compass have been classified into deterministic and stochastic error sources and the stochastic error part is the focus of this work. The influence of stochastic variation of sensors will be assessed and modeled by two different methods, namely AR (Auto Regressive) and ARMA (Autoregressive Moving Average). Numerical results from static and kinematic testing have been used to assess the performance and limitations of different modeling schemes.
