ION GNSS 2012
Session B3: Robust Navigation in GNSS-Challenged Environments

Title: An Enhanced Prototype Personal Inertial Navigation System
Author(s): Y. Ma, W. Soehren, W. Hawkinson, J. Syrstad, Honeywell Aerospace
Date/Time: Thursday, September 20, 2012, 8:35 a.m.
Room: Grand Ballroom Center (Renaissance)

Numerous pedestrian navigation applications have been proposed, including localization for a coordinating group of firefighters, first responders, or soldiers. A challenging scenario arises in GPS-denied environments when the team operates inside a building, in the urban canyon, underground, in foliage, or under the forest canopy.
Related works in the displacement measure source for the GPS denied navigation are as follows. The so called "zero-velocity updates" (ZUPTs) to reset the IMU velocity errors during the stationary phase of motion. For example, the boot mounted IMU can achieve robust navigation solutions; however, the customers are resistant to boot mounted hardware. Other aiding approaches use imaging sensors, such as cameras and lidars in combination with an IMU. This visual odometry technique is used to measure the relative motion of the person with respect to the environment in order to reduce IMU induced drift. Additional displacement measurement sources include the Doppler velocimeter from a radar or ultrasonic sensor. However, both imaging and Doppler sensors require additional hardware, and the camera navigation accuracy is limited.

As an industry leader in navigation technologies, Honeywell has been researching and developing personal navigation equipment. For instance, the DRMT 4000 is a dead-reckoning system based on the fusion of IMU and compass information. This system is low-cost and capable of GPS-denied navigation in the absence of large magnetic disturbances. Moreover, Honeywell has been developing advanced techniques for aiding personal navigation by estimating displacements using gait models. This paper presents an enhanced Personal Inertial Navigation System (ePINS) solution to the problem of personnel location in GPS-denied environments. The technical approach is based on a strapdown navigation solution maintained using a mid grade IMU and wavelet based motion classification algorithms. This is a follow up of the DARPA SUO SAS and iPINS programs, and is currently being applied to the US Department of Homeland Security (DHS) Geo-spatial Location Accountability and Navigation System for Emergency Responders (GLANSER) and the DARPA Robust Surface Navigation (RSN) programs. The ePINS device can achieve closed path performance of <2% distance traveled for both indoor and outdoor environments.

The ePINS consists of a MEMS Inertial Measurement Unit (three gyroscopes and three accelerometers), a barometric pressure sensor, and a GPS receiver. The ePINS algorithms combine inertial navigation, barometric altitude, and wavelet motion classification methods to minimize the errors inherent in sensor-based systems, while achieving the system performance, size, and cost objectives. The motion classification uses a wavelet descriptor based on IMU data and is capable of identifying typical human motion modes, including walking, running, walking upstairs, walking downstairs, stopping, crawling, etc. The motion classification in a wavelet-domain feature space is described. This process is generic enough to allow new motion types to be added to the classification scheme easily, and robust enough to ensure that motions are correctly classified. The step-length model is also described, wherein the step length of each motion as a function of frequency and biometric information of the person (e.g., height) is characterized. After that, we use an extended Kalman filter update using the step length calculated from the trained model to correct the navigation solution.

A heading initialization process is also described, which involves walking a circular path in the presence of GPS. The initial heading performance can be within 0.6 degree.

The key advantages of the proposed algorithms are: (i) The wavelet based motion classification is very flexible, and it can be extensible to any number of other motions, and (ii) The initial heading determination is robust and accurate (iii) the step distance model produces an accurate along-track measurement.

The ePINS prototype has been evaluated both in simulations and indoor/outdoor experiments. In this paper, we provide a description of the navigation results obtained around the Honeywell Golden Valley, MN facility. The navigation results are relayed and displayed to a base station over a stand-alone data radio network, or can be transmitted to a mobile android device over a cellular/Wi-Fi network. Use of a mobile device allows the user to combine a radio and control/display function into an inexpensive, compact package.

Return to Session B3 Next Abstract