Design and calibration of a neural network-based adaptive knowledge system for multi-sensor personal navigation

Dorota A. Grejner-Brzezinska, Charles K. Toth, Shahram Moafipoor, Jay Hyoun Kwon

Research output: Contribution to journalConference articlepeer-review

4 Scopus citations


This paper presents the current design and the preliminary performance analyses of the multi-sensor personal navigator prototype, currently under development at The Ohio State University Satellite Positioning and Inertial Navigation (SPIN) Laboratory. The main purpose of this research project is to develop theoretical foundations and implementation algorithms, which integrate the Global Positioning System (GPS), Micro-electro-mechanical inertial measurement unit (MEMS IMU), digital barometer and compass to provide seamless position information facilitating navigation and tracking of the military and rescue ground personnel. The system model represents an open-ended architecture, which will be able to incorporate additional navigation and imaging sensor data in the future, extending the system operations to confined and indoor environments. In addition, the current system architecture is designed to incorporate a simplified dynamic model of human locomotion used for navigation in dead reckoning (DR) mode. The adaptive knowledge system, based on the Artificial Neural Networks (ANN), is designed to support this functionality. The system is trained during the GPS signal reception and is subsequently used to support navigation under GPS-denied conditions. The stride parameters, step frequency (SF) and step length (SL) are extracted from GPS data (SF) and GPS-timed impact switches (SF) during the system calibration period. SF is correlated with several data types, such as acceleration, acceleration variation, SF, terrain slope, etc., which are extracted from other non-GPS sensors and constitute the input parameters to ANN that predicts SL during the GPS signal blockage. The predicted SL, together with the heading information from the compass and gyro, support the DR navigation. The current target accuracy of the system is 3-5 m CEP (circular error probable). This paper focuses on the design architecture of the integrated system and the preliminary performance analysis, with a special emphasis on DR navigation supported by the human locomotion model.

Original languageEnglish
JournalInternational Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
Issue number5C55
StatePublished - 2007
Event5th International Symposium on Mobile Mapping Technology, MMT 2007 - Padua, Italy
Duration: 29 May 200731 May 2007


  • Dead-reckoning
  • Human locomotion
  • Multi-sensor integration
  • Personal navigation


Dive into the research topics of 'Design and calibration of a neural network-based adaptive knowledge system for multi-sensor personal navigation'. Together they form a unique fingerprint.

Cite this