The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLIII-B1-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIII-B1-2022, 137–142, 2022
https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-137-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIII-B1-2022, 137–142, 2022
https://doi.org/10.5194/isprs-archives-XLIII-B1-2022-137-2022
 
30 May 2022
30 May 2022

RADAR/INS INTEGRATION FOR POSE ESTIMATION IN GNSS-DENIED ENVIRONMENTS

M. Elkholy1,2, M. Elsheikh1, and N. El-Sheimy1 M. Elkholy et al.
  • 1Department of Geomatics Engineering, University of Calgary, Calgary, Alberta, Canada
  • 2Department of Transportation Engineering, Faculty of Engineering, Alexandria University, Alexandria, Egypt

Keywords: FMCW Radar, Inertial Navigation System, GNSS Outage, Radar/INS Integration, Ego Motion Estimation, Micro Electrical Mechanical System

Abstract. This paper proposes a novel algorithm to use Radar in ego-motion estimation for autonomous navigation applications. This method is based on the analysis of Radar data to remove noise, ghost points, and outliers and keep the accurate features. From the detected features and the knowledge of Radar data rate and the vehicle's average speed, the change in range and azimuth between any two points can be constrained to find the corresponding points. With the help of the corresponding points, the vehicle's ego-motion can be estimated. Then, Radar is integrated with an Inertial Navigation System (INS) and odometer through an extended Kalman filter (EKF) to smooth the Radar solution and aid INS to overcome its large drifts in GNSS denied environments. Two real data were collected from frequency modulated continuous wave (FMCW) Radar sensors and Inertial Measurement Unit (IMU) in suburban areas near the University of Calgary, Canada. The proposed algorithm was tested by introducing simulated GNSS signal outages with different durations. The Root Mean Square Error (RMSE) for the horizontal position was improved by an average of 30.44% and 4.76% if it was compared with RMSE from odometer/INS solution with a percentage error less than 1% of the traveled distance which was 1.59 km and 2 km for the two datasets, respectively.