Intelligent Tuning of a Kalman Filter for INS/GPS Navigation Applications

Chris Goodall, Xiaoji Niu, and Naser El-Sheimy

Abstract:	The demand for low-cost portable civil navigation systems has been growing over the last several years. The Global Positioning System has been the backbone of most current navigation systems, but its usefulness in harsh environments, such as downtown urban areas or under heavily treed terrain is limited due to signal blockages. To help bridge these signal gaps inertial navigation systems have been commonly used. An integrated INS/GPS system can provide a continuous navigation solution regardless of the environment. For civil applications the use of MEMS sensors are needed due to cost, size and regulatory restrictions of higher grade inertial units. The Kalman Filter has traditionally been used to optimally weight the GPS and INS measurements, but when using MEMS grade sensors the a priori tuning parameters given by the designer or manufacturer are not always optimal. In these cases, the position errors during loss of the GPS signals accumulate faster than the ideally tuned case. To aid in the on-line tuning process, a reinforcement learning algorithm was used to tune the Kalman filter parameters as navigation data was collected. Tuning any Kalman filter is a difficult task and is often done before navigation with the aid of the filter designer. This process often entails much iteration using human expertise and is in no way guaranteed to result in optimal parameters. Reinforcement learning is an intelligent and adaptable solution to this problem which uses a combination of dynamic programming and trial and error exploration to develop a set of parameters that improve as navigation data is collected by the user. In comparison to a manual tuning approach it was found that using reinforcement learning led to similar estimates of the tuning parameter values for a single integrated system. This is encouraging since the reinforcement learning was done online without any need of intervention by the designer. Furthermore, when the results of tuning one sensor were applied to a second integrated sensor, the tuning converged almost immediately and improved the positioning accuracy of the system by several meters in comparison to using the values generated from tuning the first sensor.
Published in:	Proceedings of the 20th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2007) September 25 - 28, 2007 Fort Worth Convention Center Fort Worth, TX
Pages:	2121 - 2128
Cite this article:	Goodall, Chris, Niu, Xiaoji, El-Sheimy, Naser, "Intelligent Tuning of a Kalman Filter for INS/GPS Navigation Applications," Proceedings of the 20th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2007), Fort Worth, TX, September 2007, pp. 2121-2128.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In