Integrated Particle Swarm Optimization and Kalman Filter for Navigation Applications

The interest in navigation applications and locations based services (LBS) applications has experienced a considerable growth in the last few years. These applications strongly depend on the user's location, and therefore positioning is a crucial input for these applications. For most of the current navigation applications, integrated INS/GPS systems have become a standard tool. In such applications, GPS mainly provides position and velocity while the INS provides orientation. In GPS/INS integrated systems, Kalman Filter (KF) is usually implemented to optimally fuse the input measurements from different sensors. However, several shortcomings of KF have been reported, specifically, for low cost inertial sensors. To overcome some of these challenges, different approaches of KF have been implemented for navigation purposes with different implementation complexity and accuracy levels such as the Linearized KF (LKF), Extended KF (EKF) and the sigma-point or Unscented KF (UKF). Furthermore, several Artificial Intelligent (AI) techniques have been used to adapt and optimize the KF parameters such as: Particle Swarm Optimization (PSO), training Artificial Neural Network (ANN), Genetic Algorithm (GA), and Fuzzy Logic System (FLS). In this case, more complexity and extra calculations time consuming will be added to the solution which in turn affects the memory usage and the time consumption of the implemented approach. For the KF approach, the system should be modeled based on proper knowledge on both dynamic process and measurement models. Moreover, the basic assumption that both the process and measurement have white noise with zero mean must be held during the process time. The divergence due to modeling errors is a critical problem in KF applications. A conventional Kalman filter could fails to ensure error convergence due to limited knowledge of the system's dynamic model and measurement noise. In many situations, the availability of a well known model is unrealistic where there are some model uncertainties which cannot be expressed by the linear state-space model. Regardless of the implemented approach, any INS/GPS navigation system will suffer from the major problem of frequent occurrence of GPS signal blockages. To overcome this problem, the INS is usually used for navigation as a stand-alone system until GPS signals are available again. Therefore, positioning during GPS signal outages, both partial and complete outages, using INS is a partial solution. In this paper, a new technique based on Particle Swarm Optimization (PSO) is implemented to improve the INS solution during GPS signal partial outages where the number of satellites is less than four. The proposed algorithm, PSO and KF, are tested using a test field data with simulated GPS partial signal outage. In general, results of the PSO techniques showed substantial accuracy improvement of the obtained position errors during GPS partial signal outage than of the results from the KF during the same period.