Sigma-Point Kalman Filter Based Sensor Integration Estimation and System Identification for Enhanced UAV Situational Awareness and Control
Abstract
The goal of this contract was to develop and demonstrate a machine learning framework for probabilistic vehicle state and model parameter inference, aiding the sensor integration and processing for the autonomous control of UAVs. The core technology that this approach is based on is the Sigma-Point Kalman Filter (SPKF). The current industry standard and most widely used algorithm for estimation is the extended Kalman filter (EKF). The EKF combines the sensor measurements with predictions coming from a model of vehicle motion (either dynamic or kinematic), in order to generate an estimate of the current navigational state (position, velocity, and attitude). This study points out the inherent shortcomings in using the EKF and presents, as an alternative, a family of improved derivativeless nonlinear Kalman filters called sigma-point Kalman filters (SPKF). We demonstrated the improved state estimation performance of the SPKF by applying it to the problem of loosely coupled GPS/INS integration. A novel method to account for latency in the GPS updates was also developed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 25, 2004
- Accession Number
- ADA428975
Entities
People
- Eric Wan