Real Time Full States Integrated Low Cost Navigation System for Autonomous Vehicles

Abstract

Accurate measurements of angular velocities and linear accelerations are required to achieve a precise navigation solution for autonomous vehicles (AVs). These measurements are readily available from the inertial measurement unit (IMU) which is considered the most crucial component in the AV autopilot system. Inertial navigation system (INS) comprises of IMU plus complicated process that converts the IMU measurements to navigation information (position, velocity, attitude, and time (PVAT)). To use low grade IMUs for constructing a reliable INS, a precise mechanization model with an intensive aiding filter has to be implemented to integrate other sensors such as Global Positioning System (GPS) and magnetometers to insure trustable and continuous PVAT measurements. The motivation behind the work presented in this paper is to build a real time integrated navigation system using low-cost components available in the market. By using the proper calibration and error estimation techniques such as the extended Kalman filter (EKF), the system can achieve a comparable navigation accuracy with other higher performance navigation system. A linearized north-east-down (NED) error model is adopted, the GPS/INS integration using EKF is described. The algorithm is implemented on a low power ATSAM3X8E ARM Cortex-M3 series microcontrollers and integrated with an on the shelf MEMS 9-DOF IMU. The field experiments results analysis showed an outstanding real-time navigation performance if compared with high performance and much more expensive tactical grade INSs.