Neural network aided GPS/IMU Integration for Unmanned Ground vehicle navigation

Autonomously operating unmanned Ground vehicles (UGV) have a great potential for many Applications. Decreasing the cost is the important part of UGV researching. As the base of vehicle control, the navigation system provides information such as position, velocity and attitude (PVA), and is also one of the most expensive systems of UGV. So a lot of works have been done to make low-cost navigation system for UGV. Inertial measurement unit (IMU) and the Global Positioning System (GPS) are the most widely used Navigating systems. GPS provides positioning information with consistent and acceptable accuracy, if there is direct line of sight to four or more satellites .It may also suffer from outages, jamming and multi-path effects because of environment reasons. INS, on the other hand, does not have the disadvantages of GPS, but its accuracy deteriorates in the long-term due to sensor error. Especially the error of MEMS INS increases rapidly due to huge errors of low cost MEMS sensor. Integration both INS and GPS provides performance as a reliable navigation system .The measurements from MEMS INS with GPS are mixed using data fusion technologies such as Kalman filter (KF) to generate the optimal PVA solution .One existing problem of such cost-effective integration system is the heavy reliance on GPS signal availability. The system PVA accuracy would degrade sharply without GPS update due to the poor performance of low-cost MEMS inertial sensors. An integrated system of GPS and Low cost MEMS INS is often used in UGV. To overcome the limitations. AI using a Neural Network (NN) has been proposed here to perform well in long GPS outages but require long training sequences.