Extended kalman filter sensor fusion and application to mobile robot

Abstract

The main problem in a mobile robot is the error accumulation in its position in continuous navigation. Localization of a mobile robot is done with gyroscope and odometer sensors using the dead-reckoning method. To estimate precise and correct position, the dead-reckoning system is aided by an external absolute positioning sensor. The continuous error accumulation in dead reckoning is reset, the position and direction state variables are updated with absolute sensor position data. The GPS (Global Positioning System) system, which is an absolute positioning system, is used with the dead reckoning system to estimate precise and correct position data. The extended Kalman filter is used for sensor fusion. The Kalman filter has the ability to make an optimal estimate of the state variable when the data is immersed in white noise. To implement the algorithm, a mobile robot kinematic model was obtained. The kinematic model of the robot is nonlinear in nature. Thus the model is linearized for use with the Kalman filter algorithm. Finally, the data obtain from the two different navigation system is perfectly fused and shown by computer simulation.