LOW COST 3D LOCALIZATION SYSTEM FOR APPLICATIONS ON AERIAL ROBOTS

Abstract

This work describes the design of a low cost 3D localization system based on strapdown IMU (Inertial Measurement Unit), GPS (Global Positioning System) and 3-axis magnetometer. Although part of an aerial robotics project, the applications of such systems range from virtual reality systems to sports training. The prototype has been built within the university facilities with low cost devices, whose performances in standalone operation for the task proposed is prohibitive. Hence, the sensors have been selected to compose an integrated navigation system, in which the IMU, built with micromachined inertial sensors, provides high sampling rate, but unbounded errors in long term operation; the GPS receiver provides position and velocity estimates without drift in long term operation; and the 3-axis magnetometer supplies information to correct attitude estimate. To perform the sensor fusion algorithm, a Kalman Filter, a 32-bit ARM microcontroller was selected, so embedded design issues are likewise discussed in the paper. Also, the initial alignment procedure and the filter equations using quaternions are described. Experimental results of this approach, which demonstrate the system performance, are presented, as well as a discussion about future improvements and different filter configurations.