Abstract
An attitude estimation system based on inertial measurements for a mobile robot is described. Five low-cost inertial sensors are used: two accelerometers and three gyros. The robot's attitude, represented by its roll and pitch angles, can be obtained using two different methods. The first method is based on accelerometric measurements of gravity. The second one proceeds by integration of the differential equation relating the robot's attitude and its instantaneous angular velocity which is measured by the gyrometers. The results of these two methods are fused, using an extended Kalman filter. Experimental results show that the resulting system is very sensitive and accurate.<>