Design and application of real-time attitude estimation system for unstable object

The cardinal problem for balancing control of an unstable object is the real-time attitude estimation during the moving process, so this study emphasizes on walking balance control technology of an unstable object. In the paper the mathematic modeling of the parallel two-wheeled robot is described firstly. Based on the general research on multi-sensor data fusion, the data fusion concept of the complementary filter is introduced. A series of tests are made to adjust the complementary filter parameters and effectively modify the inclinometer´s lag and the gyroscope´s zero drift. The theory is applied to the two-parallel wheeled walking robot which can be finally controlled to realize the walking balancing, according to the attitude estimated from the complementary filter. The validity and rationality of the complementary filter designed is confirmed through the actual moving tests of the robot.