Control of Object in the Loop with Feedback Using Imperfect Sensors of Position and Acceleration

Abstract

The paper discusses the problem of balancing robot control with negative feedback loop. The regulator (controller) can be calculated by numerical optimization. Sensors of the controlled value are imperfect. Orientation angle sensor (gyroscope) has a limited speed, as well as binary quantization noise. An acceleration sensor (accelerometer) has the average value drift and Gaussian noise. None of these sensors is sufficient to effectively stabilize the balancing of the robot, but using them both in a single control loop can achieve the required accuracy of control in static and dynamic. Method of combining the two sensors to more accurately determine the value of a single, previously developed in theory, has been tested with simulation and with realization in acting model of the robot. The results are confirmed by their practical use with good effect, which one can see on video by the reference link.