Research On Dynamic Control Method Of Unicycle Robot And Experiment

The unicycle robot is an inherently unstable system with the characteristics ofnonlinear and under-actuated, which is recognized as a typical example in controlfield. As for the problem of the unicycle robot’s posture control, the balanceproblem of pitch and roll direction can be solved, but in the case of balance of pitchand roll direction, the control of the yaw direction is still unable to achieve based onthe current research.In this paper, based on the previous research of the issue, we will discusswhether we can design a control method which is a simple calculation and thecontrol performance is good, so that we can make the unicycle robot motion on theplane freely.Firstly, after analyzing the construction of the unicycle robot’s dynamic modelcarefully, the author thinks it is important to make a supplement for the dynamicmodel. So the dynamic equation is modified by the Mac-Millan method. Secondly,with the fuzzy theory, design a T-S fuzzy controller on the MATLAB toolbox.During the process, the fuzzy rules and membership functions is establishedautomatically by using a neural network algorithm, the learning sample data iscollected by simulation. In addition, discuss the influence of the quantization andscale factors to the control performance. Thirdly, some dynamic simulations aboutthe stability of the controller, the speed control performance and the steeringperformance are made for the controller. The results show that with the fuzzycontroller, the posture control of the unicycle robot can be achieved. Besides acomparative simulation is designed with the gain scheduling controller, which isdesigned in previous research. The result shows that under the experimentalconditions, the performance of the fuzzy controller is comparative to theperformance of the gain scheduling controller, but the fuzzy controller is easier toimplement obviously. Finally, an experiment on the real unicycle robot is done onthe improved control platform, which verified the feasibility and effectiveness ofthe control method.