A Study On The Control And Realization For Bicycle Robot Nonlinear System

The bicycle robot that is proposed by academics in the field of robot in recent years is a new conception of intelligent transport (or traffic) tool. Because it is an under-actuated non-integrated system with lateral instability, it's two wheels are longitudinal and has non-sliding contact with the ground, meanwhile it's dynamics modeling possesses a kind of symmetrical feature, and its dynamic characteristics are complicated. Therefore, it is difficult to control the bicycle robot stable. In domestic the researches about bicycle dynamics modeling, control algorithm and system identification on the robot prototype platform are few, therefore the dynamic modeling and control should be the hot topic in robotics for long time.In this paper, bicycle robot dynamics modeling, control and system identification problems are taken as the research objects. A dynamic model of adjusting bicycle handlebar control is presented, which is based on Lagrange method, and the algorithms of sliding mode control are designed, meanwhile, By using the test data of tilt angle and handlebar angle, linear and nonlinear identification have been done based on the ARX model and adaptive network-based fuzzy inference systems(ANFIS). A real bicycle robot is designed and experiments on control of bicycle robot are performed. The main contents can be summarized as follows:1. A kind of steering control single-input single-output dynamic model for bicycle robot is presented, which is based on Lagrange method. It's nonlinear system is approximately linearized, and the designed controller of LQR is presented, which can effectively control bicycle robot when the tilt angle is in small value.2. The characteristics of SISO nonlinear dynamic model are analyzed, and the cascade sliding-mode controller, hierarchical sliding-mode controller and stable sliding -mode controller are applied to the bicycle robot nonlinear system, which enable bicycle robot more stable.3. Based on linear ARX model and non linear ANFIS model, The identifications of bicycle robot system are completed through the data of handlebar angle and those of inclination angle which are gathered when bicycle robot is stable. Simulation result by ANFIS based on T-S model could be very similar to the actual test data of nonlinear bicycle robot sysytem, and it's identification precision is higher than that of linear ARX model. The obtained conclusions of fuzzy inference between input and output by above identificaton methods can provide some reference value for effective control on bicycle robot system in future.4. The bicycle robot for experiment has been designed on the platform of hardware including ARM9 embedded system, microcontroller, high-precision three-dimensional MTI sensor and so on, which has finished the sampling of systematical data. A series of experiments have been done based on the prototype, taking the bicycle robot SISO nonlinear system as example, on which the corresponding control experiment is carried.From the non-linear and underactuated point of view, the research provides a new idea for bicycle robot control. The modeling by methods of system identification enriches our laboratory researches on bicycle robot, and lay the foundation for further study.