Trajectory Tracking Control System Of Spherical Robot Based On Vision Positioning

Spherical robots are mobile robots with spherical or sphere-like shells that move by internal drive and have promising applications in the fields of military reconnaissance,environmental monitoring and planetary exploration.Vision technology has become one of the main ways for mobile robots to obtain environmental information and achieve their own navigation and positioning.In this thesis,a vision-based spherical robot is designed,and its mechanical structure design,data-driven modeling and trajectory tracking control system design are carried out.First,a spherical robot based on external vision is designed.The vision device of the robot always stands on the top of the sphere,which can adjust its posture when the sphere robot moves and obtain the environment image stably.Three omnidirectional wheels distributed at 120 degrees form the omnidirectional wheel drive device,which drives the spherical robot to achieve omnidirectional motion by eccentric torque.The designed spherical robot is characterized by unobstructed visual signal,flexible motion and small turning radius.The mechanical structure,hardware and software design of the designed spherical robot are elaborated.Then,a modeling method of the visual spherical robot based on the improved weighted least squares support vector regression machine is proposed.The dynamics of the spherical robot is modeled and its motion performance is analyzed based on the Lagrange method.In order to solve the problem that it is difficult to build an accurate dynamics model for the spherical robot due to its complex characteristics such as underdrive and nonlinearity,the generalized Gaussian kernel function is combined with polynomial kernel function,and the parameters of the designed combined kernel function are optimized by particle swarm algorithm,thus a modeling method based on improved weighted least squares support vector regression machine is proposed.Simulation experiments demonstrate the feasibility and effectiveness of the proposed modeling method.Second,a GPC-PID-based spherical robot trajectory tracking cascade control system is designed and implemented.In order to reduce the problem that the spherical robot is susceptible to perturbation and improve the trajectory tracking control accuracy,this thesis selects the position of the spherical robot as the main variable,combines the weighted least squares support vector regression prediction model of the spherical robot with the Generalized Predictive Control(GPC)algorithm,and designs an improved GPC control The GPC-PID-based trajectory tracking string-level control system is designed to achieve stable and accurate trajectory tracking of the spherical robot.The simulation experiments prove that the control system has high control accuracy and strong immunity,stability and robustness.Finally,the experiments of self-balancing,linear trajectory tracking and circular trajectory tracking of the spherical robot are carried out based on the designed experimental platform.The experimental results show that the spherical robot system designed in this thesis can quickly return to the original equilibrium state after being disturbed and can follow the desired trajectory well.