The Trajectory Planning Of Wheeled Mobile Robot Based On Moving Horizon Optimization

Trajectory planning and generation is one of the most key technologies for the control problem of Wheeled Mobile Robot (WMR), whose aim is generating the temporal motion sequences satisfying internal systemic constraints after path planning process, and ultimately finding the smooth motion trajectory which connect the initial state and terminal state. It's an important part of motion planning and also the foundation of trajectory tracking controller later.This paper introduces the main character of the Wheeled Mobile Robot, the nonholonomic constraint and differential flatness firstly; secondly, discuss the kinematics model and the dynamic model of WMR; after that, gives the one-time trajectory planning method in the based on the static environment based on the differential flatness. On the basis of this method, the further research of Wheeled Mobile Robot'real-time trajectory planning method based on the rolling optimization idea can be done. From the simulation's results we find that the trajectory not only satisfied the control inputs constraints but also solved the optimization problem like time and energy objective functions. Then the feasibility and validity of our method will be proved.The main research work in the thesis is as follows:1. Firstly, summarized the background and central control problems of Wheeled Mobile Robot, then the most important characters of Wheeled Mobile Robot which is nonholonomy and flatness are deeply analyzed, then showed how they impacting the WMR systems.2. Introduce the kinematics model of the LABMATE Wheeled Mobile Robot (WMR), and then use the Langrange principle to establish the dynamic model of it. Based on the model and the differential flatness character the one-time trajectory planning will be give. This method's advantage is that we can make use of the flatness output to transfer the plant function to the independency function. Then the problem of the trajectory planning will be transformed to the planning problem of the flatness output space and also satisfy the nonlinear constraints.3. For the WMR can't execute the instruction in sometimes, and want to optimize the performance of the system. For these problems, the flatness output is used to transfer the trajectory planning to a nonlinear planning problem. Then use the optimization toolbox to calculate the value. From the simulation's results we find that the trajectory not only satisfied the control inputs constraints but also solved the optimization problem like time and energy objective functions. So the feasibility and validity of our method will be proved.4. Based on the one-time trajectory planning method, the rolling optimization idea is introduced to solve real-time trajectory planning method in the static environment. The main idea is dividing the process into many periods, at every periods, the one-time trajectory planning method will be used. Because the obstacle is static, so the generating trajectory is as same as the trajectory which planned by the one-time trajectory planning method. From the simulation's result, the real-time trajectory planning's result is as same as the one-time trajectory planning's result. So the real-time trajectory planning method is proved correctly.5. When the Wheeled Mobile Robot is running in the real environment, the environment is variable all the time, so the research of the trajectory planning which focus on the variable environment is more significant. The real-time trajectory planning method in the variable is based on the rolling optimization idea which is obtaining the obstacle and the aim's coordinate at every sampling time, then using the information to complete the trajectory planning at this sampling time. Cyclically using this method, then the real-time trajectory planning can be achieved.Certainly, this paper also leave some problem that dose not solved. For example, achieve the dynamic trajectory planning more quickly, and adjust the parameter of the polynomial's rank for real-time.