On Multi-Objective Path Planning Of Unconstrained Mobile

Ball and plate system is an exemplification of nonlinear dynamics system, it has been received more and more attentions. The ball on a plate can be considered to be an unconstrained object, it is able to move freely. Ball and plate system is a typical object for trajectory tracking or path-planning research of unconstrained mobile.In the path planning problem of ball and plate, the ball can be considered to be an non-sensor mobile robot. The environmental information and the ball's information can obtain by visual servoing system of the ball and plate system. Therefore, the path planning and mobile robot path planning problem has many similarities. On the other hand, because of the unconstrained mobile has no constraint of mechanical in the environmental. The movement of ball with uncertainty degree. So the research of path planning problem of unconstrained mobile has brought some new problems.This dissertation studies the path planning problem of unconstrained mobile based on the ball and plate system. With the further research on multi-objective path planning problem, A lot of research works on trajectory tracking or path planning by means of evolutionary algorithm, fuzzy algorithm, artificial neural network algorithm and so on. These algorithms are mainly used in robot, mobile robots, intelligent transportation, computer networks.The major research work of this dissertation is composed of the following three aspects.1. Algorithm of ball extraction in obstacle environment. The perception of the environment and the ball is an important problem to be solved aiming at the path planning problem. After summarizing the advantages and drawbacks of the former work, a new method of extracting ball information in obstacle environment is presented. The histogram in obstacle environment has the characters as follows:(1)According to the histogram, the light intensity changes as time goes by, and the peak of the histogram appears in different places.(2)No matter how the light intensity changes and how many the obstacles are, the histogram shows the trend of two peaks as long as the obstacles exit.(3) The gray value of the ball appears almost between the two peaks.This paper presents a dynamic double-threshold method to pick up the center coordinate of the ball. Combining with the dynamic child window method, the image of the ball is managed again to get a more accurate coordinate. The final objective is to extract the ball autonomously. Simulation experiments indicate that the method has the function of getting over the illumination changes and picking up the ball information in obstacle environment.2. Environment map establishing and multi-objective programming problem. We use the ameliorative grid method to establish the map for ball path planning environment, and present a mathematical description of multi-objective problem according to the map information.Map establishing is one of the most important approaches in path planning. The paper presents a grid method to establish the map only using the ball and obstacle information, which is fully thought of the stochastic obstacle position and obstacle figure. Position information and danger information is displayed in the grid. At last, a mathematical description is set up using the ball displacement, danger tolerance and time prognostication as objective and obligation condition.2. Multi-objective path planning algorithm and trajectory smoothing are implemented on the ball and plate system. First, this paper analyzes the characters of the non-restraint path planning mission. And then, a combining of entropy method and Dijkstra method is used for path planning. At last, B-spline interpolation is used to make the discrete path into continuous one.Entropy method is used to measure the impact of the information integrated program made by a certain performance indicator. And can generate a practical weight. A path planning contest can be held on the ball and plate system. The players can plan the pre-path according to the contest rule. The referee can judge the score comprehensively considering the path planning time, path tracking time, the number of collisions between the ball and obstacles. In order to get a higher score, people can be part of the path planning.Considering the limitations of the subjective weight and objective weight respectively, a combining weight method of the two is used. Entropy method is used to generate the objective weight. The subjective weight is generated by decision-makers. Search algorithm, together with the multi-objective solution, is used to generate moving trial. B-spline interpolation is used to make discrete path into continuous one. Compared simulation on BPVS-JLU II ball and plate system shows the feasibility of the method.In summary, multi-objective path planning problems of unconstrained mobile have been studied in this paper. The main purpose of the work is to study the feasibility of multi-objective path planning algorithm of unconstrained mobile. And the algorithm be realized on the ball and plate system, and validation and analysis the result through the experimental data and simulation.