Local Path Planning For Mobile Robot Based On Binocular Vision

The most important character of mobile robot is that it programs a path to do something by detecting and dealing environmental information by itself. In other words, it has some intelligence and independent ability. The intelligent characteristic of the intelligent robot lies in it having the function that coordinate with the external world, such astarget, environment and people, each other. In this respect, the vision has great function , the vision of the robot is considered to be the important feeling ability .The path planning for intelligent mobile robot has already become the focus of studying in recent years. This paper studied the Local Path Planning for mobile robot based on binocular vision, the main work accomplished concretely follows:On the basis of binocular vision theory, put forward a kind of barrier detection method based on surface direction, and carried on the quantitative analysis of the surface direction and parallax change, discussed the concrete realizing and barrier pattern process algorithm of the barrier detection system .Studied a kind of search algorithm based on grid and used in Path Planning for the robot. Carried on the optimization of the algorithm mainly by proceed the data structure, and introduced the new concept of disconnected path without obstacle, the thought of amalgamating obstacle, and simulate the correctness , high-efficient quality, practibility and intelligent quality of this algorithm.On the basis of determining its dynamics model of the mobile robot GAIA-2, combine the real structure of GAIA-2, discussed those practical problems when this algorithm apply to GAIA-2 : the size of grids, the speed and changing direction question while advancing.Studied the structure and algorithm of BP nerve network, analyzed the local minimum question of BP algorithm, and proposed improved measure, used VC++6. 0 to design a simple BP nerve network. Useing the nerve network to path planning include following five function module : study sample, nerve network study , the export right value and valve value after training the nerve network, the path planning by the nerve network, exporting the result of path planning. At last verified its feasibility through simulation.