Study On Real-time Path Planning For Cleaning Robot Based On Monocular Vision Technique

At present, the sensor technology as one of the key technologies for mobile robot navigation, is an important means of cleaning robot to identify the surrounding environment. The robot can use the navigation methods including electromagnetic navigation, visual navigation, voice guided navigation, in which, visual navigation has become one of the main directions of development in the robot’s navigation, because of the wide range of signal detection, accession to information that is comprehensive.Based on existing studies of the visual navigation technology, this article proposes a method based on the monocular vision technology cleaning robot in real time path planning technology. Path planning is a key step in robot navigation, this article assumes that the cleaning robot has known the specific location of the goods (the global target set), and on this basis, it uses monocular vision to discovery obstacles in the indoor environment through obstacle classification and coordinate transformation, establishes raster maps, then plans the path using the ant colony algorithm for each new class. Finally, the robot can understand the global environment through local environmental information.Firstly, this paper summarizes the current development status of the cleaning robot at home and abroad, clears the background of this study, and points out the significance of the content of this paper.Secondly, it has studied a variety of robot path planning, summarizes the advantages and disadvantages of different methods; on this basis, the paper proposes its method of real-time local path planning for cleaning robots in indoor environments.Then, the technologies of monocular vision, support vector machine classification techniques and the ant colony algorithm is introduced. Using these methods and techniques, this paper studies real-time path planning of cleaning robots in the system simulation environment. First of all, use monocular vision technology to collect the image information of the local environment, through de-noising, edge detection and processing of local image information in order to identify the obstacles; Followed by, the laser range finder measures the distance between the robot and the obstacle, by calculation and coordinate conversion process to build a grid map of the local environment; takes a vector of several distances to characterize each obstacle, and uses the support vector machine classification algorithm to classify the obstacle environment; every time, a new class of obstructions environment generates, using the ant colony algorithm to plan the next step path for it; so cyclical; ultimately, real-time path planning of cleaning robots in dynamic environments is implemented.Finally, the thesis simulates cleaning robot and its indoor working environment in the MSRS robot simulation platform, uses the Microsoft Visual Studio2008C#to write the real-time path planning algorithm proposed, and verifies the effectiveness and feasibility of the proposed method through the simulated results and data analysis.