| Recent years, robotic technology is applied widely in diversification areas, such as medicine science, agriculture and home service. To serve tasks better, the robot functions are improved, which leaves the robot structures more and more complex. Take the robot navigation as an example, once one link breaks down, the whole navigation system will be affected or even crashed. And the system cannot be rapidly updated with invalid compatibility. It is to say that the traditional centralized development strategy has restricted the application and development of robot technology. So it is urgent to search for more stabilized and efficient modular design scheme.In order to solve the mentioned problems, a mobile robot navigation system based on modular technology is proposed. First of all, to improve the accuracy of robotic localization, RBPF is presented to implement precise robot pose estimation. Then this proposed navigation system is divided and implemented according to minimum function. On this basis, to guarantee the efficiency and reasonability, a modular division method based on fuzzy tree-graph clustering is proposed. According to the problems caused by Naming server losing efficacy, a self-stability method based on entropy weight and improved flooding algorithm is utilized.The main research works are as follows:(1) RBPF is adopted to obtain precise pose estimationAccording to the problems of localization accuracy, RBPF is introduced to realize the robotic localization. Then the precise robot pose estimation can be obtained to improve the accuracy of topology node position. In this way, the robot navigation task can be accomplished while constructing the environment map.(2) Module implementation of navigation system based on RTMIn the RTM framework, the robot navigation system is implemented using OpenRTM platform. The navigation system is divided into several modules according to minimum function, and these modules are packaged. By connecting the packaged modules, the system flexibility and reusability is improved.(3) Module division method based on fuzzy tree-graph clusteringTo improve the effectiveness and reasonability, a module division method based on fuzzy tree-graph is proposed. The dependency between modules is analyzed by fuzzy tree-graph clustering and analytic hierarchy process. And different module division schemes are acquired by classifying the sample sets. The comentropy is adopted to evaluate the division schemes to gain the most reasonable division scheme.(4) Modular system self-stabilization implementationAccording to the problem of modular system self-stability, a naming server self-stability method is proposed. The entropy weight method is utilized to calculate comprehensive evaluation index. And the improved flooding method is introduced to elect the backup naming server with the optimal evaluation index to ensure the safe operation of modular system and improve the system fault-tolerant capability.Based on the above methods, the experiments are done using Pioneer3-DX mobile robot in indoor real environment. Experimental results verify the validity of the robot navigation system based on modular technology. The proposed method has a certain theoretical and practical value to the research of robot navigation and modular technology. |
PDF Full Text Request