Study On Distributed Navigation Method For Robot Based On Wireless Sensor Network

Wireless sensor network (WSN) is a novel technique for monitoring physical world, the characteristics of ubiquitous could make people obtain all information in sensing region. Nevertheless, traditional WSN is still a passive monitoring system, and the robot has the function of environment interaction, it can constitute a smart response control and measurement system for environment that combines both of them. Our research is supported by Natural Science Foundation of Heilongjiang Province of China. The aim is to research the robot auto navigation method basd on WSN under unknown environment. At present, many scholars begin to study the robot autonomous navigation technology based on WSN, but most of them has to obtain the spatial coordinates of mobile robot through network or other equipments, then using the traditional method to realize the autonomous navigation for robot, and there are so many problems on location accuracy and convenience. In this dissertation, the mobile robot state and navigation space are denoted by the RSSI potential field, and it can efficiently avoid the model error while RSSI is transformed into distance. The navigation system is consists of some beacon nodes, and each of them is a distributed measurement and control unite. The main contributions of this dissertation are as follows:(1) A robot navigation space model is created relative to the beacon node, the mobile robot state and navigation space are denoted by the RSSI potential field, and the RSSI value is regarded as the controll variable for the navigation system. It efficiently avoids the influence of channel transmission model error in navigation;(2) A path planning searching algorithm of multi-Object Nodes based on the matching coefficient is designed in the situation of WSN nodes large scale deployment. The sub-navigation object nodes are obtained by the subtractive clustering algorithm. The navigation target is divided into several local object node areas, and it makes sure that robot reach the destination with weak constraint conditions.(3) In order to avoid model error, A RSSI signal pre-processing algorithm based on improved MH particle filtering is proposed, the calculation of particles iteratively is distributed to each beacon nodes and parallel execution. The resampling step is no longer necessary, so it saves the storage space for process data and effectually improves the algorithm characteristics of real-time. RSSI value and distance have the monotone function relationship, then it is easy for single node navigation decision. (4) The one-step prediction of RSSI signal on statistical significance is obtained through analysis the relevance among beacon nodes with multiple regression method, and then the channel transmission model and the sports law are used to set up the navigation sport model in coordinate space which is denoted by the RSSI potential field.(5) The final control output of robot is calculated based on multi-objective decision techniques. Navigation system consists of some beacon nodes, and each of them has its independent navigation target and can make the decision in favor of itself. Decision center seek the sub-optimal navigation strategy by multi-objective optimization theory to satisfy the demand of decision-making nodes to the greatest extent. Experiment shows that navigation accuracy is within 1m.