The Research Of Navigation Information Compression Algorithm Of Outdoor Mobile Robot

In order to improve the efficiency of autonomous mobile robots, reasonablely reducing the sampling density of navigation information is an effective method, which has been a hot topic recently. The main content of this paper is the research of navigation information compression algorithm for mobile robot in public environment.Firstly, a method based on updating sampling period with the observation matrix is proposed, to achieve the single sensor information compression. The localization principium and source of errors of single sensor is analyzed. By targeting and pretreatmenting, the positioning accuracy can be improved and the system errors reduced. Observation matrix is established according to the robot's motion state and sensor information, and the original information is compressed using the real-time updating sampling period. Then the sensor information confidence can be calculated by D-S evidence theory, making it the evaluation criteria of information compression.Secondly, an algorithm based on multi-sensor information fusion which can perceive relevance of information space is proposed. The information provided by sensors and information space relevance can be dealt to fuse the navigation angle information and location information compressively. And further more, the pose information is also fused by the principle of minimum sampling period to make sure the correspondence and effectiveness of pose information itself. Information compression confidence is calculated in real-time through the variogram and DS evidence theory, to supervise the fidelity of location information.Thirdly, how to predict and update the sampling period with the map database information is studied. In a known environment, a global topological map (in which the multi-collected GPS global pose information is stored) can be established, which makes crossroads as node to set the trace route easily. Update the largest compression sampling period dynamiclly and reset the local positioning sensor information with the database node information to reduce information collection and improve positioning accuracy effectively, and calculate the information entropy of nodes to dynamically supervise the information uncertainty.In the path tracking control experiments which is based on compressed location information, the system can still effectively complete autonomous navigation task, which shows that the algorithm can reduce redundant pose computation through reducing sampling frequency, in premise of ensuring the validity of navigation information.