Objectives:The autonomous cognitive ability of unmanned platforms to complex environments is a key issue restricting their extensive real-world application,and has become a research hotspot in cogni-tive science,artificial intelligence,cartography and other fields.From the perspective of complementary ad-vantages of human-machine,the machine maps information processing model is proposed to achieve the logical expressions of machine maps information storage,processing,interaction and learning based on the conceptual model and cognitive characteristics of machine maps.Methods:An environmental representa-tion model including perception map,working map and long-term map is constructed,and the map struc-ture is analyzed from the perspective of observation angel,reference frame,information abstraction degree,data structure and description precision.An integrated information exchange pattern for measurement,pro-duction and application is established,including environment perception,mapping,work and decision-making,and meanwhile the continuous iterative environmental information processing process of this model is analyzed.A continuous autonomous learning model is also established,and the characteristics of the model in terms of learning process,learning content and persistence mechanism are analyzed.Results:Two experi-ments are carried out to verify the feasibility of the information processing model.The first experiment im-proves the long-distance autonomous navigation capability of the benchmark model by simulating the inte-grated sensing,mapping and decision-making capacities.The second experiment enhances the correlation between environmental factors and tasks by simulating the task driven process of creating working map,thereby improving the efficiency of task decision-making.Conclusions:The proposed model is able to pro-vide a theoretical basis for the establishment of machine maps technology structure and system architecture,which in turn guides the development of machine maps system application.
information processing modelmachine mapsintegrated sensing,mapping and decision-making capacitiescontinuous autonomous learning
万方数据
维普
