Research On Key Techniques In Positioning And Path Planning Of Indoor Mobile Robots Based On UWB And Semantic Map

In recent years,service robots have been rapidly popularized in daily life,and indoor mobile robots account for a large part among them.Different from structured industrial scenarios,complex indoor environments have raised higher demands for positioning and path planning of the mobile robot.How to achieve accurate positioning in a more flexible way and get a comprehensive description of the indoor environment,are the key problems for indoor mobile robots.Among indoor positioning methods,UWB(Ultra-Wideband)can be better applied to indoor positioning of the mobile robots due to its higher positioning accuracy compared with other wireless positioning methods.Aiming at the problems in the application of indoor mobile robots,this paper adopts UWB based indoor positioning and ontology based environment modeling to carry out the research on the key techniques of indoor mobile robots,such as positioning,environment modeling and path planning.The main work and research results of this paper include:(1)In order to improve the efficiency of UWB positioning error map building,the factors affecting the accuracy of UWB positioning are studied,an adapted UWB positioning error map building method is proposed.Based on the modeling method of UWB positioning error,a particle filter based positioning method which adopts the positioning error information from the UWB positioning error map in particle initialization and weight update is optimized to improve the positioning accuracy and convergence speed of the algorithm.Referring to the experimental results of UWB positioning,the reliability of the proposed method and its applicability in dynamic indoor scenarios are analyzed.(2)The semantic modeling method of the indoor environment is explored in this dissertation.Based on path planning requirements in indoor environments,the physical information,semantic information and mobile robot path planning related knowledge are merged and described in a formal expression,the indoor mobile robot path planning task oriented ontology model of the semantic model is then built.The instantiation process of the ontology model is further discussed,together with the diverse sources of data and knowledge in ontology instances.(3)Based on the ontology model of the indoor semantic map and diversified semantic path planning task,the path planning task analysis module is studied with the SWRL rules built for path planning task reasoning.An improved Dijkstra path planning algorithm is then proposed under the framework of the semantic map.In order to build the node network for path searching,a node network extracting and simplifying method is researched.Based on the description of the interior structure in the ontology model,the edge weight allocation method of the path nodes is studied by considering the influences of the complexity of the indoor environment and the flow density.Combined with the output of the global path planning algorithm,a sub-task assignment method is designed to simplify the global path and facilitate the subsequent local collision avoidance.(4)As to the uncertainty in sensing and positioning of the mobile robots,a vertical ellipse based velocity obstacle(VEVO)algorithm is proposed to optimize the over-constraining problem in collision avoidance of the mobile robots.Dynamic window approach(DWA)is then combined with velocity obstacle method in this dissertation to integrate the dynamic and kinematic characteristics of the mobile robot in its collision avoidance.Since different robots may have different collision avoidance objectives,a multi-objective speed selection method is proposed which considers the trajectory length,time consumption and collision safety as three main objectives.Based on the collision safety requirements of different obstacles,a dynamic local path adjustment strategy is further studied to improve the efficiency of collision avoidance.The performance of the proposed collision avoidance algorithm is verified with the contrast experiments of different collision avoidance algorithms.Finally,the effectiveness of multi-objective speed selection method and dynamic local path adjustment method is verified by the experiments in which mobile robots are assigned with different collision avoidance objectives and diversified collision safety requirements.Based on above research contents,this paper studies the key techniques in positioning,environment modeling,path planning and local collision avoidance of indoor mobile robots.The effectiveness of the proposed methods is verified with real tests and simulations.In terms of UWB positioning,compared with the uniform UWB positioning error map building method,the method proposed in this dissertation can improve the positioning accuracy by nearly 20%in areas with larger positioning error,while reducing the number of positioning error measurement points by about 48%.As to global path planning,the proposed method can understand the semantic task well,and integrate the path planning related knowledge to support the decision-making in path finding.In collision avoidance,although the optimization effect on the performance of collision avoidance is not obvious in spacious environments,the proposed method can save about 12%of the time consumption of the collision avoidance in crowded environments,which improves the efficiency of collision avoidance in crowded environments.The research results of this paper have provided a good reference significance for the research of wireless positioning,path planning and collision avoidance of indoor mobile robots.