Research On The Method Of Human-Robot Cooperative Spatial Cognition In Unstructured Environment

Human-robot collaboration is an effective mode to accomplish specific tasks in unstructured environment.Spatial information cognition and interaction is one of the basic elements of human-robot collaboration to accomplish tasks.Because of the inconsistency of human-robot spatial information cognition,in the traditional human-robot collaborative tasks,most of the human-robot collaborative modes use structured,high-frequency control instructions to interact,which result in low interaction efficiency and heavy human load.This study analyzed the differences of human-robot spatial cognitive characteristics,fitted the rules of spatial information interaction between people,constructed the spatial cognitive model of robots,and realized the consistency of human-robot spatial information cognitive architecture,thus improving the efficiency of human-robot interaction.Firstly,task requirements,scenario characteristics,human-robot characteristics and existing human-robot collaboration modes were analyzed.Combined with the decision-making method of fuzzy analytic hierarchy process(FAHP),a high-level human-robot cooperation mode matching task was proposed,and the overall framework of human-robot collaboration and interaction was constructed.Next,taking the lunar exploration mission as an example,a human-human cooperative interaction experiment was carried out,focusing on the cognitive and expressive characteristics of target features,spatial relations and spatial reference frames,and establishing a relevant spatial information expression corpus to realize the parsing of natural language and the knowledge base of spatial cognitive model of robots.Then,based on the perceptual and cognitive characteristics embodied in the process of human-human collaborative interaction and the ACT-R cognitive theory framework,a cognitive model of spatial information for robots was constructed,and the cognitive and reasoning abilities of spatial information such as target recognition and location,analysis and transformation of reference frame,calculation of spatial relationship,path planning and so on were realized.Finally,a prototype interactive system of human-robot collaboration was built.Furthermore,the target location and navigation tasks of lunar exploration mission were taken as examples to verify the effectiveness of robotic space cognition modeling and the spatial cognitive method of human-robot collaboration.