Research And Realization Of Environmental Borderline Learning Method For Robots

Both robot and autonomous mobile vehicle expect to understand the workplace in advance, by reason that the more exactly they know the relative information, such as structural features of the environment, geometric conditions, spatial position, border-scale, surface property, etc., the more quickly and efficiently they can plan, monitor and guide each modules of the system and complete the mission. For the purpose of building up the precise environmental model simply, the paper set up environmental borderline tracing method and its learning strategy based on photo electronic code and detection signal of photo electronic switch. This paper discusses the tangential path tracking, curve segment tracing, flexion tracing during the process of extracting the features of environmental borderline as well as the solving strategies of key technique issues, also the helper role of common sense and experience.By analyzing the merits and limitations of various kinds of robot environmental tracing strategy, this thesis clarifies the practical utility of environmental borderline tracking method of which based on photo electronic code and detection signal of photo electronic switch, in-depth amplifies the problem faced by the borderline tracing and its solutions. And finally find out the technological line and practical strategies of which in line with the objective reality after doing a further careful analysis of the whole design concept of the experimental system , the framework structure as well as the functional division of each function modules.First of all, the system extract different environmental features and figure out the type of the environment by the aid of photoelectric switches and photoelectric encoder, which can define the carrier's position and its distance from the borderline, in further acquire the geometric conditions, spatial location and boundary-scale, then modify the environment model.In order to obtain more accurate environmental information, it uses a speedy parameter extraction algorithm during the system realization. It also designs different kinds of borderline tracing algorithm for different borderline, introduces standard deviation while modifying the environmental model, effectively eliminates the error in the course of environmental borderline tracing learning . In addition, it researches the fusion problem of more sensors. Finally, the thesis presents the experimental operating system and the test results, points out the limitations and inadequacies of the current condition thus set the future plan and orientations of efforts in conjunction with scientific development tendency.