| Navigation in known or unknown environments is still a fundamental problem for the study of autonomous mobile robots. The mobile robots can complete all kinds of tasks under non-man control, by the inside sensors, such as odometer, or outside sensors, for example vision, infrared, laser, sonar, etc. Those tasks include sensing surrounding environments, building the map of environments, locating itself position, tracking object, avoiding static or dynamic obstacles. Navigation technology has been applied to many fields, such as aviation, military scout, medical treatment, or family tends and so on. They are able to finish some special tasks, which man cannot achieve.Supported by the NSF project,"Indoor Navigation Techniques of Mobile Robot based on Imprecise Map"and the national high-tech research and development plan of China,"Collaboration and Competition Mechanism for Distributed Multi-robots and its Application Techniques", this dissertation aims to improving the efficiency of the navigation and systemically studies several critical problems existing in autonomous navigation by single robots. The main research work is as follows:Firstly, the odometric cumulate error is still the most important aspect, which affects the accuracy and efficiency during autonomous navigation. Many researchers have proposed some efficient methods. But those approaches to odometric error modeling are designed for some special driving-type robot. And can not adapt to other driving-type robot. Therefore, a general approach to odometric error modeling for mobile robots is proposed according to both synchronous-drive roller robot and differential-drive roller robot. The method presents an assumption that the robot path is approximated to circular arcs. The approximate function relationships between the process input of odometry and non-systematic errors as well as systematic errors are derived based on the odometric error propagation law in detail, further are validated during navigation and autonomous localization for mobile robots. The experiments show that this method of odometric error modeling reduces cumulate errors and improves the accuracy of localization. Secondly, for the visual localization based on features, many candidate poses are often produced when the matching happens between features in image and three dimension landmarks in map database. Those candidate poses would lead to uncertain localization. So a global localization algorithm based on Particle Swarm Optimal (PSO) is put forward. Those candidate poses are optimized to acquire the optimal solution at the current time by using PSO algorithm. The experiment shows that the algorithm acquires higher localization accuracy at a cost of sacrificing little computation time.Thirdly, In order to solve the question on local minimum and deadlock during navigation for mobile robots, an algorithm of obstacle avoidance based on fuzzy behaviors fusion is advanced here. The current optimal path need not to be acquired, but only the most satisfying solution or most efficient path is obtained using this algorithm during autonomous navigation. Therefore, some deadlock phenomena are reduced, and the real-time is improved efficiently during autonomous navigation.Finally, for improving the real-time of video image processing during pose tracking, an adaptive Adaboost algorithm is advanced to pick up the interested areas for shrinking the field of image processing and reducing the disturbing of background noises. From the application to lane departure warning and lane keeping for this algorithm, the method improves the accuracy and real-time of image processing efficiently, and provides a new idea for vision image processing.
|
PDF Full Text Request