Research On Mobile Robot Interior Corridor Recognition And Topological Graph Construction

In recent years, with the expansion of mobile robot applications, more and more mobile robots are joined in different areas for human services to replace human of complex and dangerous tasks. Mobile robot environment is often unknown, unpredictable. The existing level of intelligence of robot can not fully meet the requirements of the task in complex environment, and therefore improving the intelligence level of robots to improve their autonomy and adaptability is the key to intelligent mobile robot development.This paper makes a brief introduction to the research background of intelligent mobile robot, reviews and analyzes the current situation of the corridor types recognition for robot in indoor environment and topological map building research, pointed out the existing problems related to, the main work is as follows:1. A research on developments of the mobile robot interior corridor recognition and map building is made, and review the robot interior corridor recognition, environment space representation and the research method of topological map building aspects with technologies related to. By giving a brief introduction and comparison of various research methods, we point out that the characteristics and problems existing.2. For the indoor environment corridor type recognition problem, an improved recognition algorithm corridor is proposed. The algorithm has corridor classified and encoded into eight types, de-noising the environment depth data which is extracted from RGB-D depth sensor, and use depth difference between consecutive points and the linear least squares fit to detect existence of export corridor. Finally, complete the determination of corridor category by using the Bayesian classifier.3. Investigates interior corridor environmental topological map building problems. According to the characteristics of the environment of interior corridor, the recognized corridor type is treated as a node in the processes of building a topological map, and the corridor aisle as a branch for connecting nodes. Thus create a suitable indoor linear topological map. Compared to the Voronoi diagram, the topological map structure is very clear and concise, not easy to produce redundant node and path information.4. Investigates multi-robot built topological map merging problem, proposes a matching algorithm of using degree of nodes and vectors to improve the efficiency of matching vertices. Treating the topological map merging problem as image matching and image registration problems, we use the structure of integration of sub-graph matching and geometric aspects of image registration algorithm etc. for matching process, with matching node in traversal way, we merges these maps by finding the greatest common sub-region between the two maps, and finally select the optimal matching scheme to complete map merging.Finally, a summary of the research work is made, the areas which are needed to be improved have been analyzed, and an outlook of future is given.