A Solution To Range-only SLAM Problem Based On Scan Matching

Robot has been widely developed and applied lately because it is able to be flexible, increase productivity and improve working conditions. As the robot has been developed to the third generation, modern intelligent robot, in order to further enhance its independence, it is expected to have higher demands on flexibility and intelligence, which is capable of moving around different environments automatically and accomplishing specific tasks. Simultaneous Localization and Mapping, or SLAM, is one of the key ingredients of autonomous vehicles.Firstly, this paper conducted deep research and analysis about SLAM. Based on widely looking up to a large number of literatures, research contents, status and directions were extensively reviewed. In this paper, scan matching-based method was adopted for SLAM problem, and the details of two common algorithms were covered.Secondly, Genetic Algorithm (GA) was used in scan alignment to localize robot's pose. The algorithm simulated the principle of natural selection in nature. First of all, the transformed parameters were encoded to be chromosomes by binary encoding in search space and generated the initial population; then computed the fitness value of every chromosome and gained the next population through selection operators, crossover operators and mutation operators; at last, if the new generation satisfied the optimal condition, the best chromosome with the maximum fitness value within the generation was selected and decoded to be an optimal solution.Thirdly, GI-SLAM approach was proposed to solve SLAM problem. In order to accurately achieve the matching information between two scans, two-step matching approach was employed. The first step applied GA algorithm to find out a gross result and the second utilized ICP method to correct this result. By extracting geometrical features, rules-based technique was employed to construct environmental map.Finally, a large amount of analysis and testing was made through simulating test and practical experiment, which showed that GI-SLAM was robust to the change of environmental parameters and real-time enough for on-line processing.