Mult-robot Map Building Based On Swarm Intelligence In The Structure Environmen

Map building for mobile robot in an unknown environment is an important issue, which is a hot research field in recent years. It is the foundation of self-control for the mobile robot in unknown environment, the foundation of the mobile robot navigation, path planning, obstacle avoidance. Compared to the single robot, multi-robot map building has many advantages. Firstly, multi-robot exploration can more quickly complete the task. Secondly, multi-robot has better scalability and robustness. Finally, multi-robot is more reliable and suitable for a wide, complex and harsh environments. However, multi-robot map building is facing many challenges and difficulties because the robot needs interaction with the real environment.Firstly, multi-robot system is more complicated than the distributed system, because it requires an appropriate coordination mechanism which is formulated according to the environment and the capacity of the individual robot to complete task by the coordination between the robot. Secondly, the map information is stored in each robot. We need to merge map information in the robots into a complete map, that is to say information integration issues; Thirdly, most of the multi-robot map building methods require the global coordinates, which has high demand for hardware. Finally, the structure of our robots is simple. The communication and energy of robot is limited. So the algorithm must be easy. According to general knowledge, the process of searching target or travel for human beings, does not require accurate positioning. Human achieves the purpose of rough positioning by a particular marker in the environment. Getting inspiration from general knowledge, we use similar characteristics between the data collected by the robot to Universal Coordinate,ultimately achieving the purpose of the map integration. In the process, the robot does not depend on the global coordinate information, to complete the map Construction task. The main contributions of this paper include:1. The particle swarm optimization algorithm is applied to multi-robot self-deployment problem, to solve the problem how to make robots reach their proper place without global coordinates and central control. The characteristic of Swarm intelligence is simple structure, no central control. Therefore, the robot has not global positioning capabilities. Deployed the robot in the right position is very difficult. In this algorithm, the robot depends on the perception of distance between each other, to spread to the appropriate position by simulating gravity / repulsion.2. The paper implement the arithmetic of exploration in the local map and extraction algorithm based on line feature for the map. When the robot reaches the right position, the robot is going to explore the environment around by the explore methods based on the idea of the boundary. The information of environment will be stored in the form of the grid map, and extracted into feature maps based on line point by edge point tracking algorithm. Then, we get the local map.3. To take advantage of the characteristics that there are more line segment and angle in the office terrain, map merging approach based on line segment is proposed to complete the integration of map information. In this approach, the angle as the representative of map is used for the map matching. When the robot get local maps, the robot Universal Coordinate, relying on redundancy and similarity of the adjacent map information between the robot, and reduce the error, to achieve the prupose of the map integration. The approach can complete the task of map building without global positioning and distance detection between robots.