Autonomous Mobile Robot Environment Exploration And Navigation In Dynamic Indoor Environment

Navigation and environment exploration are two key techniques for fully autonomous and self-contained mobile robots. Although a series of practical applications for mobile navigation are already applied in static environment, there is still not a unified and integrated architecture for navigation and exploration in dynamic environment. This thesis presents a recent research on the problem of mobile robot navigation and exploration in unknown environment with multi sensors. The research work includes real-time obstacle avoidance, localization and path planning, and corresponding solutions in the real robot platform are also given in this thesis.The real-time obstacle avoidance is the essential task for mobile robot safe working. Moreover, map building in unknown environment is even crucial for mobile robot autonomous navigation. A real-time and reliable obstacle avoidance method is designed for mobile robot indoor environment exploration by considering the character and capability of Pioneer 3-DX mobile robot. During the exploration, mobile robot records raw data from laser scanner and uses them to build the environment map in later offline state.Autonomous mobile robots are required to know their position and orientation precisely in order to perform their navigation and exploration tasks. Two different localization methods are investigated in this paper. One is vision-based pose tracking, which uses the environment information from vision sensor to correct the robot's estimated pose obtained by odometry. The other localization method is Monte Carlo Localization (MCL) based on laser scanning data. By using this method, the distribution of robot poses can be represented based on a set of samples and the belief of mobile robot can be calculated by statistical algorithm. In this thesis, laser data of 2D walls and other static objects are used to estimate the robot's pose effectively.Finally, a navigation method for mobile robots is described. Geometric environment map is converted to grid map and A~* algorithm is applied to global path planning. Based on this reference path, gradient algorithm and dynamic window method are used to implement mobile robot local path planning and dynamic obstacle avoidance respectively. Autonomous indoor navigation task is realized by Pioneer 3-DX mobile robot and experiment data analysis shows the method's validity and practicability.