| Mobile robot is a high technology which synthesizes the computer technology, control theory, artificial intelligence, sensors and other disciplines. It is also an integrated system ensemble which colligated a few of techniques including environmental perception, dynamic decision and planning, behavior control and implementation etc. The navigation control in unknown environment is a hotspot and difficulty in mobile robot study, in order to secure autonomous navigation and complete the scheduled task, the robots should obtain the environment and their own state through the sensors and other technology, and then plan out a non-collision path timely to realize the obstacle avoidance and the navigation control in the obstructed environment.In unknown environment, it is ambiguous or inadequate for robot in terms of environmental information, the robot lacks or doesn't exist any prior information, including the environment size, the distribution of obstacles etc. the robot must establish a environment model through a sensor, for a appropriate environment model will help understand the environment and reduce the computational complexity of path planning and various decision making. The article presents an analysis of various environment models, and analyzes respective advantages and disadvantages. The article chooses the feature map as the environment map while carrying on Localization and Mapping, the selected environment map is consist of feature point.It is error-prone to get the robot status and environment information only by a single sensor such as odometer when carrying on robot Simultaneous Localization and Mapping. The odometer is used to achieve the flying track conjecture while the robot conducts Localization and Mapping, and the environmental geometric features are predicted according to the current known environment features and the results of the flying track conjecture. In the article, laser radar is used to sense the environmental information, and the sets of the environmental geometric features will be formed after processing the actual environment information, then the data association matching will be completed based on the predicted geometric features and the measured geometric features,the system updates the map or the robot status according to the results of data association matching of the features.On the autonomous mobile robot, to achieve the navigation control in unknown environment, the robot must be guaranteed to realize reliable localization firstly. Reliable robot localization is based on accurate environment map, and accurate environment map needs reliable robot localization, we call the problem Simultaneous Localization and Mapping (SLAM).The article conducts research on the robot SLAM in unknown static environment, and the method of realizing the mobile robot SLAM is described in detail. First, the article analyzes the application of the Kalman filter in solving the SLAM. As the Kalman filter applies only to the linear system model, the article presents and derives detailed process of a SLAM algorithm based on extended kalman filter.To verify the effectiveness of the Extended Kalman Filter (EKF) in realizing robot localization and environment modeling, the article established the Matlab simulation interface, and then conducted the simulation design with matlab7.0 to experimentize and confirm the location and tracking algorithm based on EKF. The experimental simulation can demonstrate the effect of environment map and obstacle updating as well as the moving direction updating and dynamic obstacle avoidance, the simulation results show that the SLAM algorithm based on EKF is rational and effective.Mobile robot path planning is based on the Simultaneous Localization and Mapping, furthermore, it should be an intrinsic robot behavior. The mobile robot path planning is divided into global and local path planning while the environment information is completely knew in global path planning and completely unknown or partly unknown in local path planning. The article has summarized the principles of the path planning, and elaborated the path planning based on the principles of artificial potential field. As the artificial potential field exist the problem of local minimum points, the article has conducted the thorough research on the problem, and put forward different improved measures in view of different situation.As an intelligent control system, the mobile robot needs to realize reasonable coordination between various modules, and the hardware design needs the property of openness and scalability, for a reasonable architecture design can achieve complex behavior easily and improve scalability. Considering the fact that the robot requires real-time performance and the software or hardware can be realized, the article has designed an embedded robot control system, the system architecture adopted the layering thought.The designed system takes the embedded Linux operation system and Samsung processor S3C2410 as the core, and extends the storage module, the power module, the JTAG debugging circuit, the sensor module, the wireless communication module, the servo control module, then constructs a robot software platform based on the constructed hardware platform, it mainly includes the VIVI starting software, the Linux kernel and the Cramfs root file system, the designed system can satisfy the requirements of real-time and multi task of the mobile robots well, and can simplify the application development. |
PDF Full Text Request