Autonomous Navigation And Control Research Of Unmanned Vehicle

Unmanned vehicle is called intelligent vehicle controlled by PC or microcontroller. It is one kind of mobile robots actually. The unmanned vehicle attains the information of the road or the environment with the sensors onboard. It uses satellite positioning equipment or other position sensors for self-localization. And it can arrive the destination we set with desired motion. The unmanned vehicle is equipped with satellite positioning equipment, inertial navigation system, laser sensor, encoders, visual cameras and other onboard visual sensors for moving on the road. The unmanned vehicle has the advantages of computer science, pattern recognition and intelligent control technology. It will play an important role in indoor commuter, road transportation, military patrols and city road security. And the core of this study is obstacleavoidance, location and navigation, because the unmanned vehicle is set to arrive the destination safely and intelligently.In this latter, we focus on the motion control, localization, visual tracking and collision avoidance of unmanned vehicle. In order to make the unmanned vehicle move as desired, we study on the kinematic model and dynamics model, and design the controllers based on these two models. Then, the motion control task is finished. So, the localization technique becomes the first problem to be solved. Because of the disadvantage of single localization equipment, we introduce the particle filter to finish the fusion of DGPS data and encoders data, and we take the landmarks to estimate the location of unmanned vehicle. The visual system with two cameras is equipped on the unmanned vehicle. We design the visual tracking method to finish the target tracking based on leader-follower model. Considering the performance of safety and intelligence of unmanned vehicle, we introduce the method to spatial data clustering and timespace association to separate the moving obstacle from the background environment and log the history of moving obstacle. According to these data, we propose the probabilistic potential field to avoid the moving obstacles. And we introduce the method to detect the boundary of the road which can provide more information for navigation. At last, we take experiments to test the method and controllers that we propose.