Intelligent Car Based On ROS System

The field of automatic driving is a hot topic in the current research of artificial intelligence,which has an extremely broad prospect and market.Under laboratory conditions,it's very hard to explore and research the driverless technology.However,the technology of the smart positioning and guidance car is similar to the driverless technology.The core SLAM algorithm and the sensor technology can be applied directly to the study of intelligent car.In this paper,a systematic description of hardware realization is first introduced and,combined with the existing technology,a smart navigation car hardware platform is produced and tested.In addition,core SLAM algorithm for intelligent car is systematically introduces and analyzes,including the application of extended Kalman filter and particle filter on SLAM.Moreover,an improved FastSlam algorithm is proposed,which provides a new solution to the problem of particle degeneration.FastSlam algorithm uses particle to present trajectory of the car.The weight of a particle corresponds to the probability of a trajectory at different times.The value of the weight corresponds to the value of the trajectory probability.It should be noticed that the value of weight can't determine that of the next time at previous time.Therefore,a smooth mechanism is naturally introduced.Therefore,the particle weight will be no longer only determined by relevant data at the current moment,but also by the window of historical weight information.Finally,the entire smart car system has been constructed and tested by the current popular robot operating system ROS which has powerful development tools and mature navigation and positioning algorithms.In order to overcome the shortcomings and deficiencies of the navigation algorithm in the ROS system,the A * algorithm is used to replace the original Dijkstra algorithm.The function of navigation and self-positioning has been integrated to the system by taking advantage of ROS algorithms package.The smart car can use a simple global map to travel automatically to the specified location at laboratory environment.During the driving process,it is able to bypass dynamic obstacles.Besides,the smart car can generate a detailed map through the combination and modification of dynamic local map and original global map with the data collected by laser sensor.