SLAM And Path Planning Of AGV Based On ROS System

With the increasingly large progress of electronic science and technology in the region of robotics and automation in China,followed by a great number of researches and application scenarios of moving robots,the innovated knowledge applied on moving robots used at home and abroad has been much more adopted in a wide range of fields.Real-time localization and map generation with path planner under unexplored circumstance with a certain amount of noise,on the other hand,have become hot spots for robotics exploration and development.This project focusing on the AGV application,carries out research on SLAM and path planning related technologies of AGV based on ROS system,including focusing on the realization of the whole system,from model material selection design to the software implementation of driving board circuit design,trolley control algorithm,SLAM algorithm and path planning algorithm,and improving the existing algorithms in two aspects of real-time positioning and mapping(SLAM)and path planning,and in an autonomous mobile robot system composed of a moving chassis,a 2D lidar sensor and a computer computing platform.Firstly,the hardware system and structure design of the mobile robot are completed.In regard of flexible and integrated lightweight control system design and portability among different computation architectures,an innovating ROSbased distributed control system is designed,which is combined with nodes of observation of environment and function of agile real-time information exchanging sensor in terms of precise position control during the movement and autonomous map building and navigation in the specific working operation of mobile robots.This part of the article also includes the work of functional optimization in every aspect of the above.Apart from that,couple novel methods that utilizing the sampling strategy of local sampling-based Unscented Kalman Filter(UKF)with cumulative error and particle filter with state error for comparison with the consideration of large error estimation algorithm in SLAM technologies has been made.Considering the concept of the mathematical characteristics of the current state motion vector,the partial current state motion vector localization and accumulation calculation and tracking processing in the UKF algorithm in the UT sampling process,and partial sampling of the resulting data,through the Matlab simulation and data analysis in terms of simplicity of computation,it has been proved that the mobile robot can completely ensure the position accuracy effectively with reduction of errors and cumbersome calculation under moving circumstance.To solve the problem that the traditional A* algorithm is easy to fall into local optimum and has many twists in path planning,a hybrid algorithm combining improved A*(A-Star)and dynammic window approach,DWA)is proposed.The mobile decision-making and path-planner system equipped with laser camera as the main sensor is crafted to obtained the two dimensional map info of the environment.Aiming at the problems of synchronous positioning and mapping,the Lazy Decision algorithm is improved to improve the loop quality and reduce the computation.In view of the problems existing in the path planning of robots,the prediction distance is introduced into A* algorithm,and the h(n)weight coefficient in the dynamic measurement heuristic A* algorithm is set,which solves the problem that using A* algorithm expands many nodes and easily falls into local optimum.Meanwhile,the corner in path planning is corrected,which effectively reduces the path planning time by 14%.Finally,the simulation experiments on the functions of real-time localization,map generation and path planning with navigation and guidance for the robot,using the built experimental platform for verifying the feasibility of using the UKF localization algorithm based on local sampling,the improved A* algorithm and the improved DWA algorithm are conducted.