Research On Mobile Robot Multi-task Autonomous Navigation System

With the rapid development of artificial intelligence and robotics,mobile robots play an irreplaceable role in the fields of rescue,inspection and underwater.Compared with the single task navigation that is difficult to be better applied to the complex and practical environment,the multi-task autonomous navigation is closer to reality and it has become a research hot spot in robotics in recent years.Thus research on mobile robot multi-task autonomous navigation is of great theoretical significance and application value.Firstly,on the basis of a systematical study of the general navigation method and sensors of mobile robots,the laser range finder is selected as the main sensing device in the navigation system.Simultaneous localization and mapping(SLAM)method is used to construct the environment map on the robot operating system(ROS)platform with good versatility.The design and implementation of the multi-task autonomous navigation scheme for mobile robots on ROS are completed.Then,for the problems of a large number of required particles and particle impoverishment in the Rao-Blackwellized particle filter-based SLAM(RBPF-SLAM)algorithm,an improved RBPF-SLAM(IRBPF-SLAM)algorithm is proposed.The mixed proposed distribution is optimized with the annealing parameter and incorporated the observation information of the laser range finder.Meanwhile,an adaptive partial rank-based resampling(APRR)technique is designed to mitigate particle degradation.Experimental results show that the operational efficiency of the proposed algorithm is higher than that of RBPF-SLAM algorithm.However,there are still problems for the low estimation accuracy and the poor robustness under the noise.In order to further realize the accurate and reliable navigation,an improved Rao-Blackwellized H? filter-based SLAM(IRBHF-SLAM)is proposed.The iterative unscented H? filter(IUHF)is utilized to accurately calculate the importance density function and therefore the system state mean and covariance are estimated.Through the iterative updating method,the observation information is utilized to constantly correct the system state mean and covariance to further reduce the estimate error.Experimental results prove that the improved algorithm is feasible and robust.Secondly,the multi-task path planning technology is analyzed and studied,thus a mathematical model of traveling salesman problem(TSP)for solving the multi-task path planning is put forward.The improved artificial fish swarm algorithm(IAFSA)is used to plan the path of any two tasks,thus the path consumption is solved,which is used as the input of the TSP model.With the computation of the shortest path as the whole optimization goal,IAFSA is utilized to accurately search out the optimal sequence of the mobile robot that traverses all the tasks.The robot have access to all tasks in turn in accordance with the sequence safely.Experimental results show that the algorithm is fast and effective.Finally,the design and implementation of multi-task autonomous navigation system of Pioneer3-DX robot based on laser range finder on ROS are completed.Experiments including map construction,path planning and multi-task planning are carried out under the different real environment.Experimental results reveal that the mobile robot multi-task autonomous navigation system on ROS is stable and feasible.