Indoor Mobile Robot Navigation And Motion Planning Research

With the continuous progress of technology and living standards,indoor mobile robots are gradually entering every field of life--factories,medical treatment,warehouses,shopping malls,hotels and so on.With the increasing use and demand of indoor mobile robot,the functional requirements of indoor robot are gradually improved.At present,most indoor robots applied in practice have no grasping function or can only carry out simple grasping,and the gradually diversified application scenarios put forward higher requirements for the positioning performance in the process of robot navigation.In order to meet higher service requirements and enhance the universality of robot indoor scenes,this paper carries out research on navigation and motion planning based on indoor robots.The main work is as follows.(1)The indoor mobile robot needs analysis,so as to determine the robot hardware system framework and software control system framework.According to the software and hardware framework,the hardware platform selection,navigation related algorithm analysis and implementation,and relevant model establishment of the mobile navigation system are completed respectively,as well as the hardware platform selection of the robot arm target recognition and motion planning system,the kinematics model analysis of the robot arm,camera calibration,hand-eye calibration,target detection,three-dimensional reconstruction of the grasping environment,collision detection,motion planning and other work.Finally complete the robot overall system design.(2)An adaptive Monte Carlo localization algorithm based on the improved sparrow search algorithm is proposed to solve the problem that the adaptive Monte Carlo algorithm caused by particle filter resampling is easy to cause positioning loss and the robot global positioning efficiency is low.The iterative optimization of the improved sparrow search algorithm is used instead of the resampling process to guide the particle population gradually to move towards the high likelihood domain of state estimation,which can improve the global localization efficiency and solve the particle depletion problem fundamentally.In order to solve the problems such as insufficient global search ability and unreasonable sample distribution when Sparrow search algorithm is directly applied to adaptive Monte Carlo location,corresponding improvements are made.The initial distribution of Logistic-tent map is used to improve the quality of subsequent solutions.The global search capability of the improved algorithm is enhanced by adding a proportional coefficient and using a new discoverer update strategy,and the dynamic weight is introduced to coordinate the global large range search in the early stage and the local exact search in the late stage.The follower updating strategy of sparrow search algorithm is improved to make the particle distribution more reasonable and improve the sample diversity.The ant random walk strategy is introduced to increase the search space and further help the algorithm jump out of the local optimal.Experimental results show that compared with other algorithms,the improved algorithm has higher positioning accuracy,more reasonable particle distribution,less samples required for positioning,and effectively improves the global positioning efficiency.(3)An improved RRT*algorithm is proposed to solve the problem of low stochastic tree expansion efficiency caused by strong randomness of sampling points and poor directness in path planning of manipulator.In order to solve the problem of low node expansion efficiency of random tree,a dynamic deflection Angle expansion strategy was proposed by integrating the obstacle information collected by random tree to enhance the orientation of random tree,improve the initial path quality and reduce the probability of falling into local optimal.On the basis of the above improvements,aiming at the problem of too many path turning points,the local node optimization method is used to traverse the entire path node,detect and remove the redundant nodes on the path,so as to simplify the path generated by the extended random tree,further simplify the path and shorten the path length.Cubic Bspline interpolation is used to smooth the path obtained by the improved RRT*algorithm to avoid the jitter phenomenon of the manipulator and allow the brake to track the path.The experimental results show that the path length,number of turning points and average convergence time of the improved algorithm are greatly reduced in three-dimensional path planning,and the end-effector trajectory obtained in the manipulator motion planning is shorter and smoother.