Research On Path Planning Of Mobile Robot In Complex Environment

In recent years,with the continuous progress in the fields of computer,automatic control and chip manufacturing,mobile robot technology has also entered a stage of rapid development.Path planning,as an important part of this technology,has become a hot issue that scholars at home and abroad have paid close attention to.However,planning algorithms based on sampling search still have problems such as sampling difficulties,long planning time,and large memory usage when dealing with complex environments and narrow passage planning tasks.Therefore,this thesis makes the following improvements to the representative algorithm:In order to solve the problem that the PRM(Probabilistic Road Map)algorithm cannot generate ideally distributed sampling points in complex environments and configuration spaces containing narrow passages,which affects the construction of connected graphs and leads to poor planning results,this thesis proposes a method that will improved algorithm combining PSO(Particle Swarm Optimization)and PRM algorithm is named PSO-PRM algorithm.This algorithm regards the random sampling points generated in the PRM algorithm as the particles in the PSO algorithm,divides the configuration space into multiple subspaces as the solution space of the particles,and the sampling points in the free space as the optimal position of the group.Using the solution principle of the PSO algorithm,the sampling points located in the obstacle space can be moved and explored to the free space through the position information transmitted by the sampling points in the free space.In addition,in view of the problem that the basic PSO algorithm is easy to fall into the local optimal solution,this thesis improves the position update formula involved in particle exploration,and proposes a random strategy to reselect the global optimal position.After the particle exploration is over,the total number of sampling points in the free space is increased,the connectivity of the undirected graph is improved,and the planning efficiency of the PRM algorithm is improved under the same sampling times.In order to solve the problems of low planning efficiency and poor adaptability of the RRT(Rapidly exploring Random Tree)algorithm in complex environments and configuration spaces containing narrow passages,this thesis re-examines the role of random sampling points in the RRT algorithm,and then proposes a new planning named as the RJ-RRT algorithm.The proposed algorithm designs a sampling space greedy reduction strategy,so that the sampling space is continuously reduced to the goal area,thereby reducing the expansion of the random tree in the area that does not promote the final path construction.At the same time,this thesis also proposes a local configuration recognition method to identify the current difficult to extend regional environment,especially for the identification of narrow passages.This method distinguishes narrow passages into two cases: the entrance of the narrow passage and the interior of the narrow passage,by constructing the subtree expansion space to fuzzily describe the internal trend of the narrow passages,indirectly increasing the proportion of the free space inside the passages to the entire sampling space,and then using multiple subtrees to explore the interior of the narrow passages.Through these improvements,the RJRRT algorithm can perform path search more efficiently and flexibly in complex environments and configuration spaces containing narrow passages.The improvements made to different algorithms in this thesis are compared with the same type of algorithms on different maps.The experimental results show that the algorithm proposed in this thesis has improved in many performance indicators such as planning success rate,planning time consumption and memory usage,thus verifying the effectiveness and feasibility of the proposed algorithm.Finally,a simple mobile robot path planning interactive system is designed and implemented,which improves the work efficiency of subsequent related research.