Research On Path Planning Algorithm Of Driverless Vehicle Based On MT-RRT

Unmanned driving technology mainly includes environment awareness,path planning and decision control.It is a complex system.RRT algorithm can be applied to the path planning of driverless vehicles as well.The Multi-Targeting Rapidly-Exploring Random Tree(MT-RRT)path planning algorithm is proposed by improving the RRT algorithm in this dissertation,and the obtained results shows that the proposed MT-RRT path planning algorithm performs better than traditional RRT algorithm.The contribution of this dissertation is as follows:Firstly,the research aimed at RRT algorithm and its subsequent targeted improvements,along with the advantages and disadvantages of each improved RRT algorithm were analyzed.The configuration space is proposed to analyze the dynamic constraints of the driverless vehicle during mobility.It can be seen that vehicles with ackermann steering are generally subject to the constraints of non-integrity conditions in the process of movement,and the corresponding kinematic model is established when the constraints are satisfied.Secondly,the MT-RRT path planning algorithm is proposed.Aiming at the problems of slow convergence speed and low utilization rate of sampling nodes,the MT-RRT algorithm is proposed,which is extended to multiple root nodes on the basis of one root node of the original RRT algorithm.In this algorithm,the multiple trees are created from the initial node,goal node and guided root nodes.Then merge into a tree,so as to realize the path planning.The RRT algorithm,RRT-Connect algorithm,and a kind of improved RRT algorithm based on extension domain used by artificial potential field are implemented in Matlab,and respectively in three types of maps.One is a map with a small number of obstacles,one with a narrow passageway,the other with dense obstacles.To compare the convergence rate,node utilization rate and path length of the four algorithms.Simulation results show that the MT-RRT algorithm has faster convergence speed,higher node utilization and shorter path length.Thirdly,an improved path smoothing algorithm is proposed.The path smoothing algorithm is to eliminate unnecessary nodes by short circuit,so as to reduce the folding angle and inflection nodes of the path as much as possible,to obtain a relatively smooth path.In the path planning of driverless vehicles,the map traversing area is large and the environment is complex,the path length is long and has too many broken corners.In order to reduce the timeof path smoothing,the idea of binary search was added to the path smoothing algorithm to smooth the path obtained by MT-RRT algorithm to reduce the collision detection of some unnecessary nodes.Simulation experiments were carried out on Matlab to compare the speed improvement of the improved path smoothing algorithm.The improved path smoothing algorithm can get the same path,but the speed is greatly improved.