Research On Path Planning Problems For Multiple Mobile Robots

Excellent paths for mobile robots are critical to ensure the implementation of somecertain tasks. Consequently, this thesis concentrates on the path planning problems for mo-bile robots. Research topics are sequentially studied. At the beginning, the path planningproblems for a single mobile robot are investigated. Following that, the path planning prob-lems for multiple mobile robots with conflicts contained in the paths are resolved from theperspective of two different approaches. Thirdly, the path planning problems for multiplemobile robots under double-warehouse with limited resources are implicitly detailed. Lastly,the construction and implementation of the navigational support systems for nonholonomicmobile robots under industrial environments is illustrated. The main purpose of this thesis isaimed at solving the path planning problems for multiple mobile robots while subject to thekinematics constraints, the collision-avoidance constraints, the motion boundary constraints,and the conflicts of the limited resources.Firstly, when it comes to the path planning problems for one single mobile robot, pathplanned by A*algorithm for mobile robot under the grid environment is flawed with manybroken lines, frequently turning points, and large cumulative turning angle. Thus, smoothingA*is proposed to gain the ideal path. In the Smoothing A*model, based on the initial pathplanned by A*, traversing all the nodes on the initial path, when there exist none obstacleson the line connected by the fore-and-aft nodes, deleting the nodes which prolong the lengthof the initial path, the processed path are reached consequently. Simulation results show thatsmoothing A*excels Ant, Anytime D*. When smoothing A*algorithm is adopted, length,total turning points, and cumulative turning angle of the path are almost reduced by5%,50%, and30%～60%, respectively. The path planning problem under different complexenvironments with random obstacles distribution also can be achieved by smoothing A*algorithm.Subsequently, this thesis presents a novel approach which containing two steps to deter-mine conflict-free paths for mobile robots in two and three dimensions with moving obsta-cles. Firstly, the shortest path of each mobile robot is set as the goal function which is subjectto collision-avoidance criterion, path smoothness, and velocity and acceleration constraints.This problem is formulated as calculus of variation problem (CVP). Using parametrizationmethod, CVP is converted to time-varying nonlinear programming problems (TNLPP) andthen resolved. Secondly, move sequence of the mobile robot is assigned by priority scheme;conflicts are resolved by multilevel conflict resolution strategy. Approach efficiency is con- firmed by numerical examples.Furthermore, this thesis studies the problem of automatic generation of the shortest,smooth, obstacle-avoiding and conflict-free paths for efficient guidance of multiple mobilerobots in the changing environment with obstacles static or moving. The polynomial basedpaths are subject to constraints such as motion boundaries, kinematics constraints, obstacle-avoidance, and smoothness. The shortest paths problems for the multiple mobile robots areformulated as one time-varying nonlinear goal function which is restricted with abovemen-tioned constraints. This thesis treats with all the constraints together and conflicts among thepaths are dealt within our model. A time baseline coordination method is designed to resolvethe problems. Numerical simulations show that the proposed approach can fulfill multiplemobile robots path planning problems with stationary and moving obstacles successfully.Fourthly, based on the study of the former research on path planning problems formultiple mobile robots, this thesis proposes the path planning problems for multiple mo-bile robots under double-warehouse with obstacles. Double-warehouse is composed of twoseparated warehouses which are united with two elevators. The elevators are symmetricallyinstalled in the double-warehouse which can be used to transfer mobile robots from onewarehouse to another. In the two disjoint warehouses, polynomials are used to represent theplanned paths. In the elevators, fixed straight lines are chosen to denote the planned paths.The polynomial based paths are subject to constraints such as motion boundaries, kinematicsconstraints, obstacle-avoidance, limited resource of elevators, and smoothness. The shortestpath planning problems for multiple mobile robots are formulated as one time-varying non-linear goal function which is restricted with above-mentioned constraints. Correspondingto the initial and final poses of mobile robots located in the same or different warehouses,the model is divided into two sub-models. When obstacles are taken into account, then eachsub-model is partitioned into two detailed models with or without the presence of obstacles.In terms of the limited resource conflict of elevators, each detailed model consists of thetwo models whether using the same elevator or not. We treat with all the constraints to-gether and all the conflicts are dealt with in our model. When comes to the characteristics ofthe focused problems, approach based on PSO (particle swarm optimization) with penaltyfunction is proposed. With the penalty functions, all the constraints are taken into accountin the objective function. Then, PSO algorithm is applied to achieve the solution. Two ap-proaches are devised to resolve the problems, which are the compressed factorial PSO withperturbation named Con-Per-PSO, and the PSO with simulating annealing named SA-PSO,respectively. Numerical simulations show that, the proposed approach can fulfill multiplemobile robots path planning problems in double-warehouse successfully.Finally, since navigation for nonholonomic mobile robots constitutes a critical part in robotics, then, this thesis develops a user-friendly navigation support system (NSS)developed in C to determine the conflict-free path for nonholonomic mobile robotswith stationary or moving obstacles. The NSS mainly involves three sequential phaseswhich are the layout setting module, the task assignment module and the solution browsemodule, respectively. Firstly, navigation environment with any types of robots is createdby drag-drop technique. Secondly, the mobile robot is assigned and all parameters of thenavigation system are detailed for this mobile robot. Then, the navigation path is achievedwith embedded navigation algorithm based on the parametrization method. Thirdly,the solution for this mobile robot can be browsed from graphical interfaces in the formof the conflict-free path. In our navigation algorithm IPSO (improved particle swarmoptimization, IPSO) which is based on simulated annealing (SA), the shortest navigationpath of each mobile robot is set as the goal function which is subject to collision-avoidance,path smoothness, and velocity and acceleration constraints. Efficiency of NSS is confirmedby some numerical examples. With this system, the shortest, conflict-free and smoothnavigation path can be fulfilled for nonholonomic mobile robots under any environmentpromptly.