| Robotics is a comprehensive subject developed rapidly in recent40years, and is one ofthe most active subjects in today’s technology. Robot’s path planning plays an importantrole in controlling the robot. It has universal applications in many fields, such as robotics,design of very large scale integrated circuits, geographic information system, etc. The maincontents relate to many subjects which includes environment representing, planning method,path searching and artificial intelligence. It has long been concerned by people. Manyscholars have carried on the exploration and research in many aspects and has made someachievements, but there are still many problems need to be further studied.As the research object to6-DOF manipulator PUMA560, my research includes thefollowing four aspects:(1) The kinematics of manipulator is analyzed firstly in this paper. The forwardkinematic model of the manipulator is established by the coordinate transformation and D-Hmethod. Traditional methods have some difficulties to solve the inverse kinematics since itis a complex nonlinear mapping problem, so intelligent control method is used to solve thisproblem in the paper. Method of LMBP neural network is utilized to solve the inversekinematics at first. Then, in order to improve the accuracy, least squares support vectormachine algorithm is used further to explore the problem. Through the simulation results,we can see that LMBP algorithm is available for the small curvature and samplecharacteristics of know type track. As to the large curvature and sample characteristic ofindeterminate type, LSSVM method can get higher precision.(2) I give a path planning model of obstacle avoidance. It uses the method of regularenveloping modeling of obstacles and the three-dimensional problem is transformed to atwo-dimensional one and then the path is searched by A*algorithm. It greatly reduces thecalculation. Aiming at these shortcomings that this path planning can not guarantee thesafety of the manipulator, and can not be dynamic programming, this paper presents adynamic path planning algorithm based on improved A*and gives its simulation. Thealgorithm solves two problems existing in the original algorithm. It can successfullydetermine the position of the obstacle, update the environment information and makereal-time path re-planning according to the robot’s position and orientation informationwhen it encounters an unknown dynamic obstacle. The robot can reach the target nodesecurity at last.(3) I put forward an improved PSO algorithm. It can expand the scope of random search, greatly reduces the possibility of particle trapping in local optimum. Then I give thesimulation of path planning in a static environment. The simulation results are comparedfrom three aspects of optimal path and convergence path length and dynamic convergence,showing that improved particle swarm optimization algorithm is better than the standardparticle swarm algorithm and PSO with a compression factor and inertia weight PSO infinding optimal solution, convergencing solutions faster and smaller fluctuations. Finally, Igive the simulation of improved particle swarm optimization algorithm in a dynamicenvironment. The result shows that the robot can avoid random dynamic obstaclesintelligently and can successfully reach the target location.(4) The sliding mode control algorithm is discussed and sliding mode controller withstrong robustness is designed. Two degrees of6-DOF manipulator are controlled separatelyand the system is developed based on index reaching law and the quasi-sliding mode. Theanalysis of simulation examples show that the improved control method can obtain goodcontrol effect like high tracking precision, little vibration and high stability. |
PDF Full Text Request