Research On Trajectory Planning Of Manipulator Based On Intelligent Algorithm

Trajectory planning plays an important role in robot control system. Through planning the position, velocity and acceleration of robot joints, robot can not only achieve motion stably, but also avoid vibration and impact. The robot studied in this thesis is a 6-DOF manipulator in front of the CE2 lunar rover prototype, which is developed by our laboratory. Based on the kinematic model of the manipulator, this thesis focuses the research on kinematics, inverse kinematic and trajectory planning in joint space.Firstly, kinematic model was created by Denavit-Hartenberg rules. Based on the structure characteristic of orthogonal and decoupled manipulators, the kinematical model is simplified to a 3-DOF orientation calculation and 3-DOF posture calculation. The two parts of kinematical calculation are independent. The position and attitude determination problem of this manipulator is solved base on Euclid norm. At the same time, this paper took quaternion as the control input to solve the attitude determination problem. These methods can simplify the computation, achieve the real time and accuracy requirements, and also easy to implement.Secondly, Since polynomial is high order and has not characteristic of Convex hull, it's hard to be optimized by traditional optimization methods. Considering performance and computation, how to choose the best type of piecewise polynomial and intelligent optimization methods is the basic of trajectory planning.Firstly, take advantage of the GA characteristic that are strong robustness and the iteration with global random search optimization ,proposed a Time-Optimal 3-5-3 piecewise polynomial interpolation trajectory planning algorithm based on GA for manipulator. During evolution in SGA, crossover and mutation probability are constant which cause slow convergence or premature convergence. To make up for the shortcomings of the simple genetic algorithm, AGA and RAGA are used to optimize the time of piecewise polynomial. However, traditional adaptive genetic algorithm has some limitations and does not apply to all models. Based on previous work about RAGA, some improvement are added. It turns out that the new algorithm strengthen the abilities of convergence and diversity.Thirdly, to use particle swarm simple structure, easily to adjust the parameters of the characteristics, decrease computational complexity of polynomial due to high order. Searching directly in the optimization space of the target, which reduce the search dimension , skillfully avoided the calculation variables and the dependent variable mapping and retrench computing time. PSO and Adaptive PSO algorithm with dynamically changing inertia weight are used to optimize the 3-5-3 piecewise polynomial interpolation trajectory planning algorithm. Compare to the performance by capability curves, PSO optimal results is far better than GA.Lastly, a 3D motion simulation software system was developed based on MFC and Open Inventor on VC++6.0 for manipulator. The simulation software system of 6-DOF manipulator shows the validity of the kinematics model and trajectory planning method. The result shows the effect of trajectory planning.