Study On Inverse Kinematics And Configuration Optimization For Reconfigurable And Modular Robots

With the rapid development of the modern industry, the application fields of robot is more and more extensive, so people hope they can complete more complex tasks, however, traditional robot is developed for specifically applications, how many tasks it can bring to success is constrained to its mechanical structure. When the working place or the given task is changed, the limitation of the traditional robot configuration is obviously increased. To solve this problem, a new type robot, called reconfigurable and modular robot emerges, it can changes its kinematical and dynamical parameters due to the different environment to accommodate diverse environment and complete some more complex tasks, extending the application fields of robot in military, aviation and industry. Scholars in the world have deeply studied the reconfigurable and modular robot, but focus on its automatic model generation in kinematics and dynamics, there are many challenging, just underway and marginal aspects, for example, inverse kinematics solution, assembly configuration optimization and control structure. These problems still need scholars pay much attention and effort. Hence, it is very necessary and significant to study further on reconfigurable and modular robot.In this paper, assembly configuration representation, kinematics, dynamics and configuration optimization for reconfigurable and modular robot were studied, and as the key points the inverse kinematics and configuration optimization based on genetic algorithm were studied. The important content of this paper as follows:A set of standard module were presented based on basic principle of module division and consult module study in existence, consisting of joint, link, and end-effectors. Joint module are the actuators that provide the degree of freedom of each robot, and has two mechanical input and two mechanical output ports, and will produce different types of motion depending on which input and output ports the link are connected to. To represent different assemble configurations from the module inventory, a representation called Assemble Configuration Matrix (ACM) is proposed. There will have the only robot configuration corresponding to a given ACM. Subsequently, how to automatically build kinematics and dynamics models of reconfigurable and modular robot system based on ACM is researched. A twist product-of-exponentials formula is produced to generate robot forward kinematics from a given ACM automatically. The method merely requires product of transformation matrix of every given module according to the module order of the ACM, then forward kinematics equation is gotten. The method greatly simplifies the kinematics nanlysis of different robot configurations. From the view of the engineering application, inverse kinematics may be more valuable, in that it is the base of robot motion planning, path planning and configuration optimization. To solve singularity of Jacobian matrix and solution that is closest to their starting point of the Newton-Raphson iteration method which is used currently in inverse kinematics of reconfigurable and modular robot, a genetic algorithm for solving inverse kinematics of reconfigurable and modular robot is presented. Because inverse kinematics of reconfigurable and modular robot has not an exclusive solution, considering the least joint displacements as optimal value objective and it can find the best solution among all the possible solutions. To test the validity of this method, two examples consisting of 3 DOF and 6 DOF robot configurations are given, the result of simulation shows that this method is validity, and has the least positioning and orientation error and good stability. In the follow, the dynamics of reconfigurable and modular robot is studied, an auto modeling method is used, and the dynamics function can be established in Newton-Euler equation based on ACM and rigid body geometry formula, the generalized velocity and acceleration of each joint can be calculated by forward iteration while the generalized force can also be calculated by backward iteration, then the close loop form of dynamics function is set up, which is adapt to describe reconfigurable and modular robot.In the next place, task-based assembly configuration optimization of reconfigurable and modular robot is discussed on the base of kinematics and dynamics model. The problem is formulated as a combinatorial optimization problem by introducing a task-oriented configuration optimization model, namely, including design parameters, performance constraints, and objective function. Task and module assembly specification are explained in detail, dexterity, energy cost, and multiobjective function that simultaneous consideration of dexterity and energy cost are given, considering discrete task requirement, reachability and joint limit performance constraints. A reachability evaluation method based on genetic algorithm is introduced, which can deal with joint limit constraint simultaneously, avoiding inconvenience of inverse kinematics, and is current. Because of considering dynamics objective function and that point to point motion, the joint space trajectory planning algorithm of reconfigurable and modular robot based on cubic polynomial is design, given start point and end point, the angle, angular velocity and acceleration are calculated by the calculating program. One example including two discrete task points is used to simulate, the result of simulation shows that the obtained trajectory is smooth and continuous for position, velocity and acceleration. In succession, a binary coding scheme to transform an ACM into a string structure is introduced, given the configuration optimization model, task and performance constraints are taken as input parameters to the three objective functions, the functions then respectively evaluates the task performance of an ACM of a robot configuration. Genetic algorithm is proposed for this combinatorial optimization problem. Three examples of finding task-optimal robot configuration utilizing are demonstrated. The result shows that the method is effective.At last, all the work was concluded in the summary, and combining what I have learned, prospect for some questions was laid out.