The Kinematics,dynamics,and Coordinated Motion Planning Of A Modular Omnidirectional Mobile Manipulator

In this dissertation,a modular mobile manipulator consists of an omnidirectional mobile platform(OMP)with powered caster wheels(PCWs)and a lightweight manipulator based on the integrative joints is proposed.This robot is based on a modular design in which the number of the PCWs and the configuration of the manipulator can be arbitrarily configured to satisfy specific task requirements.Meanwhile,the robot can work in narrow and clutter environment due to its omnidirectional mobility and humanrobot cooperation ability.Hence it is well-suited in the flexible manufacturing and human-robot symbiosis scenarios.In this dissertation,the design,systematic modelling and analysis,and mobility-manipulation coordination for the modular mobile manipulator is presented.The main works and contributions are summarized as below:1)Based on the graph theory,Lie group and screw theory,a configurationindependent kinematic modelling method is proposed for the omnidirectional mobile manipulator.For the manipulator subsystem,the configuration is represented by Assembly Incidence Matrix and the kinematic model is then derived by the local represented product-of-exponentials formula.For the OMP subsystem,the number and install positions of the PCWs is represented by vectors,then the kinematic model is formulated via parallel robot kinematic analysis.The systematic kinematic model of the mobile manipulator is obtained by the assembly relationship of the two subsystems.This modelling method is configuration-independent,and the kinematic model can be generated automatically when the robot configuration representation is given.Additionally,the singularity configurations of the OMP are studied.2)Based on the iterative Newton-Euler method and parallel mechanism dynamic analysis,a configuration-independent dynamic modelling method is proposed.In the dynamic modelling process,the dynamic model of an individual PCW is formulated in both joint and task spaces at first.Then the task space dynamic model of the OMP is derived by applying the AOM method.After that,the dynamic model of the manipulator which is affected by the motion of the OMP is obtained by a recursive Newton-Euler method.The dynamic modelling method for the manipulator is configurationindependent as well,the dynamic model can be generated automatically when the dynamic parameters are given.3)A load distribution model consists of constrained parameters is proposed based on kinematic constraints analysis.The OMP is a multi-chain actuated system and internal force may be induced by the PCWs.A parameterized nonsqueezing load distribution model is proposed by the kinematic constraint analysis.The nonsqueezing dynamic model is not restricted to the configuration of the OMP.Different from conventional method,the obtained model consists of constrained parameters.As a result,the dynamic performance of the OMP can be improved via parameter adjustment,while keeping the internal force as zero.4)The energy consumption optimization of the torque distribution for the redundantly actuated OMP based on modified GPM is proposed.The detailed energy consumption model of the OMP including output energy consumption and electrical energy loss is then formulated,the coefficients of the model is obtained by experimental test.A performance criterion combined with torque limits and energy consumption is proposed in the modified GPM.Compared to two torque distribution schemes which result from the Moore–Penrose pseudo-inverse method,the optimized torque distribution can significantly decrease the maximal required actuating torque as well as total energy consumption,by 11% and 7% respectively.5)A homing motion planning method is proposed.In order to improve the positioning and motion control accuracy,the OMP needs to be translated to a home state after each task.The homing motion planning is executed in a numerical way in which the states of the OMP is successively described in discrete time.A performance index function with respect to the task and joint space velocities is formulated to evaluate the trend to approach the home state between two successive states.With the given algorithm,the OMP is able to successfully move to its home configuration from arbitrary configurations while satisfying the kinematic and dynamic constraints.6)The mobility-manipulation coordination method is presented based on the different characteristics of the two subsystems of the mobile manipulator.The task can be decomposed to the two subsystems via inverse kinematic solution.A weighted fitness function is constructed for the DE method to find the optimal solution.Results show that the task can be successfully assigned in different task definitions.At last,a time-optimal trajectory planning based on the numerical integration method for the mobile manipulator is presented.